SEO and GEO Glossary

Keyword

A keyword is a word or phrase people type into a search box when looking for a topic, product or service. In SEO, pages are planned around the terms an audience is likely to use. Because Google’s language systems can match a page to a query without identical wording, clear and thorough writing matters more than repeating the term.

A veterinary clinic notices owners searching for “kitten vaccination schedule” and “first vaccine for a kitten”, and writes one thorough guide that answers both.

Filling in the meta keywords tag is still widely believed to help rankings. Google Search does not use that tag at all, and repeating the same words excessively counts as keyword stuffing under its spam policies.

Search intentLong-tail keywordKeyword stuffingSearch volume

Google Search Central

Search algorithm

A search algorithm is the set of rules and systems a search engine uses to decide which pages to show for a query and in what order. Google describes this not as a single formula but as a collection of ranking systems working together, including RankBrain, BERT, passage ranking, reliable information systems and SpamBrain.

If someone types “cracked phone screen what now”, Google’s language systems can connect the query to a screen replacement guide even when the page does not use those exact words.

RankingCore updatePageRankGoogle spam policies

Google Search Central

Search engine

A search engine is a system that discovers web pages with automated programs, organises what it finds in an index and, for each query, returns the results it judges most relevant and useful. Google describes the process in three stages, crawling, indexing and serving results, and not every page makes it through all three.

When someone searches for bike repair in Izmir, the search engine picks pages it has already crawled and indexed that fit the query and the searcher’s location, then lists them on the results page.

CrawlingIndexingRankingSERP (search engine results page)

Google Search Central

Search intent

Search intent is what a person is trying to accomplish when they type a query. Google’s Search Quality Rater Guidelines group queries into intents such as Know (find information), Do (complete a task), Website (reach a specific site) and Visit-in-person (go somewhere), and note that a single query can carry several intents.

Someone searching “coffee machine” might want to compare models, buy one or find a nearby shop, which is why guides, product pages and local stores can all show up together.

KeywordSERP (search engine results page)Long-tail keywordContent gap

Google

White hat SEO

White hat SEO is an approach that follows search engine guidelines and puts usefulness to people first. It stays within the technical requirements, spam policies and key best practices set out in Google Search Essentials, building lasting visibility through quality content, sound technical foundations and naturally earned links rather than shortcuts.

Rather than paying for links, a law firm publishes clear, up-to-date answers to the questions clients ask most, improves site speed and shows each author’s credentials on the page.

Black hat SEOGoogle Search EssentialsHelpful contentE-E-A-T

Google Search Central

Core update

A core update is a significant, broad change Google makes to its search algorithms and systems several times a year. It does not target specific sites or pages; instead it refreshes how content is assessed overall. Google announces each one on its Search Status Dashboard so site owners know when an update is happening.

If a recipe site loses traffic after a core update, Google suggests waiting at least a full week after the rollout finishes before comparing Search Console data with the period before it.

A drop after a core update is often read as a penalty. Google says these updates do not target specific sites and that pages which fall are not necessarily bad; other pages may simply be a better fit for the query.

Search algorithmHelpful contentManual action (Google penalty)Google Search Console

Google Search Central

People Also Ask (PAA)

People Also Ask (PAA) is the section of a Google results page that groups questions related to the original search. Google lists it among the exploration features that help people broaden or refine a query. Expanding one of the questions reveals an answer drawn from a web page, presented as a featured snippet with its source.

Someone searching how to get a mortgage may see questions like “what documents do you need for a mortgage” in the PAA box; content teams often use these questions to plan subheadings in their guides.

Featured snippetSERP (search engine results page)Search intentSEO competitor analysis

Google Search Central

Indexing

Indexing is the stage where a search engine analyses a crawled page, including its text, title, images and other content, and stores it in its index. At this point Google groups similar pages and picks a canonical version that may be shown in results. Not every crawled page gets indexed; content quality and robots rules can both affect the outcome.

If the page indexing report in Search Console shows that a new blog post has been crawled but not indexed, the post cannot appear in search results in its current state.

A crawled page is not automatically an indexed one. Google states that not every page it processes will be indexed, and low-quality content or robots rules such as noindex can keep a page out.

CrawlingCanonical tag (canonical URL)Page indexing reportnoindex

Google Search Central

E-E-A-T

E-E-A-T stands for experience, expertise, authoritativeness and trustworthiness, the qualities Google uses to describe content quality. Its rater guidelines call trust the most important of the four. Google says E-E-A-T is not a specific ranking factor in itself, but that its systems draw on a mix of factors that can identify content showing these qualities.

An article on drug interactions written by a pharmacist that clearly lists the author’s name, professional background and sources lets readers see who is behind the content and what it rests on.

E-E-A-T is often treated as a direct ranking score. Google says it is not a specific ranking factor in itself, and that quality raters’ assessments do not directly change how any page ranks.

YMYL (Your Money or Your Life)Helpful contentSEO-friendly content

Google Search Central

Helpful content

Helpful content is reliable, people-first material created mainly to help readers rather than to manipulate search rankings. Google frames it through three questions: who created the content, how it was produced and why it exists. Assessing helpfulness is now part of Google’s core ranking systems rather than the job of a separate system.

A camping gear shop tests its tents in real rain, writes up the results with its own photos and names the person who ran the tests, along with exactly how they were done.

Many writers believe Google prefers a certain word count and that longer content ranks better. Google says outright that it has no preferred word count, and lists writing to a target length as a warning sign of search-engine-first content.

E-E-A-TYMYL (Your Money or Your Life)Thin contentSEO-friendly content

Google Search Central

Google Search Essentials

Google Search Essentials, formerly the Webmaster Guidelines, sets out what web content needs in order to be eligible to appear in Google Search. It has three parts: technical requirements, spam policies and key best practices. The technical requirements describe the bare minimum Google Search needs from a page before it can show that page in results.

Before a new corporate site goes live, the team checks it against Search Essentials: Googlebot can reach the pages, the pages load without errors and nothing on the site breaches the spam policies.

Meeting Search Essentials is often taken to mean a site will be indexed and rank well. Google is clear that following these requirements and practices does not mean its content will be crawled, indexed or served.

Google spam policiesWhite hat SEOTechnical SEOGooglebot

Google Search Central

Google Discover

Google Discover is a feed within Google Search that shows people content related to their interests, based on signals such as their Web and App Activity. Any page that is indexed and meets Discover’s content policies is eligible to appear there, with no special tags or structured data required, although eligibility does not mean it will be shown.

A food site adds original images at least 1200 pixels wide to its seasonal vegetable recipes and keeps its headlines free of exaggeration; those recipes can then surface in the Discover feeds of people interested in the topic.

Many publishers think Discover requires special tags or structured data. Google says it does not, and adds that Discover traffic is less predictable than keyword-driven search traffic and should be treated as a supplement to it.

Image SEOGoogle Search ConsoleContent freshness

Google Search Central

Google spam policies

Google’s spam policies define practices meant to deceive users or manipulate Search, and they form part of Google Search Essentials. They cover, among other things, cloaking, doorway abuse, scaled content abuse, link spam and site reputation abuse. Google detects violations through automated systems and, where needed, human review.

A medical site hosting a low-quality third-party “best casinos” page that has nothing to do with the rest of the site, placed there only to borrow its ranking signals, is one of Google’s own examples of a site reputation violation.

Spam policies are sometimes thought to be only about buying and selling links. They actually span a much wider range, including cloaking, hidden text, keyword stuffing, scraping, sneaky redirects and scaled content abuse.

Black hat SEOManual action (Google penalty)Google Search EssentialsScaled content abuse

Google Search Central

Manual action (Google penalty)

A manual action, often called a Google penalty, is applied when a human reviewer at Google finds that pages on a site do not comply with its spam policies. Site owners are notified through the Manual actions report and the message center in Search Console. Once fixed, they can ask for a review through a reconsideration request.

An online store hit with a manual action for unnatural links to its site removes the problem links, documents the cleanup and requests a review; Google notes that such reviews can take days or weeks.

Not every ranking drop is a Google penalty. Manual actions are reported explicitly in Search Console, Google says most spam is caught automatically, and shifts after a core update are not penalties either.

Google spam policiesGoogle Search ConsoleDisavowCore update

Search Console Yardım

Organic result

An organic result is a listing that appears on a search results page because the search engine’s ranking systems chose it, not because anyone paid. To appear, a page has to be crawled, indexed and judged relevant to the query. Google states that it does not accept payment to crawl a site more often or rank it higher.

If a Bursa furniture workshop’s guide to solid wood dining tables shows up in search results with no ad spend behind it, that listing is an organic result, and the visits it brings count as organic traffic.

Paid result (search ad)Organic trafficRankingSERP (search engine results page)

Google Search Central

Featured snippet

A featured snippet is a special box in which Google reverses the usual result format, showing a passage from a page first and the source after it, when its systems judge that page a good answer to the query. They can also appear inside “People also ask”. Google’s systems choose the page, and no markup can request it.

For a query such as “what pressure should a combi boiler be at”, a plumbing firm’s short, direct answer paragraph could be shown as a featured snippet together with a link to the post.

It is widely believed that some schema markup will win a featured snippet. Google states that there is no way to mark a page as a featured snippet; its own systems decide whether a page is a good fit.

SnippetPeople Also Ask (PAA)Zero-click searchnosnippet and max-snippet

Google Search Central

SEM (search engine marketing)

SEM (search engine marketing) is the umbrella term for marketing that aims to raise a brand’s visibility on search engines. In everyday use, particularly in Turkey, it mostly refers to paid search advertising on platforms such as Google Ads. Ad performance is tracked through metrics like CTR, cost and conversions, while unpaid organic visibility is the territory of SEO.

A newly launched online stationery shop runs search ads on queries like “bulk notebooks” while its site is not yet visible organically, and starts its SEO work in parallel.

Paid result (search ad)SEO (search engine optimization)CTR (click-through rate)Conversion rate

Google Ads Yardım

SEO (search engine optimization)

SEO (search engine optimization) is the practice of helping search engines understand a website’s content and helping people find that site in search results and decide whether to visit. It spans technical setup, content quality, site structure and links from other websites, with the aim of earning unpaid visibility in organic results.

An accounting firm in Ankara publishes a guide on how to register a company, sharpens its page titles and links its service pages together, so Google understands the site better and the right people find it.

Many people assume SEO can lock in the top spot on Google. Google itself says there is no secret that automatically ranks a site first, and that no site is assured of even being added to its index.

Search engineOrganic resultTechnical SEOGEO (generative engine optimization)

Google Search Central

SERP (search engine results page)

A SERP (search engine results page) is the page a search engine shows in response to a query. Alongside standard text results it can include images, videos, featured snippets, related questions and local results. Which elements appear depends on the query, and Google also takes factors such as the searcher’s location, language and device into account.

Someone searching for boiler servicing might see ads, a map of nearby businesses, a “People also ask” box and organic results, all on the same page.

It is easy to assume everyone who types the same query sees the same page. In practice Google weighs factors such as the searcher’s location, language and device, so the results can differ from one person to the next.

Organic resultPaid result (search ad)Featured snippetPeople Also Ask (PAA)

Google Search Central

Zero-click search

Zero-click search describes a search that ends on the results page, with the user getting what they need without visiting any site. It is most common for queries with a short, clear answer, because features such as featured snippets, related questions and AI summaries can meet that need directly. A site may gain visibility here without a visit.

Someone checking the driving distance between Ankara and Istanbul sees the figure in a box on the results page and finishes without opening a single site.

Featured snippetAI OverviewsSERP (search engine results page)CTR (click-through rate)

Google Search Central

Ranking

Ranking is the order in which a search engine arranges indexed pages for a query, based on how relevant and useful it judges each one. Google relies on automated ranking systems that weigh many factors and signals across hundreds of billions of pages. Rankings are not fixed; they can shift with the query, the searcher’s location and time.

A dental clinic’s implant treatment page might sit third one month and seventh the next, simply because competing pages were updated and new content entered the index.

Search algorithmAverage positionRank trackingSearch intent

Google Search Central

Sitelinks

Sitelinks are links to other pages on the same domain, grouped beneath a single text result in Google Search. Google generates them automatically and shows them only when it thinks they will help the user. Site owners cannot currently choose which links appear, but clear titles, a logical site structure and concise internal anchor text support their quality.

A search for the brand name “examplebrand” might show its homepage result with links to “Products”, “Contact” and “Order tracking” grouped underneath.

Sitelinks are often thought to be something you can pick through a setting or markup. Google says they are currently automated; to get rid of an unwanted one, it suggests removing the page from the site or using noindex.

Internal linkAnchor textSERP (search engine results page)

Google Search Central

Black hat SEO

Black hat SEO is the umbrella term for tactics that try to gain visibility by deceiving search engines or manipulating their ranking systems, in breach of spam policies. Cloaking, keyword stuffing, link spam and hidden text are typical examples. Google says sites that violate these policies may rank lower or disappear from results entirely.

A site that serves search engines a keyword-packed page while showing visitors something entirely different is cloaking, one of the textbook black hat techniques.

White hat SEOGoogle spam policiesCloakingManual action (Google penalty)

Google Search Central

Crawling

Crawling is the process search engines use to find new and updated pages on the web and download their content. Google does this with Googlebot, which discovers URLs through links on known pages and through sitemaps submitted by site owners. Googlebot decides how often to crawl each site and limits its request rate to avoid overloading servers.

When an online store adds a new category page, links it from the main menu and lists it in its sitemap, Googlebot can discover the URL through either route and crawl it.

People often assume Google crawls every URL it finds. It does not: pages the site owner has blocked from crawling, or pages that sit behind a login, may not be crawled at all.

GooglebotCrawl budgetrobots.txtXML sitemap

Google Search Central

Long-tail keyword

A long-tail keyword is a longer, more specific search phrase that few people type on its own but that spells out exactly what the searcher needs. Because volumes are low, such phrases may not show up in keyword tools at all. In return they reveal intent clearly and make it easier to build content around one precise need.

“Shoes” is a broad term, while “waterproof women’s hiking shoes” is a long-tail phrase; an outdoor shop can write a dedicated category description that meets exactly that need.

If Keyword Planner shows nothing for a phrase, it is tempting to conclude nobody searches for it. Google notes that keywords with very low search volume, or those considered sensitive, simply cannot be discovered or forecast in the tool.

KeywordSearch volumeSearch intentKeyword difficulty (KD)

Google Ads Yardım

Paid result (search ad)

A paid result, or search ad, is a listing an advertiser pays to place on a search results page through an advertising platform. In Google Ads every ad and keyword has its own CTR, and a keyword’s expected CTR is one of the components of Ad Rank. Paid placement runs separately from organic ranking.

A hotel in Antalya can run ads on searches for all-inclusive hotels in Lara and appear on the results page as a sponsored listing; once the campaign is paused, that visibility goes with it.

A common belief is that advertising on Google lifts a site’s organic rankings as well. Google says plainly that it does not accept payment to rank a site higher or crawl it more often; ad spend does not buy organic position.

SEM (search engine marketing)Organic resultCTR (click-through rate)SERP (search engine results page)

Google Ads Yardım

YMYL (Your Money or Your Life)

YMYL (Your Money or Your Life) is the label Google’s Search Quality Rater Guidelines use for topics that could significantly affect people’s health, financial stability or safety, or the welfare of society. Pages on these topics are held to higher quality standards, and Google says its systems give even more weight to content showing strong E-E-A-T in these areas.

Pages explaining how to fill in tax forms, save for retirement or vote all fall under YMYL, whereas an article on how often to wash your jeans does not.

Not every page that mentions health or money is YMYL. The guidelines treat YMYL as a spectrum, and everyday health topics such as how often to replace a toothbrush are cited as cases where a small inaccuracy is unlikely to cause real harm.

E-E-A-THelpful contentSEO-friendly content

Google

301 redirect

A 301 redirect uses the HTTP status code saying a resource has permanently moved to the new URL in the Location header. Browsers and search engines follow it. Google treats a 301 as a strong signal that the target should become canonical and aims to show the new URL in search results; a 308 is handled the same way.

When example.com retires its old /services/seo-consulting address and republishes the content at /seo-consultancy, it sets up a server-side 301 redirect from the old URL to the new one.

A persistent myth says a 301 redirect leaks part of a page’s link value. According to Google’s site move guide, 301 and other permanent redirects do not cause a loss in PageRank.

302 redirectRedirect chainSite migrationCanonical tag (canonical URL)

Google Search Central

302 redirect

A 302 redirect uses the HTTP status code that says a resource is temporarily available at another URL. Browsers and Googlebot follow it, but Google treats it only as a weak signal that the target should be canonical, so the source URL usually stays in search results. Google handles a 307 the same way.

While a service page on example.com is down for a few days of maintenance, visitors are sent to an explanation page with a 302, so the original URL can keep its place in search results.

Some think 301 and 302 lead to the same outcome in Google. Google reads a permanent redirect as a strong canonical signal and a temporary one as a weak signal, so permanent moves call for a 301 or 308.

301 redirectHTTP status codeCanonical tag (canonical URL)

Google Search Central

404 error (page not found)

A 404 error is the HTTP status code a server returns when it cannot find the requested resource. Typical causes are mistyped URLs and pages moved or deleted without a redirect. The code does not say whether the absence is temporary or permanent. Google does not index 404 URLs, removes previously indexed ones and gradually crawls them less often.

A retired campaign page on example.com returns 404; visitors see a custom error page with a search box and links to the main categories, while the server response stays a true 404.

Many redirect every removed page to the home page. Google notes that sending lots of old URLs to one irrelevant destination may be treated as a soft 404; pages with no replacement should return 404 or 410.

Soft 404410 GoneBroken linkHTTP status code

MDN Web Docs

410 Gone

The 410 Gone status code tells clients that the requested resource is no longer available on the server and that this is likely permanent. That sets it apart from 404, which does not say whether the absence is temporary or permanent. If the site owner is unsure whether the removal is permanent, 404 is the recommended choice.

example.com returns 410 for old event announcements that have expired and will not be published again; where a current page covers the same topic, those URLs are redirected there with a 301 instead.

Some believe a 410 makes Google remove a URL differently from a 404. Google’s documentation says all 4xx codes except 429 are handled the same way: previously indexed URLs that return 404 or 410 are removed from the index.

404 error (page not found)HTTP status codeSite migration

MDN Web Docs

5xx server error

A 5xx server error is any HTTP status code saying a request failed because of a problem on the server side; 500, 502 and 503 are the most common. When Google receives these codes, or a 429, it temporarily slows down crawling. Indexed URLs are kept at first but can be dropped if the errors persist.

If example.com starts returning 503 errors during a busy campaign, Googlebot crawls less; once the server responds with 200 again, Google gradually increases the crawl rate.

A brief spell of 5xx errors is often thought to knock pages out of the index right away. Google says indexed URLs are kept at first while crawling slows; only URLs that keep returning errors are eventually removed.

HTTP status codeCrawl budgetGooglebot

Google Crawling Infrastructure

Breadcrumb

A breadcrumb is a navigation element that shows a page’s position in the site hierarchy as a trail of links, such as Home › Category › Subcategory. It helps users understand the site and move up one level at a time. Marked up with BreadcrumbList structured data, it can help Google categorize the page’s information in search results.

A product page on example.com shows the trail Men › Shirts › Linen Shirts near the top and marks up the same path with BreadcrumbList JSON-LD that uses the position, name and item properties.

Structured dataInternal linkSEO-friendly URL structureRich result

Google Search Central

Canonical tag (canonical URL)

A canonical tag is a rel=canonical link annotation that tells search engines which URL you prefer to have shown among duplicate or very similar pages. It goes in the page’s head or in an HTTP header. Google treats it as a strong signal, but it is a hint rather than a rule: Google makes the final choice of canonical URL.

If a product on example.com opens both at /product/blue-shirt and at a URL with a campaign parameter, both versions get a canonical tag in their head pointing to https://example.com/product/blue-shirt.

Many treat the canonical tag as a command Google must obey. Google reads it as a strong hint, yet it can pick a different canonical URL after weighing other signals such as redirects, sitemaps and internal links.

Duplicate contentnoindexhreflang tag301 redirect

Google Search Central

CLS (Cumulative Layout Shift)

CLS (Cumulative Layout Shift) is the Core Web Vitals metric for visual stability. It measures the largest burst of layout shift scores from unexpected shifts during a page’s lifetime. A CLS of 0.1 or less is good and above 0.25 is poor. Images without dimensions, late-loading ads and web fonts are common causes.

If an ad slot that loads late above the article text pushes the content down on example.com, CLS goes up; reserving a fixed height for the slot in advance reduces the shift.

Core Web VitalsLCP (Largest Contentful Paint)INP (Interaction to Next Paint)Lazy loading

web.dev

Core Web Vitals

Core Web Vitals are a set of three metrics that measure real-world user experience for loading performance, responsiveness and visual stability: LCP, INP and CLS. Good values are an LCP of 2.5 seconds or less, an INP of 200 milliseconds or less and a CLS of 0.1 or less. Google’s ranking systems use these metrics.

If the Core Web Vitals report in Search Console shows poor mobile LCP on example.com’s product pages, the team compresses the large hero image and tracks whether the 75th percentile meets the 2.5-second target.

Some expect a perfect Core Web Vitals score to bring the top ranking. Google says good results do not ensure top positions; page experience is more than these scores, and chasing a perfect score purely for SEO may not be the best use of time.

LCP (Largest Contentful Paint)INP (Interaction to Next Paint)CLS (Cumulative Layout Shift)Page experience

Google Search Central

Faceted navigation

Faceted navigation is a site feature that lets visitors narrow lists of products, articles or events by attributes such as color, size or price. Because it is usually built with URL parameters, filter combinations can generate a near-infinite number of URLs, leading to overcrawling and slower discovery of new pages.

example.com uses robots.txt to block crawling of the parameter URLs created by its color and size filters, leaving only product pages and the unfiltered category listing open to crawlers.

Many expect a canonical tag alone to solve the crawling load from filter URLs. Google says canonical and nofollow are generally less effective in the long run than blocking those URLs in robots.txt or handling filters with URL fragments.

URL parameterCrawl budgetDuplicate contentrobots.txt

Google Crawling Infrastructure

Lazy loading

Lazy loading defers resources such as images that sit outside the visible viewport until they come within a calculated distance of it, instead of loading them with the page. Browser-level lazy loading uses the loading=lazy attribute on img elements. It lightens the initial load but should not be used for images visible in the first viewport, especially the LCP image.

On a long category page at example.com, product images in the first row load normally while those further down get loading=lazy; every image also has its width and height set to limit layout shift.

Adding lazy loading to every image is often assumed to speed things up. Images visible in the first viewport, especially the LCP image, then appear later; they should keep the browser’s default eager loading.

LCP (Largest Contentful Paint)CLS (Cumulative Layout Shift)Image SEOCore Web Vitals

web.dev

Googlebot

Googlebot is the generic web crawler Google Search uses to discover and fetch pages. It finds URLs through links and sitemaps, downloads pages and renders them by running JavaScript in a recent version of Chrome. It comes in two types, Googlebot Smartphone and Googlebot Desktop, and it respects robots.txt rules on automatic crawls.

When the server logs of example.com show requests with a Googlebot user agent, the team can confirm that they really come from Google by checking the source IP address and its reverse DNS hostname.

Crawlingrobots.txtCrawl budgetServer log analysis

Google Crawling Infrastructure

hreflang tag

An hreflang tag is a rel=alternate hreflang annotation that tells Google about the language or regional versions of a page. It can be added in the HTML head, in HTTP headers or in a sitemap. Each version should list itself and every other version, helping Google point users to the version that best fits their language or region.

Each Turkish /tr/ and English /en/ page on example.com carries link tags pointing to both versions with hreflang=tr and hreflang=en, plus an x-default entry for the language selector page.

Many think hreflang tells Google what language a page is in. Google detects language algorithmically from visible content; hreflang only connects the localized versions of the same content.

Canonical tag (canonical URL)XML sitemapDuplicate content

Google Search Central

HTTP status code

An HTTP status code is the three-digit code a server returns to report the outcome of a request from a browser or a crawler. 2xx means success, 3xx a redirect, 4xx a client error and 5xx a server error. Google uses these codes to decide whether to process content, follow a redirect, drop a URL or slow down crawling.

When the page of a discontinued product on example.com returns 404 or 410 instead of 200, Google drops the URL from its index over time and crawls it less and less often.

404 error (page not found)Soft 4045xx server error301 redirect

Google Crawling Infrastructure

HTTPS and SSL certificate

HTTPS is the encrypted version of the HTTP protocol, using TLS to encrypt all communication between a browser and a server. The certificate people often call an “SSL certificate” is what makes this secure connection possible. Serving pages securely is one of Google’s page experience considerations, and Google prefers equivalent HTTPS pages when choosing canonicals.

When moving to HTTPS, example.com installs a valid certificate, permanently redirects every HTTP URL to its HTTPS counterpart and uses only HTTPS URLs in its sitemap and canonical tags.

Page experienceCanonical tag (canonical URL)Site migration301 redirect

MDN Web Docs

IndexNow

IndexNow is a protocol that lets website owners notify participating search engines with a simple ping when a URL has been added, updated or deleted. It helps those engines prioritize crawling of the changed URLs. The protocol’s site lists Microsoft Bing, Naver, Seznam.cz, Yandex and Yep as supporting search engines.

Whenever a product price changes or a new blog post goes live, example.com’s content management system sends the URL to an IndexNow endpoint, while it relies on an up-to-date sitemap for Google.

Some expect URLs submitted through IndexNow to reach Google as well. Google does not appear in the list of supporting search engines on the protocol’s official site; for Google, sitemaps and the URL Inspection tool are the usual routes.

XML sitemapBing Webmaster ToolsCrawlingGoogle Search Console

IndexNow.org

INP (Interaction to Next Paint)

INP (Interaction to Next Paint) is the Core Web Vitals metric that assesses a page’s overall responsiveness by observing the latency of pointer and keyboard interactions throughout a visit. An INP of 200 milliseconds or less is good, and above 500 milliseconds is poor. It replaced FID, which measured only the input delay of the first interaction.

If the button that opens the filter panel on example.com responds late because a heavy JavaScript task keeps the main thread busy, INP rises; splitting that task into smaller pieces can speed up the response.

INP is sometimes thought to measure only the first interaction on a page. That describes FID, the metric it replaced; INP observes all interactions during a visit and, on most pages, reports the slowest one.

Core Web VitalsLCP (Largest Contentful Paint)CLS (Cumulative Layout Shift)JavaScript SEO

web.dev

JavaScript SEO

JavaScript SEO is the technical work of making sites that build their content and links with JavaScript crawlable, renderable and indexable. Google runs JavaScript with an evergreen version of Chromium, but rendering is queued and not every bot can execute JavaScript. Meaningful status codes, crawlable links and the History API are core concerns.

If category addresses in the single-page app at example.com use the #/category pattern, Googlebot cannot reliably resolve them; switching to /category URLs with the History API makes the pages easier to discover.

Some believe Google cannot see JavaScript-generated content at all. Google renders pages with an evergreen Chromium; still, because rendering is queued and not every bot runs JavaScript, server-side rendering or pre-rendering remains a good idea.

RenderingServer-side rendering (SSR)GooglebotSoft 404

Google Search Central

JSON-LD

JSON-LD is a format for adding structured data to a page as JavaScript notation inside a script tag in the HTML. Because the markup is not interleaved with the visible text, nested data items are easy to express. Google also supports Microdata and RDFa but recommends JSON-LD as the easiest option to implement and maintain.

example.com adds a script tag with the type application/ld+json to the head of each blog post, describing the post’s headline, author and publication date with the Article type.

Structured dataSchema.org (schema markup)Rich result

Google Search Central

Broken link

A broken link, also called a dead link, points to a page that no longer exists or has moved, so it returns an error such as a 404. Common causes are mistyped URLs and pages deleted or moved without a redirect. Broken links interrupt the visitor’s path and keep the intended content from being reached through that link.

If a product on example.com moves to a new URL while blog posts still link to the old address, those links hit a 404; updating them to the new URL and adding a 301 on the old one fixes the problem.

404 error (page not found)Internal link301 redirectOrphan page

Google Search Central

Duplicate content

Duplicate content is the same or very similar content being accessible at more than one URL. Common causes include regional and device variants, HTTP and HTTPS versions, sort and filter parameters, and demo sites accidentally left open to crawlers. Google clusters such pages, picks one as canonical and usually shows that one in results.

When a category page on example.com produces dozens of URLs through price sorting and color filters, those versions show the same product list, so they count as duplicate content and Google picks one of them as canonical.

A widespread belief is that duplicate content automatically triggers a penalty. Google says some duplication on a site is normal and not a spam policy violation; the real costs are a confusing user experience and harder performance tracking.

Canonical tag (canonical URL)URL parameterFaceted navigationKeyword cannibalization

Google Search Central

LCP (Largest Contentful Paint)

LCP (Largest Contentful Paint) is the Core Web Vitals metric that reports how long it takes, from the moment the user starts navigating to a page, for the largest image, text block or video visible in the viewport to render. It reflects perceived load speed. A good LCP is 2.5 seconds or less at the 75th percentile of page loads.

If the largest element on example.com’s home page is the hero image, keeping that image out of lazy loading and reducing its file size can improve LCP.

Core Web VitalsTTFB (Time to First Byte)Lazy loadingCLS (Cumulative Layout Shift)

web.dev

Mobile-first indexing

Mobile-first indexing means Google uses the mobile version of a site’s content, crawled with its smartphone agent, for indexing and ranking. Text, headings, structured data or metadata missing from the mobile version may not be part of what Google evaluates. Google recommends responsive design, which serves the same HTML on every device.

If the specification table on example.com’s desktop product page is removed on mobile, Google may not see that information; placing the table inside an accordion on mobile keeps the content while saving space.

People sometimes assume the full desktop content will still be indexed even if the mobile version is trimmed. Google indexes from the mobile version and warns that a mobile page with less content can lead to some traffic loss.

GooglebotPage experienceCore Web VitalsIndexing

Google Search Central

noindex

noindex is an indexing rule that asks search engines not to show a page or file in their results. It is set with a robots meta tag or an X-Robots-Tag HTTP header. Once Googlebot crawls the page and sees the rule, Google drops it from Search results even if other sites link to it, provided the page can be crawled.

Order confirmation and thank-you pages on example.com offer nothing to searchers, so the team adds a robots meta tag carrying the noindex rule to the head of those pages.

People often add a robots.txt block on top of noindex to be safe. Because a blocked page cannot be crawled, Google never sees the noindex rule, and the URL can keep appearing in results if other pages link to it.

Robots meta tagX-Robots-Tag headerrobots.txtIndexing

Google Search Central

Rendering

Rendering is the process in which a browser or search engine processes a page’s HTML, CSS and JavaScript to produce the final version a user would see. Google queues pages that return a 200 status code for rendering and, when resources allow, runs the JavaScript in a headless Chromium, then uses the rendered HTML for links and indexing.

If product descriptions on example.com are fetched with JavaScript after the page loads, the initial HTML does not contain that text; Google can see it only after rendering the page.

JavaScript SEOServer-side rendering (SSR)IndexingGooglebot

Google Search Central

Robots meta tag

A robots meta tag is a meta element placed in the head of an HTML page to tell search engines how that page may be indexed and shown in results. It carries rules such as noindex, nofollow, nosnippet and max-snippet; a name value of robots addresses all crawlers, while googlebot targets Google only.

To keep its internal search result pages out of Google, example.com adds a meta tag with the name set to robots and the content set to noindex to the head of those pages.

noindexX-Robots-Tag headerrobots.txtnosnippet and max-snippet

Google Search Central

robots.txt

robots.txt is a plain text file at the root of a site that tells crawlers which URLs they may access. It follows the Robots Exclusion Protocol, standardized as RFC 9309. Its main job is managing crawler traffic and avoiding server overload; it is not a mechanism for keeping pages out of Google.

To stop crawlers from fetching list URLs with a sort parameter, example.com adds a Disallow rule covering those addresses for all user agents in its robots.txt file and lists its sitemap URL at the end.

A common belief is that a Disallow rule removes a page from Google’s index. A blocked URL that other pages link to can still appear in results without a description; keeping it out takes noindex or password protection.

Robots meta tagnoindexGooglebotCrawl budget

Google Search Central

Page experience

Page experience describes how good a page is to use beyond its content: good Core Web Vitals, a secure HTTPS connection, proper display on mobile devices, and no excessive ads or intrusive interstitials getting in the way of the main content. Google’s core ranking systems look to reward content that provides a good page experience.

If blog posts on example.com show a newsletter pop-up that covers the whole screen on mobile as soon as they load, turning it into a small banner improves page experience and keeps the main content accessible.

Many think Google uses a single “page experience signal”. Google says there is no single signal; its core ranking systems look at a variety of signals that align with overall page experience.

Core Web VitalsHTTPS and SSL certificateMobile-first indexing

Google Search Central

Pagination

Pagination splits a long list of content or products across several pages linked with “next”, “previous” and page-number links. For Google to find everything, each page needs its own URL, the pages should link to each other sequentially with a href links, and each page should carry its own canonical URL.

The blog archive on example.com is paginated with separate URLs such as ?page=2 and ?page=3; each page links to the next one and back to the first, and page numbers live in a query parameter rather than a # fragment.

A common mistake is pointing every page in a paginated series to page one as canonical. Google advises against it and recommends that each page have its own canonical URL; it also no longer uses rel=next and rel=prev.

Canonical tag (canonical URL)Internal linkURL parameterFaceted navigation

Google Search Central

Schema.org (schema markup)

Schema.org is a shared vocabulary for describing things on web pages, such as people, organizations, products and events, in a way the major search engines (Google, Microsoft, Yandex and Yahoo!) can understand. It is applied with the Microdata, RDFa or JSON-LD formats. For Google Search behavior, Google’s own structured data documentation is the definitive reference.

example.com uses the schema.org vocabulary to mark up its company details with the Organization type, its blog posts with the Article type and its page paths with the BreadcrumbList type.

Structured dataJSON-LDRich resultEntity

schema.org

SEO-friendly URL structure

An SEO-friendly URL structure builds addresses that people and search engines can easily understand, using readable words and a logical hierarchy. Google recommends descriptive words instead of long ID numbers, hyphens rather than underscores to separate words, and trimming unnecessary parameters that do not change the content.

example.com structures product addresses as /mens-clothing/linen-shirt instead of /product.php?id=4821&cat=7, keeps every URL in lowercase and drops tracking parameters that do not change the content.

URL parameterOn-page SEODuplicate contentBreadcrumb

Google Search Central

Site migration

A site migration changes the URLs of existing pages, for example when moving from HTTP to HTTPS, changing or merging domain names, or changing URL paths. A successful move relies on mapping old URLs to new ones, server-side permanent redirects, and updated internal links, canonical tags and sitemaps.

Moving from example.com to example.net, the team first builds a table that maps old URLs to new ones, redirects each old address to its counterpart with a 301 and submits a Change of Address in Search Console.

301 redirectRedirect chainCanonical tag (canonical URL)Google Search Console

Google Search Central

Soft 404

A soft 404 happens when a server returns a 200 success response but the content looks like an error, such as an empty page or a “not found” message. Google reports these as soft 404 errors in Search Console. Because such pages keep getting crawled, they waste crawl budget; pages that truly do not exist should return a 404.

If the URLs of permanently removed products on example.com show a template saying “Product not found” while returning 200, Search Console may flag them as soft 404s; the fix is to return a real 404.

Soft 404 is sometimes mistaken for a status code the server sends. There is no such HTTP code: the server returns 200, and Google applies the label when the page content looks like an error or an empty page.

404 error (page not found)HTTP status codeCrawl budgetJavaScript SEO

Google Crawling Infrastructure

Server log analysis

Server log analysis is the review of a web server’s access and error logs to see how bots and users actually visit a site. It shows which URLs Googlebot crawls and how often, which URLs return unexpected error codes and where crawling is wasted. Because Google’s crawlers run on many machines, the logs may show visits from several IP addresses.

When example.com’s access logs show that most Googlebot requests go to list URLs with sort parameters, the team blocks those parameters in robots.txt and tracks how often product pages get crawled.

GooglebotCrawl budgetSEO auditBot traffic

Google Crawling Infrastructure

Server-side rendering (SSR)

Server-side rendering (SSR) means generating a page’s HTML on the server and sending it complete with content, rather than building it in the user’s browser. The content is then present in the initial HTTP response, so search engines do not need to run JavaScript to understand it. Google recommends server-side or pre-rendering because it is faster for users and crawlers.

Built with React, example.com renders its product and category pages on the server, so Googlebot and other bots that do not run JavaScript see the product name, price and description in the first response.

RenderingJavaScript SEOTTFB (Time to First Byte)

Google Search Central

Crawl budget

Crawl budget is the set of URLs that Google can and wants to crawl on a site. It combines the crawl capacity limit, which keeps Google from overloading the server, with crawl demand, driven by the site’s size, update frequency, popularity and quality. It matters mainly for very large or frequently updated sites.

If Googlebot spends its time on hundreds of thousands of duplicate filter URLs at the example.com store, newly added product pages may be discovered later; trimming unnecessary URLs steers crawling toward the pages that matter.

Many assume every site needs to manage crawl budget. Google’s guidance mainly targets very large or rapidly changing sites; for smaller sites, keeping the sitemap current and checking the Page Indexing report is generally enough.

GooglebotFaceted navigationServer log analysisXML sitemap

Google Crawling Infrastructure

Technical SEO

Technical SEO is the work of making a website’s infrastructure easy for search engines to crawl, render and index. It covers server responses, robots.txt, XML sitemaps, canonical tags, redirects, HTTPS, page speed, mobile readiness and structured data, so that the content a site publishes can actually reach Google’s index.

If new products on an online store at example.com stay invisible in search for months, a technical SEO audit starts by checking robots.txt rules, noindex tags, server errors and the URLs listed in the sitemap.

GooglebotCrawl budgetIndexingSEO audit

Google Search Central

TTFB (Time to First Byte)

TTFB (Time to First Byte) measures the time between the start of navigation to a page and the moment the first byte of the response begins to arrive. It reflects connection setup and server responsiveness, and it comes before loading metrics such as LCP. According to web.dev, 0.8 seconds or less is good and more than 1.8 seconds is poor.

If example.com’s category pages are rebuilt from the database on every request and TTFB is high, page caching or a content delivery network can speed up the server’s first response.

TTFB is often listed as a Core Web Vital. It is not part of that set; meeting its “good” threshold is not strictly necessary as long as it does not keep a site from scoring well on the metrics that matter.

LCP (Largest Contentful Paint)Core Web VitalsServer-side rendering (SSR)5xx server error

web.dev

URL parameter

A URL parameter is the part of an address after the question mark, made of key=value pairs separated by ampersands. Parameters handle sorting, filtering, sessions and campaign tracking. Those that do not change the content, or that can be stacked on top of each other, can create large numbers of URLs for the same content and make crawling inefficient.

In example.com/shoes?color=black&sort=price, the color parameter narrows the list while the sort parameter only reorders the same products; since the latter creates no new content, it may not need to be crawled.

Faceted navigationDuplicate contentCrawl budgetUTM parameters

Google Search Central

X-Robots-Tag header

The X-Robots-Tag header delivers robots meta tag rules as an HTTP response header. It is the way to apply indexing rules such as noindex to non-HTML files like PDFs, images and videos, and it can be set in the server configuration to cover an entire file type or a whole site with a single rule.

Because it does not want its price list PDFs in search results, example.com configures its server to send an X-Robots-Tag: noindex header with every file ending in .pdf.

Robots meta tagnoindexCanonical tag (canonical URL)robots.txt

Google Search Central

XML sitemap

An XML sitemap is a file that gives search engines information about a site’s pages, images, videos and other files so they can crawl it more efficiently. It can flag which URLs the owner considers important, when pages were last updated and which language versions exist. It helps most on large, new or poorly linked sites.

example.com builds a sitemap that lists only the canonical product and category URLs it wants indexed, adds a lastmod date for each one and submits the file in Search Console.

It is often assumed that every URL in a sitemap will be crawled and indexed. A sitemap helps search engines discover URLs, but it does not mean every listed item will be crawled or indexed.

CrawlingIndexingCanonical tag (canonical URL)IndexNow

Google Search Central

Structured data

Structured data is a standardized format for providing information about a page and classifying its content, such as the ingredients, cooking time and calories on a recipe page. Google uses it to understand pages and to enable rich results. Most of it relies on the schema.org vocabulary and is often written in JSON-LD.

Product pages on example.com mark up the product name, price, availability and rating with the Product type, and the code is validated with the Rich Results Test before it goes live.

Some think any accurate information can be marked up, even if it is not on the page. Google asks site owners not to add structured data about information that is not visible to users, and not to create empty pages just to hold it.

Schema.org (schema markup)JSON-LDRich resultEntity

Google Search Central

Redirect chain

A redirect chain is a URL reaching its final destination through several intermediate redirects instead of one, such as page A redirecting to B and B to C. Chains add latency for users and have a negative effect on crawling. In a redirect loop, the path leads back to an earlier URL and the request never reaches a destination.

If example.com moved from HTTP to HTTPS, then to the non-www host and later to a new URL structure, an old link may take three hops; the fix is to redirect every old URL straight to its final address.

301 redirectSite migrationCrawl budget302 redirect

Google Search Central

Rich result

A rich result is a search result that Google shows with a richer appearance, using structured data on the page; review ratings, breadcrumbs and event details are typical examples. To be eligible, the markup must include the required properties for that feature type and follow Google’s structured data guidelines.

When event pages on example.com mark up the date, venue and ticket details with the Event type, Google may present them as rich results that show those event details.

Many assume markup that passes the Rich Results Test will show up as a rich result. Google makes no such commitment even for correctly marked-up pages; its algorithm decides based on what it considers the best experience for the user.

Structured dataSchema.org (schema markup)SERP (search engine results page)Featured snippet

Google Search Central

Landing page

A landing page is the page where a visitor first enters a site, whether from a search result, an ad, an email or a social post. In marketing it often means a page built around one goal, such as a quote request or a purchase. For SEO, a good landing page directly answers the intent behind the visitor’s query.

Someone searching for “corporate flower subscription” lands not on a florist’s home page but on a page that brings together subscription plans, delivery areas and a quote form.

Search intentConversion rateOrganic traffic

Google Search Central

Alt text (alt attribute)

Alt text is the description written in an image’s HTML alt attribute, which can replace the image when it isn’t displayed. Screen readers rely on it, and Google uses alt text along with computer vision to understand what an image shows. An empty alt value marks a purely decorative image that isn’t part of the content.

A pet store describes a collar photo as “red leather dog collar with metal buckle” instead of “product image”, and gives the decorative divider graphic on the same page an empty alt value.

Some believe that packing alt text with keywords helps image rankings. Google warns that keyword stuffing in alt attributes results in a negative user experience and may cause the site to be seen as spam.

Image SEOKeyword stuffingOn-page SEO

Google Search Central

Keyword stuffing

Keyword stuffing is the practice of filling a web page with keywords or numbers in an attempt to manipulate rankings, and it violates Google’s spam policies. Typical examples include repeating phrases so often that they sound unnatural, blocks of text listing cities and regions a page wants to rank for, and lists of phone numbers without substantial added value.

A moving company that stacks dozens of lines like “London movers, Camden movers, Hackney movers” at the bottom of its page is keyword stuffing. Its service areas could be presented in a useful coverage table instead.

Keyword densityGoogle spam policiesBlack hat SEOAlt text (alt attribute)

Google Search Central

Keyword cannibalization

Keyword cannibalization is when two or more pages on the same site answer the same search intent and compete for the same queries. It is an industry term rather than one Google documents by name. Which page appears can shift from period to period, and internal and external links end up split between the competing pages.

A software company’s blog has two posts, “What is CRM” and “What is CRM software”, answering the same question. Search Console shows both URLs taking turns on the same query, so the posts are merged into one page.

Search intentCanonical tag (canonical URL)301 redirectInternal link

Search Console Yardım

Keyword density

Keyword density is the number of times a keyword appears in a text divided by the total word count, usually expressed as a percentage. Google notes that excessively repeating the same words is tiring for readers and that keyword stuffing violates its spam policies, so natural, complete coverage of a topic matters more than any ratio.

If “steel door models” appears 40 times in an 800-word article, the copy reads as forced. The writer cuts the repetition and covers related subtopics such as sizes, locking systems and insulation instead.

Many believe there is an ideal keyword density to hit for rankings. Google’s guidance defines no such ratio; it says excessive repetition is tiring for users and that keyword stuffing goes against its spam policies.

Keyword stuffingKeywordSEO-friendly content

Google Search Central

Heading tags (H1–H6)

Heading tags are the HTML elements from H1 to H6 that mark the hierarchy of sections on a page. The H1 usually carries the main heading, while H2 and lower levels structure the subsections. Headings in semantic order help screen reader users and make a page easier to skim, and Google may draw on them when generating title links.

In a coffee guide, the H1 reads “How to Brew Filter Coffee”. “What you need”, “Grind size” and “Brewing steps” follow as H2s, and the individual steps can be broken out with H3s where needed.

People often think headings used out of order hurt Google rankings. Google says the order doesn’t matter from a Search perspective and there is no ideal number of headings, although semantic order is genuinely helpful for screen readers.

Title tagOn-page SEOSEO-friendly content

Google Search Central

Cloaking

Cloaking is the practice of presenting different content to users and search engines with the intent to manipulate rankings and mislead users, and it violates Google’s spam policies. Examples include showing search engines a page on one topic while users see another, or inserting keywords only when the request comes from a search engine.

A site serves Googlebot a travel guide while showing visitors an unrelated product sales page. Because users don’t get the content the search result promised, this is cloaking.

Paywalled content is often thought to count as cloaking. Google says it doesn’t consider a paywall cloaking if Google can see the full content behind it and the site follows its Flexible Sampling guidance.

Black hat SEOGoogle spam policiesGooglebotManual action (Google penalty)

Google Search Central

Image SEO

Image SEO is the work of making images discoverable and understandable in Google Images and web results. It includes short but descriptive file names, meaningful alt text, placing images near relevant text, supported file formats, responsive images and image sitemaps, while making sure large files don’t slow the page down.

A recipe site renames “IMG_0042.jpg” to “oven-roasted-chicken-vegetables.webp”, places the photo beside the matching recipe step and adds it to an image sitemap.

Alt text (alt attribute)Lazy loadingXML sitemapCore Web Vitals

Google Search Central

Internal link

An internal link is a link from one page on a site to another page on the same site. Google uses links as a signal for judging page relevance and to find new pages to crawl. Every page you care about should be linked from at least one other page on the site, with anchor text that describes the destination.

A yoga studio’s “Yoga for beginners” article links to its timetable page within the body text, using the anchor text “beginner class schedule”.

It’s often assumed that anything acting like a link can be followed by Google. Google says it can crawl a link only if it is an a element with an href attribute, so other link-like elements may be missed.

Anchor textOrphan pageTopic clusterLink equity

Google Search Central

Content gap

A content gap is a topic your audience searches for, or that competitors already answer, which your site doesn’t cover or covers poorly. Gaps are usually found through keyword research and competitor analysis. Closing one means more than adding the missing topic: the new page should offer original information and real value compared with what already ranks.

A bike shop’s site has product pages but no guide answering “how to choose the right bike size”. Seeing that competitors cover it, the team publishes an original guide with its own sizing chart.

SEO competitor analysisKeywordTopic cluster

Google Search Central

Content freshness

Content freshness is how well the information on a page reflects the current state of its topic. Google’s ranking systems include freshness systems designed to show more recent content for queries where it would be expected. Not every query needs fresh results, but searches about a newly released film or a recent event clearly do.

A tech site’s “Best mid-range phones” list goes stale as new models launch. The team revises it with new models and fresh testing, and only then changes the updated date.

Some think changing a page’s date makes it look fresh. Google lists updating dates when the content hasn’t substantially changed among the warning signs of search engine-first content.

Evergreen contentHelpful contentSearch algorithm

Google Search Central

Evergreen content

Evergreen content is content whose subject isn’t tied to a particular moment, so it stays relevant and useful for a long time. Explainers, how-to guides and glossaries are typical examples. Unlike news or campaign pages it keeps its relevance, but its details can still date, so it needs periodic review.

A travel blog’s “Documents you need for a passport application” post is searched for year after year. Because it includes details that change, such as fees, the team reviews and updates it at regular intervals.

Content freshnessPillar pageHelpful content

Google Search Central

Doorway page

A doorway page is created to rank for specific, similar search queries and leads users to intermediate pages that are less useful than the final destination. Google’s spam policies define this as doorway abuse. A typical example is a set of pages targeting different cities or regions that all funnel visitors to one page.

A carpet cleaning company publishes 40 pages that differ only by district name, each funneling visitors to the same contact page. This shouldn’t be confused with location pages that offer genuinely local information.

Google spam policiesLocation page (city page)Programmatic SEOBlack hat SEO

Google Search Central

Topic cluster

A topic cluster is a content structure in which a pillar page covering a broad topic and detailed articles on its subtopics are connected through internal links. It is a content planning term, not one defined by Google. The links help readers move between related pages and help Google find those pages and judge their relevance.

A garden supply site builds a hub around its “Growing vegetables on a balcony” guide. Articles on tomatoes, peppers and potting soil link to the guide, and the guide links back to each with descriptive anchor text.

Pillar pageInternal linkAnchor textContent gap

Google Search Central

Meta description

A meta description is an HTML meta tag that summarizes a page’s content in a short, relevant way. Google creates snippets primarily from the page content itself, but it may use the meta description when it gives users a more accurate summary. That is why each page benefits from its own readable, unique description.

Instead of repeating one description across hundreds of product pages, an online store generates descriptions programmatically from its database, combining each product’s name, material and color.

A common belief is that a meta description stops working past a certain character count. Google says there is no limit on its length; the snippet shown in results is truncated as needed, typically to fit the device width.

SnippetTitle tagCTR (click-through rate)On-page SEO

Google Search Central

Scaled content abuse

Scaled content abuse is generating many pages primarily to manipulate search rankings rather than to help users, and it violates Google’s spam policies. The method doesn’t matter: large amounts of unoriginal content that provides little to no value fall under it whether produced with AI, other automation or any other means.

A site stitches together articles from other sources and publishes hundreds of new pages every day. The pages give readers nothing new, which makes this a textbook case of scaled content abuse.

Many think this policy only targets AI content. Google defines scaled content abuse regardless of how the content is created; what counts is whether mass-produced pages add value for users.

AI-generated contentProgrammatic SEOThin contentGoogle spam policies

Google Search Central

Programmatic SEO

Programmatic SEO is an approach that uses a data source and page templates to generate large numbers of pages automatically, for example by city, product or feature combination. If each page doesn’t offer users original, useful information, the approach can drift toward what Google defines as scaled content abuse or doorway abuse.

A currency information site generates a page for each currency pair from a single template, with the current rate, a historical chart and a converter. The template is shared, but each page answers a distinct question with its own data.

Scaled content abuseDoorway pageThin contentLong-tail keyword

Google Search Central

On-page SEO

On-page SEO is the optimization of a page’s own content and HTML elements for search engines. It covers the title tag, meta description, heading structure, image alt text, internal links and the body copy itself. The goal is to help search engines understand what the page is about and help users recognize it in results and decide to visit.

A furniture store gives its “wooden dining tables” category a page-specific title, adds descriptive alt text to the product photos and links from the category copy to its wood care guide with descriptive anchor text.

Many still believe that filling in the keywords meta tag helps rankings. Google’s SEO Starter Guide states that Google Search doesn’t use the keywords meta tag at all.

Title tagMeta descriptionHeading tags (H1–H6)Technical SEO

Google Search Central

SEO-friendly content

SEO-friendly content answers the reader’s question accurately and fully while being structured so search engines can easily understand its topic. It has a scope that matches search intent, descriptive headings, natural language and internal links to related pages. Google recommends people-first content over content made primarily to gain search rankings.

An insurance comparison site opens its “what is comprehensive car insurance” article with a short definition, explains coverage in a table, groups common questions under their own heading and links to its quote comparison page.

Helpful contentSearch intentHeading tags (H1–H6)E-E-A-T

Google Search Central

Snippet

A snippet is the description or summary part of a Google search result, shown below the title link. Google creates snippets primarily from the page content itself and may show different snippets for the same page in different searches. The meta description can be used when it describes the page more accurately than on-page text.

For a search like “when should a boiler be serviced”, Google may show a passage from a heating service page’s paragraph on service intervals as the snippet instead of its meta description.

Many assume the text under a result is simply their meta description. Google builds snippets primarily from page content and uses the meta description only when it thinks it describes the page more accurately.

Meta descriptionFeatured snippetSERP (search engine results page)nosnippet and max-snippet

Google Search Central

Pillar page

A pillar page is a comprehensive guide that covers a broad topic as a whole and links out to detailed pages on each of its subtopics. It sits at the center of a topic cluster. The subpages link back to it, so readers can move from the overview to the details and related pages stay connected across the site.

An accounting software site’s “Bookkeeping for small businesses” guide briefly explains invoicing, expense tracking and cash management, and each section links to a separate article that covers that topic in depth.

Topic clusterInternal linkEvergreen content

Google Search Central

Title tag

A title tag is the HTML title element that names a page. When Google automatically generates the title link for a search result, it may use the title element alongside other sources, such as the main visual title, H1 headings and the anchor text of links pointing to the page. Good titles are descriptive, concise and unique to each page.

If every page on a software company’s site is titled “Home | Company”, the pages are hard to tell apart in results. Page-specific titles such as “Inventory Tracking Software Features | Example Software” solve the problem.

Many assume Google enforces a fixed character limit for titles. Google says there is no limit on how long a title element can be; the title link is simply truncated in results as needed, typically to fit the device width.

Meta descriptionSnippetHeading tags (H1–H6)SERP (search engine results page)

Google Search Central

AI-generated content

AI-generated content is text, images or metadata created wholly or partly with generative AI tools. Google expects it to meet Search Essentials and its spam policies, and stresses manually fact-checking it for accuracy before publishing. That review also applies to titles, meta descriptions, structured data and image alt text.

A cosmetics brand drafts product descriptions with AI. The content team checks ingredient details against the product labels, corrects errors and adds usage tips based on its own experience.

A common belief is that Google penalizes content simply for being written with AI. Its spam policy targets mass-produced content that adds little value, no matter how it’s created, and Google notes AI can help with research and structure.

Scaled content abuseHelpful contentGenerative AIHallucination (AI)

Google Search Central

Orphan page

An orphan page is a live page that receives no internal links from any other page on the same site. Visitors can’t reach it by browsing, and because Google also uses links to discover pages, it may only be found through other routes such as a sitemap. Google recommends linking every important page from at least one other page.

After a “Back to school deals” page is removed from the menu, no other page links to it. A site crawl surfaces the orphan, and the team links it from the relevant category page.

Internal linkXML sitemapCrawlingSEO audit

Google Search Central

Thin content

Thin content is page content that offers users no original information or meaningful added value. Typical examples are product descriptions copied from elsewhere, summaries compiled without adding anything, and near-identical template pages. Google’s spam policies specifically address affiliate pages that copy merchant descriptions without original content, as well as mass-produced content that adds no value.

An electronics store copies manufacturer descriptions for 300 products word for word. The team then adds its own test notes, use cases and comparison tables to priority products to give those pages original value.

Thin content is often equated with short content, as if adding words fixes it. Google says it has no preferred word count; what matters is the value a page offers users, not its length.

Helpful contentDuplicate contentScaled content abuseGoogle spam policies

Google Search Central

Backlink

A backlink is a link from a page on another website to a page on your site, also called an inbound or incoming link. Google follows links to discover new pages and uses them as a signal when determining how relevant a page is. Backlinks are among the main things tracked in off-page SEO.

When a food blog links from one of its recipes to a kitchenware store’s cast-iron pan page, that link is a backlink for the store.

Off-page SEOReferring domainLink buildingAnchor text

Google Search Central

Link gap

A link gap is a comparison that reveals domains linking to your competitors but not to you. It is run with link data from SEO tools and suggests which sources may be worth approaching. Google describes links as one of the signals it uses when determining how relevant pages are.

A yoga studio lists wellness blogs that link to its three competitors, then contacts those that don’t link to it yet with a content idea about its new class program.

SEO competitor analysisReferring domainLink buildingContent gap

Google Search Central

Link equity

Link equity is an industry term for the ranking value a link is assumed to pass to the page it points to, often called link juice. Google’s spam policies speak of links that pass ranking credit, and ask that paid links be qualified with sponsored or nofollow.

A news site adds sponsored to a link in a piece it runs as advertising, so the paid link is marked as one that should not pass ranking credit.

PageRankNofollow (rel="nofollow")rel="sponsored"Backlink

Google Search Central

Link scheme

A link scheme is the practice of creating links to or from a site primarily to manipulate search rankings, which Google’s spam policies call link spam. The policy’s examples include buying or selling links for ranking purposes, excessive link exchanges, links created with automated programs and links widely distributed in the footers of various sites.

Fifteen unrelated sites set up “our partners” pages purely to link to one another. That arrangement matches Google’s example of excessive link exchanges.

Many believe any paid link breaks Google’s rules. Google accepts buying and selling links for advertising and sponsorship as a normal part of the web’s economy, provided the links are qualified with sponsored or nofollow.

PBN (private blog network)Google spam policiesrel="sponsored"Manual action (Google penalty)

Google Search Central

Anchor text

Anchor text is the visible text of a link, also called link text. Google describes good anchor text as descriptive, reasonably concise and relevant both to the page it sits on and to the page it links to. Text like that helps people and search engines understand what the linked page contains.

In a bike maintenance guide, linking with “how to lubricate a chain” rather than “here” tells readers and search engines exactly what the target page covers.

Many assume that packing anchor text with keywords helps. Google advises writing naturally and resisting the urge to cram every related keyword into the link text.

BacklinkInternal linkKeyword stuffing

Google Search Central

Digital PR

Digital PR is the work of earning coverage for a brand in online publications through newsworthy content, original research and expert commentary. Its goals are visibility, reputation and naturally earned links. Google’s spam policies list links with optimized anchor text in press releases distributed on other sites among examples of link spam.

A courier company shares anonymized data on holiday shipping peaks with journalists. Stories that use the data cite the company as the source, and some of them link to it.

Link buildingBrand mentionOff-page SEO

Google Search Central

Disavow

Disavow is a Search Console tool for submitting a list of questionable pages or domains linking to your site that you want to reject. Google calls it an advanced feature that can harm search performance if used incorrectly, and limits it to cases where many spammy links have caused, or are likely to cause, a manual action.

After a manual action, a site lists the spammy links it could not get removed in a UTF-8 .txt file, one URL or domain: entry per line, and uploads it in Search Console.

Disavowing is often treated as routine maintenance for every site. Google says that in most cases it can assess which links to trust on its own, so most sites will not need the tool.

Toxic backlinkManual action (Google penalty)Google Search ConsoleSpam Score

Search Console Yardım

Dofollow link

A dofollow link is the informal SEO name for a standard link that carries no qualifying rel value such as nofollow, sponsored or ugc. HTML has no “dofollow” rel value. Google says that links you expect it to fetch and parse without any qualification need no rel attribute at all.

A city guide links to a café its editor recommends with a plain link and no rel attribute. SEO tools list that link as dofollow.

Many people think a link needs rel=“dofollow” to be treated normally. No such rel value exists, and Google says links that need no qualification need no rel attribute.

Nofollow (rel="nofollow")rel="sponsored"rel="ugc"Link equity

Google Search Central

Domain Authority (DA)

Domain Authority (DA) is a Moz metric that predicts how likely a domain is to rank in search results compared with its competitors. It runs from 1 to 100, with higher scores indicating greater ranking potential. Moz calculates it with a machine learning model based mainly on link data, especially linking root domains.

A dental clinic’s DA only means something when compared with other clinics’ sites in the same city; on its own, the score is neither good nor bad.

DA is often mistaken for a Google ranking factor. Moz describes it as its own predictive metric and says Google does not use it to determine rankings.

Page Authority (PA)Domain Rating (DR)Spam ScoreReferring domain

Moz

Domain Rating (DR)

Domain Rating (DR) is an Ahrefs metric that shows the relative strength of a website’s backlink profile on a logarithmic scale from 0 to 100. The calculation looks at domains with at least one followed link to the site, the DR of each of those domains and how many other sites each one links to.

An online store comparing two blogs it would like links from checks their DR, then weighs the score alongside each blog’s topic and readership.

A high DR is often read as proof of a clean link profile. Ahrefs notes that DR doesn’t account for backlink spam and that link spam has been used to inflate it.

Domain Authority (DA)Referring domainPageRank

Ahrefs

Guest posting

Guest posting is publishing an article under your own name on someone else’s website. It is a way to bring expertise to a new audience. Google’s spam policies list links with optimized anchor text in guest posts distributed on other sites, and paid links that pass ranking credit, among examples of link spam.

An occupational safety specialist writes a guest post on scaffold inspections for a construction industry blog, with a descriptive link to their own site in the author bio.

Link buildingAnchor textSponsored article (advertorial)Link scheme

Google Search Central

Link building

Link building is the work of getting other websites to link to your pages, also known as link acquisition. Common approaches include original research, reference-worthy content and digital PR. Google’s spam policies define creating links primarily to manipulate search rankings as link spam.

An energy consultancy publishes a report on regional electricity use. News sites and blogs that cite its figures link to the report page as their source.

BacklinkDigital PRLink schemeLink gap

Google Search Central

Brand mention

A brand mention is any reference to a brand’s name on another website, news story or forum, with or without a link. Google’s Search Quality Rater Guidelines ask raters researching a site’s reputation to look at what outside, independent sources such as news articles, reviews and expert recommendations say, rather than at what the site says about itself.

A design studio is named, without a link, in a city guide’s roundup of standout local studios. That is a brand mention, and the studio may ask the writer to add a link.

Digital PRE-E-A-TOff-page SEOEntity

Google

Nofollow (rel="nofollow")

Nofollow is a rel attribute value telling Google that you would rather not have your site associated with the linked page, or have that page crawled from your site. Google recommends it when the more specific sponsored and ugc values don’t apply. Links marked with these values are generally not followed.

A forum moderator marks a link to a site they don’t trust with nofollow instead of deleting it, signalling that the forum doesn’t vouch for that page.

A common belief is that nofollow stops the target page from being crawled or indexed. Google says the page may still be found through sitemaps or other links and crawled; noindex is the way to keep it out of the index.

Dofollow linkrel="sponsored"rel="ugc"noindex

Google Search Central

Page Authority (PA)

Page Authority (PA) is a Moz metric that predicts how well an individual page is likely to rank in search results. It is scored from 1 to 100 on a logarithmic scale and calculated the same way as Domain Authority, with a machine learning model, but for a single page rather than a whole domain.

A content team compares the PA of two guides on the same topic to see which page has the stronger link profile.

People often look for a universally “good” PA. Moz says the metric is comparative, has no fixed good or bad value and can fluctuate as its algorithm is updated.

Domain Authority (DA)Domain Rating (DR)Backlink

Moz

PageRank

PageRank is the ranking system Google has used since it first launched to assess how pages link to one another. Google lists it among the link analysis systems that help it understand what pages are about, and says that how PageRank works has evolved a lot while it remains part of its core ranking systems.

When a museum’s exhibition guide links to an entry in an art history glossary, that link becomes part of the web of links Google analyzes to understand how pages relate.

PageRank is often thought to be retired. Google says the way it works has evolved a lot since launch, but it continues to be part of its core ranking systems.

Link equitySearch algorithmRankingBacklink

Google Search Central

PBN (private blog network)

A PBN (private blog network) is a group of websites controlled by one person or group and run mainly to link to target sites. The sites may look like independent publications, but their content typically exists to carry links. Google’s spam policies treat creating links, or low-value content, primarily to manipulate rankings as link spam.

Someone buys several expired domains, adds short posts to each and links every post to the same product page. The result is a typical PBN.

Link schemeExpired domain abuseBlack hat SEOGoogle spam policies

Google Search Central

rel="sponsored"

rel=“sponsored” is the rel value added to a link to mark it as an advertisement or paid placement. Google says it prefers this value for paid links. Under Google’s spam policies, links bought or sold for advertising or sponsorship are not a violation when they are qualified with sponsored or nofollow.

A tech blog publishes a headphone review the manufacturer paid for and marks the link to the product page with sponsored.

Some assume nofollow is no longer acceptable on paid links. Google says sponsored is preferred, but nofollow is still an acceptable way to flag them.

Nofollow (rel="nofollow")rel="ugc"Sponsored article (advertorial)Link scheme

Google Search Central

rel="ugc"

rel=“ugc” is the rel value used to mark links inside user-generated content, such as comments and forum posts. Google recommends qualifying these links with ugc, and notes that sites may choose to remove it for members who have consistently made high-quality contributions over time.

A photography forum adds ugc to links in members’ posts automatically, but leaves it off links from moderators who have shared high-quality content for years.

Nofollow (rel="nofollow")rel="sponsored"Dofollow link

Google Search Central

Off-page SEO

Off-page SEO is the work of supporting a site’s search visibility through factors outside the site itself. Links from other websites, brand mentions in independent sources, digital PR and recognition within a community all fall under it. Google notes that it primarily finds pages through links from pages it has already crawled.

An architecture practice gets a project featured in a trade magazine, speaks at an industry event and is interviewed by the local press. All of that visibility happens off its own site, which is what off-page SEO covers.

On-page SEOBacklinkDigital PRBrand mention

Google Search Central

Site reputation abuse (parasite SEO)

Site reputation abuse, often called parasite SEO, is publishing third-party content on a host site mainly to benefit from ranking signals the host earned with its own content. Google addresses it under the site reputation policy in its spam policies, which notes that hosting third-party content is not in itself a violation.

An educational site hosting sponsored payday loan reviews written by a third party that distributes the same page to other sites is one of the examples in Google’s policy.

Google spam policiesSponsored article (advertorial)Expired domain abuseBlack hat SEO

Google Search Central

Spam Score

Spam Score is a Moz metric showing the percentage of sites with features similar to a given site that Moz has found to be penalized or banned by Google. It comes from a machine learning model trained on features common to such sites, and Moz classes 61% to 100% as a high score.

A backlink audit shows a high Spam Score for one linking domain. Rather than disavowing it outright, the team reviews the site’s content and context.

A high Spam Score is often taken as proof that a site is spam. Moz explains that it reflects correlation rather than causation, so disavowing links from high-scoring sites isn’t automatically the answer.

Toxic backlinkDisavowDomain Authority (DA)

Moz

Expired domain abuse

Expired domain abuse is buying an expired domain name and repurposing it primarily to manipulate search rankings with content that offers users little to no value. It is part of Google’s spam policies, whose examples include casino content on a former elementary school site and affiliate content on a site once used by a government agency.

Someone buys the old domain of a non-profit medical charity and uses it to sell commercial health products. That matches one of the examples in Google’s policy.

PBN (private blog network)Site reputation abuse (parasite SEO)Google spam policies

Google Search Central

Sponsored article (advertorial)

A sponsored article, or advertorial, is editorial-style content that a brand pays to have published on another site or publication. Google’s spam policies list paid advertorials containing links that pass ranking credit among examples of link spam, and ask that such links be qualified with sponsored or nofollow.

A garden equipment maker pays a home and lifestyle magazine to run a watering guide on its site, and the link to the product page is marked sponsored.

Some think publishing an advertorial breaks Google’s policies in itself. Google does not treat advertorials meant to reach readers, rather than manipulate rankings, as a site reputation violation, but expects their paid links to be qualified.

rel="sponsored"Link schemeSite reputation abuse (parasite SEO)Digital PR

Google Search Central

Referring domain

A referring domain is a unique domain with at least one link pointing to your site. Many links from the same site still count as one referring domain, so the figure shows how many different sources link to you. It is a key input in link metrics such as Moz’s Domain Authority and Ahrefs’ Domain Rating.

If a travel blog links to a hotel’s site from twenty different posts, the hotel gains twenty backlinks but only one referring domain.

BacklinkDomain Authority (DA)Domain Rating (DR)Link gap

Google Search Central

Toxic backlink

A toxic backlink is an industry term for a spammy, artificial or low-quality link pointing to a site. Google’s spam policies treat links created primarily to manipulate rankings as link spam. Where a site has a considerable number of such links that have caused, or are likely to cause, a manual action, Google points to the disavow tool.

A furniture store notices hundreds of links from unrelated low-quality directories, all with the same anchor text, and first asks for them to be removed.

A link flagged as toxic by an SEO tool is often assumed to be penalized by Google. Such flags come from the tool’s own model; Moz, for example, says its Spam Score is based on correlation.

DisavowSpam ScoreLink schemeManual action (Google penalty)

Google Search Central

Primary business category

The primary business category is the main category on a Google Business Profile that best describes what the business is as a whole. Google’s guidelines ask owners to use as few categories as possible, choose the most specific ones that fit, and pick categories that complete the sentence “this business is a…”, rather than adding keywords or features.

A dental clinic picks the specific dentistry category that best describes it, and lists amenities such as free parking or a children’s play area as profile attributes rather than categories.

Adding as many categories as possible is often seen as a way to show up for more searches. Google asks for as few categories as needed to describe the core business, and notes that choosing a specific category already covers the more general ones.

Google Business ProfileLocal ranking factors (relevance, distance, prominence)Local SEO

Google İşletme Profili Yardım

Google Business Profile

Google Business Profile is a free listing that shows a business’s name, address, phone number, opening hours, category and photos on Google Search and Maps. Owners can create a new profile or claim an existing one, and verification shows Google they are authorised to manage it. Google’s guidelines allow only one profile per business.

A newly opened café in Izmir creates and verifies its profile, uses the exact name on its sign, and adds its opening hours and interior photos, so people searching on Maps find accurate details.

Adding a neighbourhood or service to the business name is a popular trick for more visibility. Google’s guidelines require the name used on the storefront, website and stationery, and do not allow unnecessary information in it.

Local SEONAP consistencyPrimary business categoryBusiness reviews (Google reviews)

Google İşletme Profili Yardım

Geo-grid rank tracking

Geo-grid rank tracking measures a business’s local ranking from many points laid out across a map rather than from a single location. Because Google factors in how far a searcher is from each business, the same query can rank differently from one neighbourhood to the next. The grid shows where a business is strong and where its visibility fades.

A dental clinic in Kadıköy tracks the query “dentist” from points spread around its location; it may rank near the top in nearby streets but slip further down a few kilometres away.

Rank trackingLocal pack (map pack)Local ranking factors (relevance, distance, prominence)“Near me” searches

Google İşletme Profili Yardım

Service-area business

A service-area business serves customers at their locations rather than at its own address; plumbers, locksmiths and home cleaning services are typical examples. On Google Business Profile, these businesses hide their address and set the area they cover. Businesses that also welcome customers on site can show their storefront address and define a service area as well.

A boiler repair technician working from home leaves the home address off the profile, lists the districts covered as the service area, and can then appear in searches across those districts.

Drawing the widest possible service area can look like an easy way to reach more searches. Google’s guidelines say the boundaries of a service area should not extend beyond about two hours of driving time from where the business is based.

Google Business ProfileLocal pack (map pack)Location page (city page)NAP consistency

Google İşletme Profili Yardım

Business reviews (Google reviews)

Business reviews are the ratings and comments customers leave about a business on Google, shown next to its Business Profile in Maps and Search. Google says more reviews and positive ratings can help local ranking. Businesses can invite reviews with a link or QR code and reply to them briefly and politely.

A car repair shop prints a QR code linking to its review page on every invoice, and answers a negative review with a short note explaining how it fixed the problem, without sharing any of the customer’s personal details.

Offering a discount or small gift in exchange for a review can seem harmless. Google treats free or discounted goods or services offered for reviews as fake and misleading content, and does not allow it.

Google Business ProfileLocal ranking factors (relevance, distance, prominence)Local SEO

Google İşletme Profili Yardım

Location page (city page)

A location page, or city page, is a page on a business website dedicated to a specific branch or service area, with details unique to that place. Address, opening hours, directions, local services and the on-site team make it genuinely useful. Near-identical pages that only swap the city name, however, can fall under Google’s doorway abuse policy.

A physiotherapy group with branches in three cities gives each branch its own page, listing that branch’s address, map, practitioners and the treatments it offers.

Spinning up pages for every district with only the district name changed is a common local tactic. Google defines pages that target specific regions or cities and funnel users to one page as doorway abuse.

Doorway pageService-area businessLocal SEOLocalBusiness schema

Google Search Central

LocalBusiness schema

LocalBusiness schema is a schema.org structured data type that tells search engines a business’s name, address, opening hours, phone number and location in machine-readable form. Google lists name and address as required properties, with geo coordinates, telephone, url and opening hours among those recommended, and advises using the most specific subtype available, such as Restaurant.

A restaurant adds Restaurant markup in JSON-LD to its contact page, keeping the name, address, phone number and opening hours identical to its Business Profile.

Adding LocalBusiness markup is often expected to make business details appear in a special format in search results. Google makes no promise that features using structured data will show up.

Structured dataJSON-LDNAP consistencyGoogle Business Profile

Google Search Central

NAP consistency

NAP consistency means a business’s name, address and phone number appear the same way everywhere: on its Google Business Profile, its website, directories and social accounts. Google’s guidelines ask businesses to present themselves as they are consistently known in the real world, and conflicting details can mislead both customers and search engines.

After moving, a veterinary clinic updates its new address with identical wording on its Business Profile, its website footer, its contact page and every directory that lists it.

Google Business ProfileLocal citationLocalBusiness schemaLocal SEO

Google İşletme Profili Yardım

“Near me” searches

“Near me” searches are queries such as “pharmacy near me” or “breakfast near me”, where people look for nearby businesses or services. Distance is weighed together with relevance and prominence. When a searcher does not share their location, Google uses what it knows about it, and the results appear as local listings on Maps and Search.

Someone in Beşiktaş searching for a vegetarian restaurant near me at lunchtime sees nearby places with up-to-date profiles laid out on a map.

Local ranking factors (relevance, distance, prominence)Local pack (map pack)Google Business ProfileSearch intent

Google İşletme Profili Yardım

Local citation

A local citation is any mention of a business’s name, address and phone number on another website, such as a directory, an industry platform or a local news site, with or without a link. Google says it can pick up category information from mentions across the web, and that prominence takes into account how many websites link to it.

A physiotherapy centre in Eskişehir makes sure its listings on a health directory and a local business directory show its name, address and phone number exactly as they appear on its Business Profile.

NAP consistencyGoogle Business ProfileBrand mentionBacklink

Google İşletme Profili Yardım

Local pack (map pack)

The local pack, or map pack, is the block of local business results shown alongside a map on a Google results page for location-based searches. Details such as each business’s name, rating, address and hours come from its Google Business Profile, and which businesses appear depends on relevance, distance and prominence.

Someone in Kadıköy searching for a pide restaurant sees a map with a handful of places, their ratings and opening hours, right on the results page; that block is the local pack.

Local SEOGoogle Business ProfileLocal ranking factors (relevance, distance, prominence)Geo-grid rank tracking

Google İşletme Profili Yardım

Local SEO

Local SEO is the work of making a business more visible in location-based searches, on Google Maps and in the local sections of search results. Its core elements are a complete and accurate Google Business Profile, active review management and clear location details on the website. Google determines local results using relevance, distance and prominence.

An accounting office in Çankaya keeps its Business Profile hours and category up to date, replies to customer reviews and adds a contact page with its office address to its website.

Some business owners believe they can pay Google for a higher spot in map results. Google states plainly that there is no way to request or pay for a better local ranking.

Google Business ProfileLocal pack (map pack)Local ranking factors (relevance, distance, prominence)NAP consistency

Google İşletme Profili Yardım

Local ranking factors (relevance, distance, prominence)

Local ranking factors are the three main criteria Google uses for local results: relevance, distance and prominence. Relevance is how well a profile matches the search, distance is how far the business is from the searcher, and prominence is how well known it is, informed partly by how many websites link to it and how many reviews it has.

A bakery slightly further from the centre, with a complete profile, plenty of reviews and a strong local reputation, can appear ahead of a closer competitor whose details are patchy.

It is often assumed that the closest business takes the top local spot. Google weighs distance together with relevance and prominence, and better-known businesses are more likely to show up.

Local pack (map pack)“Near me” searchesBusiness reviews (Google reviews)Local citation

Google İşletme Profili Yardım

Key event (conversion)

A key event is a Google Analytics 4 event marked as especially important to the business, typically a conversion moment such as a form submission or purchase. An event can be marked as a key event when it is created, with a default value and a counting method. Google Ads conversions can also be created from key events.

A dental clinic tracks appointment form submissions with an event named appointment_form and marks it as a key event, so appointment requests from organic search can be followed on their own.

Conversion rateGoogle Analytics 4 (GA4)Engagement rate

Google Analytics Yardım

Keyword difficulty (KD)

Keyword difficulty (KD) is a score that SEO tools calculate to estimate how hard it will be to rank in the top organic results for a query. Because each tool uses its own data and method, the same keyword can score differently across tools. It is used with search volume and intent to decide which keywords to prioritise.

A newly launched travel blog passes on the high-difficulty “Italy trip” and starts with lower-scoring long-tail queries such as “two-day itinerary in Florence”.

Search volumeKeywordSEO competitor analysisLong-tail keyword

Google Ads Yardım

Search volume

Search volume is an estimate of how many times a keyword is searched over a period, usually per month. Keyword Planner in Google Ads provides estimates of monthly searches, and keywords with very low volume or considered sensitive may not be available. It helps prioritise topics by the demand behind them.

A garden supplies store compares monthly volume estimates for “drip irrigation kit” and “how to install drip irrigation”, then sets separate targets for the product page and a how-to guide.

Search volume is often treated as an exact count. Keyword Planner gives estimates of monthly searches; the figures are for planning and don’t translate directly into the traffic a site will receive.

KeywordLong-tail keywordKeyword difficulty (KD)

Google Ads Yardım

Bing Webmaster Tools

Bing Webmaster Tools is Microsoft’s service for site owners, combining performance data for a site in Bing search with free SEO tools. It includes Site Scan, an on-demand technical audit, plus backlink and keyword research reports. Verified sites and submitted sitemaps can be imported from Google Search Console, although analytics data is not.

A furniture store imports its verified property from Search Console into Bing Webmaster Tools, bringing its sitemap along, then runs Site Scan to list technical issues on the Bing side.

Google Search ConsoleIndexNowXML sitemap

Bing

Bot traffic

Bot traffic is requests that come from automated software rather than people, whether to a website or a search engine. Alongside legitimate bots such as search crawlers, it includes fake visits that inflate analytics data. Google treats sending automated queries to its Search without permission, including scraping results for rank checking, as a violation of its spam policies.

On a news site, sessions from one source jump tenfold overnight while engagement rate stays close to zero. The team doesn’t report the spike as organic growth and looks into its source separately.

SessionEngagement rateRank trackingGoogle spam policies

Google Search Central

CTR (click-through rate)

CTR (click-through rate) is the ratio of clicks to impressions, showing what share of the people who saw a result went on to visit it. Search Console reports it for organic search results, and Google Ads for ads and free product listings. It helps judge how relevant and appealing a result looks to searchers.

A bike shop’s “city bikes” page receives 1,000 impressions and 30 clicks; dividing 30 by 1,000 gives a CTR of 3%.

Many look for one “good CTR” that fits every site and query. Google Ads Help calls a good CTR relative to what you advertise and on which networks, and organic results also vary with time, place, device and search history.

ImpressionAverage positionTitle tagMeta description

Search Console Yardım

Direct traffic

Direct traffic is the sessions that Google Analytics 4 reports with the source and medium “direct / (none)”, meaning no referring source is attached to them. Visits where someone types the address straight into the browser are the typical case. It is reported separately from organic search, paid ads, email and referral traffic.

Visitors who type a café chain’s address from a printed flyer land in the “direct / (none)” row. Once the flyer carries a short link with UTM parameters, those visits show up under the campaign name.

Organic trafficUTM parametersSession

Google Analytics Yardım

Conversion rate

Conversion rate is the share of sessions or users that complete a defined goal out of all sessions or users. In Google Analytics 4 the goal is usually set up as a key event, such as a purchase, form submission or quote request. It shows whether visits are useful to the business, not just how many there are.

A pet supplies store records 2,000 organic sessions and 50 purchases; dividing 50 by 2,000 gives the organic channel a conversion rate of 2.5%.

Key event (conversion)SessionLanding pageSEO KPI

Google Analytics Yardım

Engagement rate

Engagement rate is the percentage of sessions in Google Analytics 4 that were engaged. A session counts as engaged if it lasts longer than 10 seconds, has a key event, or has two or more page or screen views. It shows how many visits involved meaningful engagement and is the inverse of bounce rate.

An accounting firm’s site logs 200 sessions in a week, 120 of them engaged. Dividing 120 by 200 gives a 60% engagement rate, which leaves a 40% bounce rate.

It’s easy to assume that any event marked as a key event makes a session engaged. Events such as first_visit, first_open and session_start are excluded from the engaged-session calculation even when marked as key events.

Bounce rateSessionKey event (conversion)

Google Analytics Yardım

Google Analytics 4 (GA4)

Google Analytics 4 (GA4) is Google’s analytics platform that measures user interaction on websites and apps through sessions and events. It collects a session_start event automatically when a session begins, and reports metrics such as engaged sessions, engagement rate, bounce rate and key events to show what visitors do on a site.

For a yoga studio’s website, GA4 shows sessions from organic search, how many of them were engaged and how often the trial class form was submitted, all within one property.

SessionEngagement rateKey event (conversion)Google Search Console

Google Analytics Yardım

Google Search Console

Google Search Console is a free Google service that helps site owners monitor and troubleshoot how their site appears in Google Search. It confirms that Google can find and crawl the site, surfaces indexing problems, shows which queries bring impressions and clicks, and sends alerts about indexing, spam or other issues.

Once a furniture store adds its site to Search Console, the Performance report shows impressions and clicks for “wooden dining table”, while the Page indexing report lists the category pages that are not indexed.

Many assume a site must be registered in Search Console to appear on Google. Google states that signing up isn’t required for inclusion; the tool helps you understand how Google sees the site.

URL Inspection toolPage indexing reportImpressionBing Webmaster Tools

Search Console Yardım

Share of voice

Share of voice is a measure of how much of the total search visibility for a defined set of queries a brand captures. It is usually calculated by comparing rankings or impressions on tracked keywords with competitors, and the method varies by tool. It describes a brand’s relative standing in a market rather than the rank of one keyword.

A coffee machine brand tracks a set of 50 keywords. When it captures a quarter of the total visibility in that set, its share of voice is reported as 25%.

Rank trackingImpressionSEO competitor analysisAI visibility

Search Console Yardım

Impression

An impression is the Search Console metric counting how many times a site appeared in Google Search results. It is the denominator of CTR and the first measure of visibility. Data grouped by query, country, device or date is aggregated by property, while data grouped by page or search appearance is aggregated by page.

A stationery shop gets 2,000 impressions and 40 clicks for “fountain pen” in a month; dividing 40 by 2,000 shows that 2% of impressions turned into clicks.

Two results from one site for the same query are often assumed to count twice. In the property-level chart total they count as a single impression, while data grouped by page is aggregated page by page.

CTR (click-through rate)Average positionGoogle Search ConsoleSERP (search engine results page)

Search Console Yardım

Bounce rate

Bounce rate is the percentage of sessions in Google Analytics 4 that were not engaged, making it the inverse of engagement rate. A session counts toward it when it lasts no longer than 10 seconds, has no key event and has fewer than two page or screen views. It shows the share of visits without meaningful engagement.

A visitor views a florist’s homepage for eight seconds and leaves without doing anything else. The session isn’t engaged; if it were the site’s only session, bounce rate would be 100%.

Every single-page visit is often counted as a bounce. In GA4, a one-page session that lasts longer than 10 seconds or includes a key event is engaged and doesn’t count toward bounce rate.

Engagement rateSessionGoogle Analytics 4 (GA4)Key event (conversion)

Google Analytics Yardım

Organic traffic

Organic traffic is the visits a site receives from the unpaid results of search engines. In Google Analytics 4 it appears with the “organic” medium in session-scoped source dimensions, and in user-scoped “First user” dimensions as users whose first visit came from organic search. It is tracked apart from ads, email and referrals.

On a language school site, a user first arrives through organic Google search, then starts a later session from a newsletter. First user source shows “google / organic”, while the second session shows “email / newsletter”.

Organic resultDirect trafficSessionGoogle Analytics 4 (GA4)

Google Analytics Yardım

Average position

Average position is the Search Console metric giving the average position of a site’s topmost result in Google Search. In the chart it covers the whole site; in the table it applies to the URL or query in that row. Because results depend on time, place, device and search history, it summarises a period rather than one fixed rank.

A hotel site shows an average position of 4.2 for “boutique hotel”. That doesn’t mean every searcher saw it in fourth place; it is the average of its topmost result over the period.

Average position is often read as the average of all of a site’s results for a query. Search Console uses the position of the site’s topmost result, so a second, lower listing doesn’t pull the figure down.

ImpressionRank trackingCTR (click-through rate)SERP (search engine results page)

Search Console Yardım

Session

A session in Google Analytics 4 is a period during which a user interacts with a website or app. It starts when a page is viewed while no session is active and, by default, ends after 30 minutes of inactivity, a timeout that can be adjusted. The session_start event collected at the beginning generates the session ID and session number.

On a recipe site, a user reads one recipe in the morning and comes back for another in the afternoon. With more than 30 minutes of inactivity in between, GA4 records two separate sessions.

Sessions are often treated as the number of visitors. One user can start several sessions, which is why GA4 keeps a session number counting how many sessions each user has started.

Google Analytics 4 (GA4)Engagement rateBounce rate

Google Analytics Yardım

Field data and lab data

Field data and lab data are two ways of measuring page performance: field data comes from real users, while lab data measures the same metrics in a simulated environment without a user. The Chrome UX Report (CrUX) and Search Console’s Core Web Vitals report provide field data; Lighthouse provides lab data. Assessment uses the 75th percentile of page loads.

A real estate site scores well in a Lighthouse test, yet Search Console’s Core Web Vitals report rates its mobile URL group as Poor. The team gives priority to the real-user data.

A strong lab score is often taken to mean real users have a good experience too. Tools that run without a user, such as Lighthouse, can’t measure INP; only field measurement captures the complete picture.

Core Web VitalsINP (Interaction to Next Paint)LCP (Largest Contentful Paint)CLS (Cumulative Layout Shift)

web.dev

Page indexing report

The Page indexing report is a Search Console report showing which pages Google can find and index on a site and which indexing problems it encountered. URLs are split into indexed and not indexed, with reasons such as “Crawled - currently not indexed” or a noindex directive. Fixed issues can be sent to Google for validation.

On a sample online store, the report lists filtered listing URLs as “Alternate page with proper canonical tag”. Since the canonical pages are indexed, the team leaves that row alone.

It’s tempting to aim for every URL on a site to be indexed. Google advises not to expect that; the goal is to get the canonical version of every important page indexed.

Google Search ConsoleURL Inspection toolnoindexCanonical tag (canonical URL)

Search Console Yardım

SEO audit

An SEO audit is a systematic review of the issues that stop a site from being found, crawled, indexed and properly understood by search engines. It covers technical setup, on-page elements, content and links, and checks whether Google sees pages the same way users do and whether titles and meta descriptions are unique to each page.

An audit of an architecture firm’s site finds that project pages share one title and some images are blocked by robots.txt; the findings are then ranked by priority.

Some audits flag a missing meta keywords tag as a problem. Google Search doesn’t use the keywords meta tag, so audit effort is better spent on unique titles, descriptive meta descriptions and crawlability.

Technical SEOPage indexing reportURL Inspection toolSEO competitor analysis

Google Search Central

SEO KPI

An SEO KPI is a key performance indicator chosen to track how organic search work contributes to business goals. Common examples are impressions, clicks, CTR and average position in Search Console, alongside organic sessions and key events in Google Analytics 4. A good KPI states clearly which stage, from visibility to conversion, it measures.

A law firm’s site picks three KPIs: impressions on non-branded queries, organic clicks, and consultation form submissions from organic search. Each report compares the three with the previous period.

ImpressionCTR (click-through rate)Conversion rateShare of voice

Search Console Yardım

SEO competitor analysis

SEO competitor analysis is the study of sites that rank well for target queries to find gaps in content, technical setup and links. It compares which queries competitors appear for, which visual elements their results show and which topics they cover. Because results look different by device, country and language, comparisons are made under the same conditions.

An organic skincare brand notices that the top three results for “moisturiser for sensitive skin” show review stars, and plans structured data for its own product page.

SERP (search engine results page)Keyword difficulty (KD)Share of voiceContent gap

Google Search Central

Rank tracking

Rank tracking is the regular monitoring of a site’s position in search results for a chosen set of keywords. Search Console’s average position summarises a period, while rank tracking tools show the rank of individual keywords for a selected location and device. Read together, the two give a clearer view of changes in visibility.

A plumbing business in Izmir tracks “boiler service” weekly for the Izmir location on mobile, then checks any drop in rank against its clicks in Search Console.

The rank you see in your own browser is often taken as everyone’s rank. Results depend on time, place, device and the searcher’s recent history, so a single manual check doesn’t reflect the overall position.

Average positionSERP (search engine results page)Share of voice

Search Console Yardım

URL Inspection tool

The URL Inspection tool is a Search Console feature that reports on Google’s indexed version of a single page. Its results reflect the most recently indexed version, while the live test fetches the URL in real time to check whether it may be indexable. It can also send an indexing request, subject to a daily limit.

A food blog updates its “coffee brewing methods” post. The indexed version still shows the old title, the live test picks up the new one, and the editor then submits an indexing request.

A valid live test is often read as a promise of indexing. It only shows that Google can access the page; other conditions such as duplication and quality still apply, and an indexing request doesn’t ensure inclusion either.

Google Search ConsolePage indexing reportIndexingCanonical tag (canonical URL)

Search Console Yardım

UTM parameters

UTM parameters are tags added to campaign and referral links that tell Google Analytics which source, medium and campaign a visit came from. Google recommends always using utm_source, utm_medium and utm_campaign, with parameters such as utm_term and utm_content adding detail. Missing parameters lead to “(not set)” values in reports.

A bookshop’s newsletter link ends in ?utm_source=newsletter&utm_medium=email&utm_campaign=summer_sale, so visits from that newsletter appear in GA4 under the campaign name.

Capitalisation in UTM values is often assumed not to matter. Values are case sensitive, so “google” and “Google” count as different sources, and using lowercase consistently is the recommended practice.

Direct trafficGoogle Analytics 4 (GA4)Organic traffic

Google Analytics Yardım

AEO (answer engine optimization)

AEO (answer engine optimization) is an industry term for work aimed at having content surface as the direct answer to a question, typically in featured snippets, voice assistants and AI-generated answers. It is not an official Google term, and its meaning overlaps heavily with GEO.

A garden supply site places a two-sentence answer directly under the heading “when should tomato seedlings be planted?”, then covers the details and regional differences below it.

Many assume special markup can get a page chosen as a featured snippet. Google says no such markup exists; its systems decide whether a page would make a good featured snippet for a search.

GEO (generative engine optimization)Featured snippetPeople Also Ask (PAA)AI Overviews

Google Search Central

AI Mode

AI Mode is Google’s AI search experience for questions that need further exploration, reasoning or complex comparisons, letting people go deeper through follow-up questions. Google says it uses a query fan-out technique that breaks a question into subtopics, and it shows helpful links to the web. Eligibility to appear as a source is the same as for AI Overviews.

A user asks AI Mode which electric cars suit a family of three for city driving with a large boot, then narrows the search with the follow-up “which of these charges the fastest?”

AI OverviewsQuery fan-outConversational search

Google Search Central

AI Overviews

AI Overviews is a Google Search feature that, for some queries, uses AI to summarize the gist of a complex topic and links to supporting web pages. Google says they appear only when its systems judge them additive to classic Search. To appear as a link, a page must be indexed and eligible to show with a snippet.

A search for “why does boiler pressure drop?” returns an AI summary of likely causes, with a heating brand’s maintenance guide and a manufacturer’s support page shown as supporting links.

Some believe an llms.txt file or special schema is needed to appear in AI Overviews. Google states that no new machine-readable files, AI text files or special schema.org structured data are required for these features.

AI ModeQuery fan-outnosnippet and max-snippetZero-click search

Google Search Central

Citable content

Citable content consists of passages that stay clear when lifted out of context, answer a question directly and contain verifiable information. Because AI answers may draw on relevant sections rather than whole pages, GEO works at the passage level. The GEO paper reports that adding sources, quotations and statistics were among the most effective methods.

Instead of “this feature makes life easier”, a software company writes “Example CRM automatically links incoming emails to the matching customer record”, naming the product so the sentence stands on its own.

Repeating keywords is often thought to boost visibility in AI answers. In the GEO paper, keyword stuffing gave little to no improvement in generative engines, while adding sources, quotations and statistics proved more effective.

GEO (generative engine optimization)Citation (AI source attribution)Keyword stuffingFeatured snippet

arXiv (Aggarwal vd., 2023)

Semantic search

Semantic search is a retrieval method that finds results by closeness in meaning rather than exact word matches. A common approach converts the query and documents into embeddings, measures their similarity and returns the closest documents; OpenAI recommends cosine similarity for this comparison.

On a help center, someone searching “why was my card declined?” is shown the article “what to do when a payment isn’t approved” first, even though its title shares none of those words.

EmbeddingRAG (retrieval-augmented generation)Search intent

OpenAI

Citation (AI source attribution)

A citation, in AI answers, is the link, source card or inline marker showing which web page an AI-generated answer relies on. Google’s AI Overviews and AI Mode show links to supporting sites, and the Gemini API returns citations that map spans of the answer to source URLs. Citations are a core GEO metric.

Someone asks an AI assistant how often a home water filter cartridge should be replaced; a numbered source next to the relevant sentence points to a filter brand’s maintenance page.

AI visibilityGroundingBrand mentionCitable content

Google Search Central

Knowledge Graph

The Knowledge Graph is Google’s knowledge base that models real-world entities and the relationships between them. When Google introduced it in 2012, it said the graph was built on public sources such as Freebase, Wikipedia and the CIA World Factbook. Knowledge panels are shown for its entities, and developers can look them up through the Knowledge Graph Search API.

A search for “Taj Mahal” could mean the monument or the musician of the same name; in Google’s launch example, the Knowledge Graph helps tell which entity the searcher has in mind.

EntityKnowledge panelStructured data

Google Blog

Knowledge cutoff

A knowledge cutoff is the latest date covered by a language model’s training data; the model cannot know about later events from its own parameters. The way past this limit is to ground answers in current sources. Google’s documentation says Grounding with Google Search helps Gemini give more accurate answers and cite verifiable sources beyond its knowledge cutoff.

A model won’t know the address of a shop that opened after its cutoff, but an assistant with web search can find the shop’s current location page and cite it in the answer.

Training dataGroundingRAG (retrieval-augmented generation)Large language model (LLM)

Google AI for Developers

Knowledge panel

A knowledge panel is the information box that appears in Google when someone searches for a person, place, organization or thing in the Knowledge Graph, giving a quick overview. Panels are generated automatically from various sources across the web, and the subject or an official representative can get verified and suggest changes.

Searching a publisher’s name shows a knowledge panel with its founding year, logo and official site; as a verified representative, the publisher suggests a correction to a wrong founding year.

Many think brands write their own knowledge panel. Google says panels are generated automatically from various sources across the web, while verified representatives can suggest changes.

Knowledge GraphEntityGoogle Business ProfileSERP (search engine results page)

Knowledge Panel Yardım

Large language model (LLM)

A large language model (LLM) is an AI model trained on very large amounts of text that generates language by estimating the probability of the next token, or sequence of tokens, in a text. Assistants such as ChatGPT, Gemini and Claude run on such models. How well a model predicts depends on the size and diversity of its training data.

When someone asks an assistant what to check before commissioning a company website, the language model builds the reply token by token; if web search is enabled, it can also draw on the pages it finds.

People often think a language model looks facts up like a database. As Google notes, generative models don’t retrieve facts; they predict a likely sequence of words based on training data, so unverified answers can be wrong.

TokenTraining dataGenerative AIHallucination (AI)

Google for Developers

ClaudeBot

ClaudeBot is Anthropic’s crawler that gathers web content for its generative AI models. Anthropic says blocking ClaudeBot in robots.txt signals that the site’s future materials should be excluded from its AI model training datasets. Anthropic also runs Claude-SearchBot for search indexing and Claude-User for user questions, each managed separately.

An e-learning platform blocks ClaudeBot in robots.txt but allows Claude-SearchBot and Claude-User, so its content stays out of training while Claude can still read its pages when users ask questions.

Blocking ClaudeBot is often thought to cut off all Claude access. Anthropic says ClaudeBot relates to training, while Claude-SearchBot handles search indexing and Claude-User handles user requests, each controlled by its own robots.txt rule.

AI crawlerGPTBotTraining datarobots.txt

Anthropic

Training data

Training data is the body of text and other content an AI model processes during training to learn patterns in language. Large language models are trained on very large text collections, and their predictions depend on the size and diversity of that data. Web content can end up in training data; controls such as GPTBot and Google-Extended manage this use.

A research firm doesn’t want summaries of its paid reports used for model training, so it disallows GPTBot and ClaudeBot in robots.txt and adds a Disallow rule for the Google-Extended token.

Large language model (LLM)GPTBotGoogle-ExtendedKnowledge cutoff

Google for Developers

GEO (generative engine optimization)

GEO (generative engine optimization) is the work of getting a brand and its content cited as a source in answers written by generative AI systems such as ChatGPT, Gemini, Perplexity or Google AI Overviews. The term was introduced in a 2023 paper by Aggarwal et al. It builds on classic SEO but measures visibility inside the answer, not rank position.

A bookkeeping software site restructures its e-invoicing guide so each section opens with a direct answer, lists the steps and cites official sources, aiming to be named when AI assistants answer “how does a small business issue an e-invoice?”

A common belief is that GEO replaces SEO. Google says SEO best practices remain relevant for AI Overviews and AI Mode, and a page must be indexed and eligible to show with a snippet to appear there.

AEO (answer engine optimization)AI visibilityCitable contentSEO (search engine optimization)

arXiv (Aggarwal vd., 2023)

Google-Extended

Google-Extended is a standalone product token used in robots.txt, not a separate crawler. It lets publishers manage whether content Google crawls with its existing user agents may be used to train future Gemini models and for grounding in Gemini Apps and Vertex AI. Google says it does not affect inclusion in Google Search and is not a ranking signal.

A magazine site adds “User-agent: Google-Extended” and “Disallow: /” to robots.txt; with Googlebot’s rules unchanged, its pages continue to be crawled and shown in Google Search.

Blocking Google-Extended is often thought to remove pages from AI Overviews and AI Mode. Google says the token does not affect inclusion in Search; to limit what appears in Search’s AI features, publishers use controls such as nosnippet, max-snippet or noindex.

robots.txtGooglebotnosnippet and max-snippetTraining data

Google Crawling Infrastructure

Embedding

An embedding is a way of representing the meaning of text as a vector, a list of floating-point numbers. The distance between two vectors measures how related the texts are, so similar meanings sit close together. Embeddings power tasks such as search, clustering, recommendations and classification.

“Terminating a lease” and “can my landlord end the contract?” share few words, yet their embeddings land close together, so a search system can match both to the same topic.

Semantic searchRAG (retrieval-augmented generation)Token

OpenAI

GPTBot

GPTBot is OpenAI’s web crawler that collects content which may be used to train its generative AI foundation models. According to OpenAI, disallowing GPTBot in robots.txt indicates that a site’s content should not be used for that training. OpenAI says its bot settings are independent and publishes GPTBot’s IP addresses in a separate list.

A publisher adds “User-agent: GPTBot” followed by “Disallow: /” to robots.txt; since no rule blocks OAI-SearchBot, access for ChatGPT search stays open.

Blocking GPTBot is often thought to remove a site from ChatGPT search. OpenAI says the settings are independent: a site can disallow GPTBot and still allow OAI-SearchBot in order to appear in search results.

OAI-SearchBotAI crawlerTraining datarobots.txt

OpenAI

Hallucination (AI)

A hallucination is when a language model confidently produces a statement that sounds plausible but is false or unsupported. OpenAI researchers argue that one cause is training and evaluation procedures that reward guessing over acknowledging uncertainty. Google advises fact-checking AI-generated content before publishing it.

Asked when a café closes, an assistant invents a time the café never published. If opening hours appear as plain text and consistently across the site, a search-connected engine may be more likely to find the right answer.

Answers grounded in web search are often assumed to be error-free. Google says grounding reduces hallucinations; reducing is not eliminating, so claims in an answer should still be checked against their sources.

GroundingRAG (retrieval-augmented generation)Large language model (LLM)Knowledge cutoff

OpenAI

Prompt

A prompt is the input a user or application gives a language model, describing the task, the context and the expected format of the output. The clarity of a prompt shapes the scope and form of the answer. Anthropic recommends defining success criteria, and ways to test against them, before refining a prompt.

A team tracking AI visibility writes prompts from real customer questions, such as “which firms offer corporate accounting services?”, and runs the same prompts across several assistants at regular intervals.

Large language model (LLM)AI visibilityToken

Anthropic

llms.txt

llms.txt is a proposed file that summarizes a site’s key pages in Markdown for language models and AI agents. Jeremy Howard published it on llmstxt.org in September 2024. Its only required section is an H1 with the site’s name; it sets no access rules like robots.txt and does not replace a sitemap.

A software company adds a /docs/llms.txt file for its developer documentation, with a short summary followed by links to Markdown versions of its setup, API reference and release notes pages.

Some believe llms.txt is required to appear in Google’s AI summaries. Google says no AI text files are needed for AI Overviews or AI Mode; llms.txt is a proposal published on llmstxt.org.

robots.txtAI crawlerXML sitemap

llmstxt.org

nosnippet and max-snippet

nosnippet and max-snippet are robots meta rules that limit the text preview Google Search shows from a page. nosnippet removes the text snippet and stops the content being used as direct input for AI Overviews and AI Mode; max-snippet caps how many characters can be used. The data-nosnippet attribute excludes only the marked section.

A membership site wraps the summary box of a paid report in a div with the data-nosnippet attribute, while the rest of the page remains available for search snippets and AI summaries.

nosnippet is sometimes thought to affect only AI summaries. Google says it applies to all forms of search results, including web search, Images and Discover, so the classic text snippet disappears too.

SnippetRobots meta tagAI OverviewsFeatured snippet

Google Search Central

OAI-SearchBot

OAI-SearchBot is the crawler OpenAI uses to surface websites in ChatGPT’s search features. OpenAI says sites that opt out of it will not be shown in ChatGPT search answers, though they can still appear as navigational links, and that robots.txt changes can take about 24 hours to be reflected in search results.

A hotel group investigating why ChatGPT never names it for “family hotels in Bodrum with a kids’ club” finds in its server logs that the firewall returns 403 to OAI-SearchBot requests.

GPTBotAI crawlerrobots.txtServer log analysis

OpenAI

PerplexityBot

PerplexityBot is the crawler Perplexity uses to surface and link websites in its search results. Perplexity says the bot is not used to crawl content for AI foundation models and recommends allowing it in robots.txt. Its separate fetcher, Perplexity-User, retrieves pages for user requests and generally ignores robots.txt rules.

A software review site allows PerplexityBot in robots.txt so it can be cited in Perplexity, and checks that its CDN isn’t blocking requests from Perplexity’s published IP ranges.

Blocking PerplexityBot is often seen as protecting content from model training. Perplexity says the bot does not collect content for foundation models; blocking it mainly risks limiting how the site appears in Perplexity’s search results.

AI crawlerrobots.txtBot traffic

Perplexity

Query fan-out

Query fan-out is a technique in which an AI search system breaks a question into subtopics, runs multiple related searches across them and builds its answer from the results. Google says AI Overviews and AI Mode may use it, which lets them show a wider and more diverse set of links than a classic web search.

A question about a three-day winter trip to Cappadocia can fan out into searches on transport, hotels, balloon flight conditions and winter clothing; a page that only answers the balloon question clearly may still be cited.

It is often assumed that only the top result for the main query can be cited in an AI summary. Because Google may split a question and run several searches, a page that answers one subtopic well can also appear as a link.

AI ModeAI OverviewsLong-tail keywordTopic cluster

Google Search Central

RAG (retrieval-augmented generation)

RAG (retrieval-augmented generation) is a method in which a language model first retrieves relevant passages from an external knowledge source and then bases its answer on them. It was proposed in a 2020 paper by Lewis et al. RAG reduces reliance on what the model learned in training and can make it easier to show where an answer comes from.

An insurer’s support assistant, asked whether a policy covers earthquake damage, first retrieves the relevant clauses from the current policy terms, then writes its answer based on them and links to the clause.

GroundingLarge language model (LLM)EmbeddingHallucination (AI)

arXiv (Lewis vd., 2020)

Conversational search

Conversational search is a way of searching in which people ask natural-language questions instead of single keyword queries and go deeper with follow-up questions that build on the previous answer. Google’s AI Mode is one example; Google says it is especially helpful for questions that need further exploration, reasoning or complex comparisons.

A user asks for quiet printers suited to a small office, then continues in the same conversation with “which of these print double-sided?” and “which has the lowest toner cost?”

AI ModeQuery fan-outSearch intentLong-tail keyword

Google Search Central

Grounding

Grounding is the practice of basing a language model’s answer on current, verifiable sources such as web search results. In Google’s Gemini documentation, the model decides whether a search would help, generates and runs queries, then synthesizes the results, letting it answer beyond its knowledge cutoff and cite sources.

A developer enables Grounding with Google Search in a Gemini-based app; asked about a regulation that took effect this week, the model runs a search and ties sentences in its answer to source pages.

RAG (retrieval-augmented generation)Citation (AI source attribution)Hallucination (AI)Knowledge cutoff

Google AI for Developers

Token

A token is the smallest unit a language model works with when processing text; it can be a word, part of a word or a single character. Models read text as sequences of tokens and generate answers by predicting one token after another. How a text splits into tokens varies by language and tokenizer.

A long, heavily suffixed Turkish word such as “optimizasyonlarımızdan” may be split into several subword tokens, while a short English phrase with a similar meaning can take fewer.

A token is often taken to mean a word. By Google’s definition, a token can be a word, a subword or even a single character, and most modern models tokenize text into subwords.

Large language model (LLM)PromptEmbedding

Google for Developers

Generative AI

Generative AI is the umbrella term for AI systems that create new content, such as text, images, audio or code, based on patterns in their training data. Text generation runs on large language models. Google notes that generative models predict a likely sequence of words rather than retrieving facts, so their output should be fact-checked before publishing.

An online store drafts product descriptions with generative AI, then checks every dimension, material detail and return condition against the product data before the pages go live.

Many assume Google penalizes any AI-written content by default. Google says the tool itself is not the issue; generating many pages without adding value for users may violate its scaled content abuse policy.

Large language model (LLM)AI-generated contentScaled content abuseHallucination (AI)

Google for Developers

Entity

An entity is a person, place, organization, product or concept that search engines and AI systems recognize as a distinct, identifiable thing. Google summed up the idea as “things, not strings” when it introduced the Knowledge Graph in 2012. Being recognized as an entity can help a brand’s name be matched with the right facts rather than other meanings.

A producer called “North Olive” uses the same legal name, logo and address on its website, in its Organization markup and on its social profiles, so its name isn’t confused with a direction or a product.

Knowledge GraphKnowledge panelStructured dataSchema.org (schema markup)

Google Blog

AI visibility

AI visibility describes how often, in what context and how accurately a brand or website is mentioned in answers from systems such as ChatGPT, Gemini, Claude, Perplexity and Google’s AI features. The GEO paper measures it with metrics such as how much of the answer the cited sentences take up and where they appear.

A dental clinic asks three assistants the same twenty patient questions every month and records how often it is named, which competitors are recommended and whether its address is given correctly.

GEO (generative engine optimization)Citation (AI source attribution)Share of voicePrompt

arXiv (Aggarwal vd., 2023)

AI crawler

An AI crawler is a bot that fetches web content for model training, for an AI-powered search index, or for a user’s request during a chat. OpenAI and Anthropic run separate bots for these purposes, and each can be addressed by name in robots.txt; GPTBot, for example, is for training and OAI-SearchBot for search.

A news site disallows GPTBot and ClaudeBot in robots.txt but allows OAI-SearchBot and Claude-SearchBot, opting out of training while keeping a path to being cited in AI search.

People often assume user-triggered bots follow robots.txt too. OpenAI says robots.txt rules may not apply to ChatGPT-User, because those actions are initiated by a user.

robots.txtGPTBotClaudeBotBot traffic

OpenAI