The Tasadr glossary · GEO terms in plain words

Also known as generative engine optimization · optimising for AI answers · AI SEO · generative engine optimisation

GEO is the work of getting your site cited inside the answer an AI engine writes, rather than ranked in a list of links.

It stands for Generative Engine Optimization, and we call it optimising for AI answers. Some call it "AI SEO". Its measurable result is a citation inside the answer, not a position on a results page, so it is measured by question rather than by keyword. The full definition, and why it matters now, is in what GEO is.

Also known as answer engine optimization · answer engine optimisation

AEO stands for Answer Engine Optimization: preparing your page to be the direct answer an answer engine lifts or builds its reply on.

The goal is the same as GEO's: your page as the source the answer is built on. Its definition, other names and an Arabic example are in what is AEO, or answer engine optimisation?, and the three names are compared in SEO vs GEO vs AEO.

Also known as Q&A engine · question and answer engine · answer engine

An answer engine is any engine that replies to a question with a ready answer rather than a list of links, such as ChatGPT, Gemini, Perplexity and Google AI Overviews.

It is sometimes searched for as a "Q&A engine"; in Arabic it is «محرك الإجابة» or «محرك الأسئلة والأجوبة». The work of preparing your page for one is called answer engine optimisation (AEO).

Also known as search engine optimisation · search engine optimization · SEO

SEO (search engine optimisation) is the work of raising your page's rank among the links on a results page; its result is a position in a list, not a citation inside an answer.

Many Arabic speakers write it «السيو», and someone asking «وش يعني سيو؟» ("what does SEO mean?") means this same work. It is the foundation of GEO, not its replacement. An AI engine reads the same web that search engines index, so a slow page, or one blocked to crawlers, is cited by nobody.

The difference between the two results, a rank and a citation, is set out in how GEO differs from SEO.

Also known as keywords · search term

Keywords are the words a searcher types into a search engine, the unit classic SEO is built on, such as "dental clinic Riyadh".

It still matters, because an engine searches with words before it writes its answer. But the unit of measurement in answers is the whole question in your customer's words, not the word alone. More in do keywords still matter in GEO?

Also known as LLM · large language models

A large language model is software trained on a very large body of text until it can write new text that answers a question; engines such as ChatGPT, Gemini and Claude are built on one.

LLM is its short form. How it writes its answer, and why engines connect it to search when they need a source, is in what is a large language model, and how does it answer?

Also known as RAG · retrieval augmented generation

Retrieval-augmented generation is when a system first retrieves pages or documents related to the question, then the language model writes its answer from them, rather than relying on what it learnt alone.

RAG is its short form, and it is the principle that has an engine read your page before it answers. How it chooses what to retrieve is in what is retrieval-augmented generation (RAG)?

Also known as grounding with search · grounded answers

Grounding is an engine tying its answer to content it finds on the web at the time of the question, building the answer on it and linking it, rather than writing from what the model learnt alone.

Google's Arabic documentation for Gemini developers calls it «الاستناد إلى مصادر», and says it improves factual accuracy, gives access to current information and adds citations to the sources. A page search cannot find is a page an engine cannot ground on, as in how AI engines choose their sources.

Also known as query fan-out · sub-queries · fan-out queries

Query fan-out is when an engine breaks your customer's question into subtopics, runs several searches at once, and writes its answer from all their results.

It is Google's name for it; Google's Arabic pages call it «توسيع طلبات البحث». Google's definition, an AlUla example and what it changes on your page are in what is query fan-out, and what does it change on your page?

Also known as hallucination · AI errors

An AI hallucination is an engine stating something false with confidence, as if it were true, such as an old price, a closed branch or a service your business does not offer.

So it matters that an engine finds a page on your site that states the facts about your business clearly. Why answers get things wrong, and how to report an error, is in what is an AI hallucination, and why do answers err?

Also known as zero-click search · no-click search

Zero-click search is a search that ends on the results page or the answer: the asker reads what they need without opening any site.

So visits alone are not enough of a measure. What it means for your site is in what is zero-click search, and what does it mean for you?

Also known as brand entity · entities

An entity is a thing with one identity a search engine knows, a person, place, company or product, however its name is spelt.

An engine understands that your business's Arabic name, its Latin name and its domain are one thing when your pages and profiles across the web say so with the same name. The details are in what is an entity in search, and why does it matter?

Also known as experience, expertise, authoritativeness, trust · EEAT

E-E-A-T is the four qualities Google uses to describe trustworthy content: experience, expertise, authoritativeness and trust, of which Google says trust matters most.

Google says E-E-A-T is not a specific ranking factor. How it relates to AI answers is in what is E-E-A-T, and how does it relate to AI answers?

Also known as GenAI · generative artificial intelligence

Generative AI is artificial intelligence that writes new text, draws an image or produces audio in response to a request a user types, rather than picking from results that already exist.

Google's guidance for site owners describes generative models as not retrieving facts but predicting a likely sequence of words from their training data. That is why engines such as AI Overviews and Perplexity lean on pages they find in search before they write, then link what they relied on. A customer in Riyadh who asks "what's the best oud shop in Riyadh?" reads one paragraph and its sources, not ten links.

Related terms in this glossary: the large language model that writes the answer, grounding, which ties it to real pages, and AI hallucination, when the model gets things wrong. Where your page fits into all this is in how does an engine build an answer, and where do its sources come from?

Also known as chat bot · AI chatbot

A chatbot is a program you type or speak a question to, which replies in sentences as if in conversation and builds on what was said earlier in the same chat.

ChatGPT, Gemini and Claude are chatbots your customer asks directly. We call them chat engines, because they search the web and link their sources. OpenAI says ChatGPT may search the web automatically when a question would benefit from current information, and only then can it cite your site with a link; an answer it writes from memory alone cites no source.

Take a customer in Kuwait who asks for the best family seafood restaurant in Salmiya. If ChatGPT searches before it answers, it links the restaurants whose pages it read, and yours may be among them; if it answers from memory, it may write names without a single link.

The chatbot you put on your own site to serve customers is a different thing: it answers people who have already reached you, while ChatGPT and its peers answer your customer before they know you, as in what we mean by the questions AI answers about you. Close to it in this glossary: AI engine, large language model and AI agent.

Also known as AI prompt · prompts

A prompt is what a user types to an AI model to ask it for something, whether a question, an instruction or a text to summarise.

On AI engines the prompt is usually your customer's own question, in their words and with their conditions. A family in Jeddah may write "I want a mandi restaurant that suits a big family and has private seating", not "mandi restaurant Jeddah". That is why we measure the buying question as the customer writes it, not a clipped keyword.

Improving the prompts you write for an AI assistant is another skill altogether: optimising for AI answers works on your page, not on how the question is phrased, as the steps to lead AI answers in Arabic explain.

Also known as model training data · AI training data

Training data is the text, images and other material a language model learnt from before its release; it is where the model's knowledge comes from when it answers without searching.

OpenAI says its models learn from three sources: information publicly available on the internet, such as public pages, forums and blogs; information it accesses through partners; and information its users, human trainers and researchers provide. It says it does not intentionally gather data from sources known to be behind paywalls.

This matters because what a model learnt about your business stays with it until it is trained again. If a beauty centre in Jeddah changed its name this year, a model answering from memory may still write the old one, and only an engine that searches and finds the centre's new page corrects it. So write your business's facts the same way everywhere on the web.

You can ask that your pages not be used for training by disallowing a training crawler in robots.txt, such as OpenAI's GPTBot. That is independent of the search crawler that puts you in its answers, and the difference is set out in which crawlers should you allow, and which should you block? Read it with knowledge cutoff and grounding.

Also known as knowledge cut-off date · cutoff date

A knowledge cutoff is the last date covered by the data a language model was trained on, so the model does not know from memory anything that happened after it.

OpenAI and Anthropic publish this date for each model on their model pages. If you moved your clinic from Mohandessin to Sheikh Zayed after that date, or opened a new branch in Khobar, the model will only know if it searches the web at the time of the question and finds a page that says so. That is grounding.

So write the new fact on a page of your own site that a crawler can read, with its date, for the engine to find when it searches. Two related terms in this glossary: large language model and training data.

Also known as user intent · searcher intent · query intent

Search intent is what a searcher wants from their question: to understand something, to compare options, to buy, or to reach a site they already know by name.

For example, "How much is a dental implant in Maadi?" is about buying, and "Does a dental implant hurt?" is about understanding. The page that answers the first needs the price and the district in its first sentence; the page that answers the second needs an explanation that reassures the patient before any price.

Microsoft's guide to writing content for AI search answers tells you to write for intent, not just keywords, and to use phrasing that directly answers the questions people ask. At Tasadr we start measuring with the questions whose intent is to buy, because they are the closest to a paying customer.

Both kinds, with examples, are in what is the difference between a buying question and an informational one? Next to it in this glossary: buying question, informational question and long-tail queries.

Also known as long-tail keywords · long tail

Long-tail queries are specific searches of many words that few people type each month, each saying exactly what its searcher wants.

Many people type "dental clinic"; few type "clear aligners for adults in Al Yasmin, payable in instalments", and those few know what they want. That is the shape of the question your customer types to an AI engine: a full sentence with its conditions. Google's guide to its generative AI features adds that its systems understand synonyms and the general meaning of what people seek, so you need not worry about covering every way someone might search for what you offer.

Write one page that answers the conditions your customers actually state, such as price, district and timing, and gather those conditions as how do you choose the questions you measure on? explains. Two related terms: query fan-out, when an engine splits a question into sub-questions, and search intent.

Also known as AI agents · agentic AI

An AI agent is a system that completes a task for its user step by step: it searches, opens pages and compares, and may fill in a form or make a booking, rather than only answering.

Google's guide to its generative AI features describes agents as autonomous systems that perform tasks on behalf of people, such as booking a reservation or comparing product specifications, and says browser agents may access your website to gather the data they need to complete those tasks.

So an agent reads your page like a visitor in a hurry. If a family in Riyadh asks an agent to book a chalet in Khobar for the weekend, it needs to read the price and the availability as plain text, and to find the booking button. Google's web.dev recommends real buttons, and form fields linked to their labels, so the agent knows what to press and what to fill in.

Some content delivery networks tell agents apart: since 1 July 2026 Cloudflare sorts AI traffic into search, agent and training, and a site owner can block each kind on its own. The crawlers that visit a page when a user asks for it are listed in what are AI crawlers, and who sends them?

Related terms: user-triggered fetcher, chatbot and CDN.

Also known as AI share of voice · SOV

Share of voice is the proportion of the answers to a fixed set of questions that your brand gets, compared with the other sites those answers cite or mention.

The number follows the questions you choose: your share in "best accounting firm in Dammam" questions is not your share in price questions, and it is a different measure from the Tasadr score. How to count it yourself and how to read it are in what is share of voice in AI answers?

Also known as Google Knowledge Graph

The Knowledge Graph is Google's database of billions of facts about people, places and things, which its search draws on when the answer to a question is a fact.

Google's Arabic pages call it «الرسم البياني المعرفي». Google says it takes facts from a variety of sources that compile factual information, and that its search creates knowledge panels from it automatically when there is enough information about something on the open web. Businesses have their Google Business Profile instead.

Google's own Arabic page gives an example: how tall the Burj Khalifa is. What matters for you is that Google gathers facts from many sources across the web, not from your site alone, so write your business's name and address the same way on your site, your profile and the directories.

The full explanation is in what is the Knowledge Graph, and how does Google know you? Next to it in this glossary: entity, Wikidata and Google Business Profile.

Also known as content chunking · text chunking

Chunking is cutting a long text into small passages that can each be retrieved on their own, a step retrieval systems take before a model writes its answer.

Here Google's own guidance contradicts a common claim. Its guide to generative AI features says there is no requirement to break your content into tiny pieces for AI to understand it, that its systems can understand several topics on one page and show the relevant piece, and that there is no ideal page length.

What helps is what helps a reader anyway: a heading that asks the question and a paragraph that answers it in its first sentence, as when a dental clinic in Jeddah titles a section "How much does a cleaning cost?" and opens it with the price. That is an answer paragraph; the method is in how do you write a paragraph AI engines will quote?, and chunking itself comes from retrieval-augmented generation.

Also known as machine translation · machine-translated content · auto-translated pages

Machine-translated pages are pages whose text was moved from one language to another by translation software, then published without review by someone who knows both languages and the market.

Under Google's spam policies, generating many pages through automated transformations, translation among them, is scaled content abuse when little value is provided to readers. Google's guidance for multilingual sites explains that it determines a page's language from the visible content, and that translating only the boilerplate while keeping the bulk of the content in one language makes for a poor experience.

An illustrative example: a Dubai clinic's page was machine-translated, and its "Book now" button became «كتاب الآن», which reads as "a book, now". Then a patient asks an engine in Arabic what a dental cleaning costs in Al Barsha. The engine finds no Arabic sentence on the clinic's page that answers, and may cite a smaller clinic that wrote its page in Arabic from the start, as in what if your site is English first?

Read it with hreflang and helpful content.

Also known as people-first content · helpful, reliable, people-first content

Helpful content is Google's name for content written to benefit its reader first rather than to manipulate rankings, the subject of its guidance page "Creating helpful, reliable, people-first content".

Google says its ranking systems are designed to prioritise such content, and its guide to generative AI features says those features are rooted in the same ranking and quality systems. What makes a page helpful in search is also what puts it forward as a source in an answer: a Riyadh developer's page that states unit sizes, the payment schedule and the handover date helps more than one repeating "the best investment in the capital".

One of the self-assessment questions Google suggests: after reading your content, will someone feel they have learned enough about the topic to help them reach their goal? Read it with E-E-A-T, YMYL and scaled content abuse.

Also known as Your Money or Your Life · YMYL topics

YMYL topics are subjects that could significantly affect people's health, financial stability or safety, or the welfare of society, such as medical treatment, investment and the law.

Google calls them "Your Money or Your Life" topics, YMYL for short, in its page on helpful content, whose Arabic edition calls them «مالُك أو حياتك». It says its systems give even more weight to content with strong E-E-A-T on these topics, and that trust is the most important part of E-E-A-T. Its guide to its generative AI features says those features are rooted in its core Search ranking and quality systems.

A Riyadh dermatology clinic's page on hair-loss treatment and a Cairo accounting firm's page on the tax return are both YMYL. So say on them who wrote the page and what qualifies them, when it was last reviewed and where its figures come from, and promise no outcome, as in what should a clinic's page say for the engine to cite it?

Near it in this glossary: helpful content and AI hallucination.

Also known as scaled content

Scaled content abuse is a site generating many pages primarily to manipulate search rankings rather than to help readers, which Google's spam policies forbid however the pages are made.

Google's Arabic page calls it «إساءة استخدام المحتوى الواسع النطاق». Its examples include using generative AI tools to create many pages that add no value for readers, and building pages from scraped content through automated transformations such as synonymising or translating. Its guide to generative AI features adds that a page for every variant of a question, made primarily to manipulate rankings or AI responses in its Search, breaks the same policy.

An illustrative example: a cleaning company in Dubai publishes two hundred pages titled "apartment cleaning in Al Barsha", "apartment cleaning in Al Quoz" and the like, with only the district changed. One page per real service, with its prices and the areas it truly covers, serves better, and one strong page can answer a question and its sub-questions, as in should you write a page for every sub-question? Its opposite is helpful content; close to it are machine-translated pages.

Also known as original research · proprietary data

Original data is figures and facts you gathered yourself from your own work, from a measurement, a survey, a sales record or a price list, which a reader cannot find on any other page.

Among the questions Google asks you to judge your content by is "Does the content provide original information, reporting, research, or analysis?", and its guide to its generative AI features says a first-hand review provides a unique perspective based on personal experience.

The research paper the name GEO comes from found in its experiments that adding statistics, quotations and cited sources to a page raised its presence in engines' answers by roughly 30 to 40%. That is a research result, not a promise for your site.

Picture an estate agent in Riyadh that publishes the average days to sell in the districts it covers, from its own deals, or a clothing shop in Cairo that publishes the sizes it sells most. Those are numbers an engine can only find on your page. We do the same at Tasadr and publish what we measured, with its date, as in a real report for an electronics store in Egypt.

Read it with content freshness and answer paragraph.

Also known as AI engines · AI search engine

An AI engine is software that reads pages from the web, then writes an answer in its own words and credits the sources it relied on, such as ChatGPT, Gemini and Perplexity.

We sort the eight engines we measure into two kinds, and the difference is how your customer reaches them. What sets each one apart is in the AI engines we measure.

• Chat engines, which are ChatGPT, Gemini, Perplexity, Claude and Grok: your customer opens one and asks it directly, and its answer changes from one measurement to the next.
• Search surfaces, which are Google AI Overviews, Google AI Mode and Copilot: they reach your customer while they search, inside the results page itself.

Also known as AI Overviews · Google AI overviews · AI summaries in Google

Google AI Overviews is a summary Google writes above the search results, with links to the sources it was built from.

Your customer does not choose this surface: they type a question into Google and find the summary waiting above the links. It is not the old featured snippet either. A snippet quoted one page word for word; an overview is new text Google writes after reading several pages.

How it picks its references in the Saudi market is in the guide to Google AI Overviews in Saudi Arabia.

Also known as AI Mode · Google AI mode

Google AI Mode is a conversation inside Google Search, behind a tab of its own, that writes a longer answer with its own references.

It arrived in Arabic on 8 October 2025 as «وضع AI», and we measure it separately from AI Overviews because one question can draw different sources on the two. The detail is in what is Google AI Mode, and how does it choose its sources?

Also known as SERP · search results page · results page

A search engine results page is the page a search engine shows after you type your query, with the usual links and other elements around them, such as an AI-written summary, a map, images and ads.

Google's Visual Elements gallery says the page can look different depending on the device, the country and the language of the query, so what your customer in Jeddah sees on a phone may not be what you see at your desk. And when Google writes a summary at the top, your customer may read the answer and stop: you can be the first link and still be absent from the summary above it, as in a rank or a citation: what changes in the result?

Its elements include Google AI Overviews, the local pack and the featured snippet, while Google AI Mode puts in its place an answer you can talk to, with its links.

Also known as featured snippets · position zero · answer box

A featured snippet is a box in Google's results that puts text quoted from a single page above that page's title and link, the reverse of a regular result.

Google's Arabic pages call featured snippets «المقتطفات المميَّزة». Google states that its systems decide whether a page would make a good featured snippet for a search, and that a site owner cannot mark a page as one. It differs from Google AI Overviews: the snippet quotes one page, while the overview is text Google writes from several.

Someone may ask Google how many days furniture delivery takes inside Riyadh and find, above the results, a sentence from one shop's page, with the page's name and link under it. The name AEO grew out of direct answers like these, as in where does the name AEO come from?, and you can keep a page out of featured snippets with nosnippet, one of the snippet controls.

Also known as local results · map pack

The local pack is a box on Google's results page showing a map and a few nearby businesses, with their names, ratings and opening hours, when someone searches for a service in their area, such as "car wash near me".

Google's help pages call these local results and say they rest mainly on relevance, distance and prominence, that prominence also draws on how many websites link to your business and how many reviews it has, and that there is no way to request or pay for a better local ranking. The profile they are built from is your Google Business Profile.

Google's guide to generative AI features says a Business Profile can help your products and services be shown in AI responses as well as in other search results, as in what part does your Google Business Profile play in AI answers? Related terms in this glossary: NAP, online reviews and local SEO.

Also known as ChatGPT web search · SearchGPT

ChatGPT search is ChatGPT's ability to search the web so it can answer with current information and links to the sources it relied on.

OpenAI says ChatGPT may search automatically when a question would benefit from current information; that when it works with search providers it typically rewrites the question into one or more targeted search queries; and that it may use the asker's approximate location from their IP address to give local results.

For your site to be eligible, OpenAI asks for two things: let its OAI-SearchBot crawl, and make sure your host or content delivery network allows traffic from the IP addresses it publishes for that crawler. It says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, and that no site's placement is guaranteed.

For example, a customer in Dubai asks for the best car service centre in Al Quoz. ChatGPT may search with phrases that look nothing like their words, then link the centres whose pages it read. More in how does ChatGPT pick which businesses to recommend? Close to it: AI search crawler and query fan-out.

Also known as Google Extended

Google-Extended is a name a site owner writes in robots.txt to decide whether Google may use what it gathers from the site to train Gemini models and to ground answers in the Gemini apps; it is not a crawler of its own.

Google says it does not affect a site's inclusion in Google Search and is not used as a ranking signal there, so blocking it does not take you out of AI Overviews or AI Mode, as in does blocking Google-Extended take me out of AI Overviews? Its nearest neighbour is the training crawler.

Also known as AI search bot · search crawler

An AI search crawler is a crawler an AI engine sends to read web pages and put them in its search index, so the engine can then cite them in its answers.

OpenAI and Anthropic each run one crawler for search and another for training, independent of each other: OAI-SearchBot reads for ChatGPT search, and Claude-SearchBot to improve search results in Claude. PerplexityBot reads for Perplexity's results, and Perplexity says it does not collect content to train models. OpenAI says a site that blocks its search crawler is not shown in ChatGPT search answers, and that blocking its training crawler does not take you out of them.

An illustrative example: a shipping company in Dammam launches its new site with a Disallow line its developer left in robots.txt from the test phase, blocking every crawler. No search crawler reads the site, and no training crawler either, while the pages work in the browser as if nothing were wrong.

All the names, and how to allow them, are in which crawlers should you allow, and which should you block? Next to it in this glossary: crawler, training crawler and user-triggered fetcher.

Also known as user-initiated fetcher

A user-triggered fetcher is a program that opens a specific page because a user asked for it at that moment, by pasting a link into a chat or asking an assistant to read a site, rather than crawling the web of its own accord.

Many engines have one, such as ChatGPT-User, Perplexity-User and Claude-User, and Google and Perplexity say theirs generally ignore robots.txt rules because a user asked for the page. Such a visit may come from a customer comparing your site with a competitor's site before deciding. How it differs from a crawler, and what each engine says about its own, is in what are AI crawlers, and who sends them?

Also known as direct answer · answer-first paragraph

An answer paragraph is the one or two sentences that answer the reader's question directly at the top of a section, before any detail, and make sense without what comes before them.

It is the part an engine can most easily carry into its answer: Microsoft says AI systems extract concise, clear pieces of content from pages and weave them into answers. How to write one is in how to write content AI quotes.

Also known as schema · schema markup · structured data markup

Structured data is a standard format for describing your page to search engines: it tells them, in a form machines read, what kind of page it is and whose, such as a local business, a product or a question and answer.

Google says it is not required for its generative AI features and must match what a visitor reads on the page. The type your business needs, and how to test it, is in which structured data does your site need, and what can it not do?

Also known as rich results · rich snippets

Rich results are Google search results that show more than a title and description, such as star ratings, a price or questions, and Google builds them from the structured data on the page.

Google's Arabic pages call them «النتائج الغنية بصرياً», and its search gallery lists the kinds its Search supports. Which one fits your business is in the structured data your site needs.

Also known as web crawler · bot · spider

A crawler is a program that reads web pages on behalf of an engine: it reaches your page and reads its text before the engine can cite it.

Each engine has its own, such as Googlebot for Google Search, OAI-SearchBot for ChatGPT search, PerplexityBot for Perplexity and Claude-SearchBot for Claude. Their names, how to allow them and how to check your site are in can AI engines read your site, and how do you check?

Also known as AI training crawler

A training crawler is a crawler that collects web pages to train AI models, not to put your site in search results or answers.

GPTBot is one: OpenAI says disallowing it means your content should not be used to train its models, and that this is independent of OAI-SearchBot, which puts sites in ChatGPT search. So you can block the first and allow the second, a choice set out in which crawler should you allow, and which block?

Also known as robots.txt file · robots file

robots.txt is a text file at the root of a site that tells crawlers what they may read on your site and what to leave.

Google says it is used mainly to manage crawler traffic to your site, and that it is not a way to keep a page out of Google: noindex does that. How to write it for AI crawlers is in how do you write a robots.txt that allows them?

Also known as llms.txt file

llms.txt is a proposed text file at the root of a site that sums it up and links its key pages for AI models.

Google says its Search ignores such files, so it is useful documentation, not a route to a citation. When your site is worth one is in what is llms.txt, and does your site need one?

Also known as indexation · Google indexing · getting indexed

Indexing is a search engine storing your page in its index after crawling it, so it becomes eligible to be shown in its results.

Many Arabic speakers call it «أرشفة الموقع». It is Google's condition for its generative features: a page must be indexed and eligible to be shown with a snippet to be a supporting link in AI Overviews or AI Mode. More in is indexing in Google a condition for AI answers?

Snippet controls are rules you put on your page to set how much of its text Google shows as a snippet, or to stop it, such as nosnippet, max-snippet and data-nosnippet.

Google says nosnippet stops the page being used as direct input for AI Overviews and AI Mode, and that max-snippet limits how much of it is used there, so you can take a page out of answers without meaning to, as in what do nosnippet and max-snippet do to your page?

Also known as Google Business Profile · Google My Business · GMB · Google Maps business listing

A Google Business Profile is your business's listing on Google Search and Maps, with its name, address, opening hours and reviews.

Google's Arabic name for it is «الملف التجاري على Google», and some Gulf searchers write «قوقل ماب بزنس». It is not the same as Tasadr's own business profile feature. How AI answers read it is in how do AI answers read your Google Business Profile and reviews?

Also known as name, address, phone · NAP consistency

NAP stands for name, address and phone: your business's name, address and phone number, written the same way everywhere on the web.

Google's guidelines for representing a business ask you to present it consistently, under the name it uses in the real world. So write the name, address and phone on your site exactly as they are on your Google Business Profile, as in what to write in your profile so it matches your site.

Also known as sitemap · sitemap.xml

An XML sitemap is a file that lists the addresses of your site's important pages, each one optionally with the date it last changed, so crawlers find them without following every link.

Google says a sitemap helps most on a large site, or on a new one with few external links to it; that it does not guarantee everything in it will be crawled and indexed; and that Google uses the last-modified date if it is consistently and verifiably accurate, while it ignores priority and change frequency.

Bing said on 31 July 2025 that sitemaps are a foundation for complete URL coverage in AI-powered search, that it typically revisits submitted sitemaps at least once a day, and recommended pairing them with IndexNow.

Tasadr's own sitemap, as we read it on 17 October 2026, lists every page twice, in Arabic and in English, each version dated by its own last real change rather than by when the file was generated, and each pointing to its twin with hreflang. The sitemap is one item in how do you prepare your site for AI questions?

Also known as canonical · canonical tag

A canonical URL is the address a search engine picks to represent a set of duplicate or near-duplicate pages, so it indexes that one and shows it in place of the others.

Google's Arabic documentation calls it «عنوان URL الأساسي». It says you can suggest it with rel=canonical, a redirect or your sitemap, but it may pick another page: your preference is a hint, not a rule. An Arabic page and its English version are not duplicates, while two Arabic pages with the same text, one for customers in Saudi Arabia and one for customers in the UAE, need a canonical and hreflang together.

This matters because the page Google picks is the one it indexes, and indexing is the condition for being a supporting link in AI Overviews and AI Mode: edit a version it did not pick, and its index may still hold the version you left unchanged. See what it chose in indexing statuses and the URL Inspection tool in Google Search Console, read more in does a page need to be indexed by Google for AI answers?, and pair it with hreflang.

Also known as hreflang tag · hreflang attribute · hreflang annotations

hreflang is an annotation in a page's head or in a sitemap that tells Google the page has a version in another language or for another country, and where to find it.

According to Google, it helps its search send each searcher to the version that fits their language or region. Each version must list itself and all the others, and if two pages do not point to each other the annotations are ignored. The value starts with the language, such as ar, which a country code may follow, as in ar-EG; a country code on its own is not allowed, and x-default names the page to open when no version fits.

It matters when your site has a version for each country with different prices: without it, a customer in Cairo may land on the UAE price page. On Tasadr's site every page has two versions that point to each other, and we write the language as plain ar rather than ar-SA, because the same Arabic page is written for every Arab country.

Which countries need a page of their own is in how do you lead AI answers in more than one Arab country? Read it with canonical URL and machine-translated pages.

Also known as meta robots tag · robots meta

The robots meta tag is a line in the head of an HTML page that tells search engines whether to index the page and whether to show a snippet of it, and it applies to that page alone, while robots.txt is a single file for the whole site that tells crawlers which pages they may read.

Google's Arabic documentation calls it «العلامة الوصفية لبرامج الروبوت». Its best-known rule is noindex, which keeps a page out of search results, and so out of AI Overviews and AI Mode, since Google requires a page to be indexed. Google also says these rules are read only if the crawler is allowed into the page: block a page in robots.txt and mark it noindex, and its address can still end up in Google's results through links from other sites, because Google never read the tag.

An illustrative example: a restaurant chain in Kuwait moves its new menu page from staging to its live site with a noindex its developer forgot, and the page stays out of Google and its AI summaries however good its text. So check the head of every important page after publishing. The other controls are in snippet controls and what do nosnippet and max-snippet do to your page?

Also known as SSR · server rendering

Server-side rendering is the server building a page's full text before sending it, so the text arrives in the HTML itself and whoever reads it does not have to run JavaScript.

In Google's words, server-side or pre-rendering is still a great idea, because it is faster for users and crawlers, and because not all bots can run JavaScript. Google recommends server-side rendering over dynamic rendering, which it calls a workaround, and its page on AI features asks that important content be available in textual form.

OpenAI's, Perplexity's and Anthropic's pages about their crawlers, as we read them on 17 October 2026, do not say whether those crawlers run JavaScript.

An illustrative example: a residential project's site in Jeddah is built as a single-page app, and its unit prices load with JavaScript after the page opens. View the page source and you will not find a single price, and a crawler that does not run JavaScript reads only the source. How to check in a minute is in why can an engine's crawler find your page empty?

Related terms: crawler, indexing and CDN.

Also known as CWV · web vitals

Core Web Vitals are three metrics by which Google judges a visitor's real experience of a page: how fast its largest element loads, how quickly it responds when the visitor interacts with it, and how steady its layout stays while it loads.

Google sets a line for a good experience on each: the largest contentful paint within 2.5 seconds of the page starting to load, an interaction to next paint under 200 milliseconds, and a cumulative layout shift score under 0.1. Its reports show them as LCP, INP and CLS, including a report of that name in Google Search Console.

Google does not say they are a condition for being cited in AI Overviews or AI Mode, but it lists a good page experience, reduced latency included, among the practices it recommends for its generative AI features, and a reader who taps your link in an answer on a phone in Riyadh or Cairo expects a page that opens quickly. The full checklist is in how do you prepare your site for AI questions? A content delivery network is one way to speed a page up.

Also known as CDN · content distribution network

A content delivery network is a set of servers spread across many countries that keeps copies of your site's pages and hands each visitor the copy from the point nearest them, so the site opens faster and holds up under load and attack.

It is abbreviated CDN. Because it stands between your site and everyone who requests it, it also decides which crawlers get in. That is why Google lists among the basics for its generative AI features that crawling be allowed in robots.txt and by any CDN or hosting infrastructure; OpenAI asks that your network allow the IP addresses it publishes for its search crawler; and Anthropic says its crawlers will not try to get past CAPTCHAs.

Some networks sort AI traffic. Since 1 July 2026 Cloudflare splits it into search, agent and training, and it says domains joining it from 15 September 2026 block training and agent bots by default on pages that display ads, while search crawlers stay allowed.

Picture a Kuwaiti shop that switches on bot protection in its network: the search crawler now meets a challenge page instead of the product page, while the shop sees nothing wrong in its own browser. So ask whoever runs your network, as in can a firewall or CDN block AI crawlers? Close to it: AI search crawler, AI agent and Core Web Vitals.

Also known as IndexNow protocol · Index Now

IndexNow is an open protocol through which a site tells participating search engines at once that a page was added, changed or deleted, instead of waiting for a crawler to visit.

The protocol's site names Microsoft Bing, Naver, Seznam.cz, Yandex and Yep among its supporters, and Google is not among them. It says IndexNow helps keep your content current in AI-powered search results, and that a submission does not guarantee immediate indexing. It concerns you because Copilot is Microsoft's answer inside Bing search: Bing learns of your edit without waiting for its crawler.

We send the addresses of the pages that changed to IndexNow with every release of our own site, since 11 October 2026. For Google, rely on your XML sitemap and Google Search Console: IndexNow does not replace them. Underneath it all is indexing, and what your site earns in Copilot you read in Bing's AI Performance report.

Also known as internal linking · internal link

Internal links are links from one page of your site to another page on the same site, such as a link from a service page to its prices.

Google uses links, it says, to find new pages to crawl and as a signal of a page's relevance, and it asks that every page you care about have a link from at least one other page on your site, with anchor text that says concisely what is behind it. It lists making your content easy to find through internal links among the basics for its generative AI features, and Microsoft says internal linking remains essential for your content to be discovered.

If a Jeddah clinic wrote a plain page on clear-aligner prices and linked to it from nowhere else, neither a crawler moving from link to link nor a patient starting on the orthodontics page would reach it.

On Tasadr's site every page links to at least three others, with words that say what the reader will find, never "click here". The rest of preparing your site is in how do you prepare your site for AI questions? See also backlinks, XML sitemap and indexing.

Also known as inbound links · backlink

Backlinks are links on other sites' pages that point to your site, such as a Kuwaiti restaurant guide writing about your branch and linking to your page.

Google says it uses links as a signal when it determines how relevant pages are, and to find new pages to crawl. It counts buying or exchanging links to raise rankings as link spam, unless the links are marked nofollow or sponsored. A backlink is not a citation, the link an engine puts in its answer, as is a citation in an AI answer the same as a backlink? explains.

In AI answers the gain may lie in the page itself rather than its link: an article in a Gulf newspaper that names your business is a page an engine may cite, so your name reaches the answer from outside your site, as in how do AI engines learn about you beyond your own site? Do not confuse them with internal links, which join your own pages, and one honest way to earn backlinks is digital PR.

Also known as fresh content · content updates · updated content

Content freshness is your page's information being right today, its prices, hours and figures, and the page saying when it last changed.

Suppose a Dubai school's fees page still lists a past school year's fees: an engine may pass the old figure on to a parent. Changing the page's date does not fix that; updating the fees does.

Bing says freshness signals directly influence how quickly your updates reach search results and AI-generated answers. Google, for its part, says its generative AI features retrieve relevant, up-to-date pages from its index, while asking whether you change the date of pages to make them seem fresh when the content has not substantially changed.

A date alone does not make a page fresh, and on our own pages the "Updated" date moves only with a substantive change. Where freshness sits among the other factors is in which factors affect leading AI answers? Read it with XML sitemap and original data.

Also known as listicles · best-of articles

Best-of lists are articles that rank the businesses of one kind in one city, such as "the 10 best mandi restaurants in Riyadh" or "the best international schools in the Fifth Settlement, New Cairo", published by news sites, directories and blogs.

When your customer asks "what's the best mandi restaurant in Riyadh?", an engine may build its answer on such a list and cite it, so a place on it is earned by being worth naming, not bought, and Google says seeking inauthentic mentions across the web is not as helpful as it might seem. How to earn your way onto these lists honestly is in how does your business earn a place on "best of" lists?

Also known as Wikidata knowledge base

Wikidata is an open knowledge base that people edit together in many languages, in which every thing has an item with a number starting with Q, together with facts about it.

Wikidata accepts an item if it has a link to a page on Wikipedia or one of its sister projects, or refers to a clearly identifiable entity that can be described with serious and publicly available references, or fulfils a structural need. So a business does not get in on its own page alone, but through independent sources that write about it.

In our own structured data we link each engine we name to its item there: for us Gemini is item Q116698014, Google's chatbot, not the constellation. Where Wikidata sits among the places that mention your business is in how do AI engines learn about you beyond your own site?

Next to it in this glossary: entity, Knowledge Graph and structured data.

Also known as online PR · digital public relations

Digital PR is the work of earning mentions of a business on news sites, magazines and directories online, through real news, a figure from its own data or an expert view from its team, not through a paid advertisement.

In our measurements of Arabic questions about our own field in October 2026, Gulf news sites were among the sources Perplexity cited, as in where does Perplexity get its sources for Arabic questions? A paid article is another matter: Google counts its links as link spam unless they are marked sponsored or nofollow. What to give a journalist so they write about you is in when will the press write about you, and what should you give them?

Also known as local search optimisation · local search engine optimization

Local SEO is the work that lifts your business in searches tied to a place, such as "near me" or the name of a city or district, in Google's results and Maps.

According to Google, local results are based mainly on relevance, distance and prominence, meaning how well known a business is; more reviews and positive ratings can help your local ranking; and there is no way to request or pay for a better local ranking.

AI engines read the same facts. Google says Google Business Profiles can help your products and services appear in both its AI responses and its other search results, and OpenAI says ChatGPT may use the asker's approximate location to give local results. When a customer in Khobar asks for the nearest car wash open right now, the engine needs the district and the opening hours written as text on your page and your profile.

How local search differs from local answers is in what is local GEO? Close to it in this glossary: Google Business Profile, NAP and local pack.

Also known as customer reviews · ratings and reviews

Online reviews are the comments and star ratings customers leave about your business on your Google Business Profile, booking and delivery apps, and review sites.

Google says more reviews and positive ratings can improve your ranking in the local pack, and in our October 2026 research into Google's autocomplete suggestions, "top-rated" recurred in Saudi Arabia, Kuwait and the UAE, so the rating becomes part of the question itself, as in how do reviews enter a top-rated question?

Google's Maps content policy bans reviews paid for in cash or in kind, and incentives such as discounts, goods or services at no charge in exchange for a review. The honest route is to ask every customer after the service, as in how do you grow your reviews without breaking Google's policies? Reviews are one kind of user-generated content.

Also known as business directory · business listings site

An online directory is a site that lists businesses by field and city, each with its address, phone and opening hours, and often its customers' reviews.

A directory may be the site that leads the answer instead of you. In a real report for an electronics store in Egypt, measured between 12 and 26 October 2026, the sites most often cited instead of the store's site included online marketplaces and a classifieds platform.

So write your business's name, address and phone in every directory that lists you exactly as they are on your site, and correct any detail that is out of date. How to earn a real mention in directories and platforms is in how do AI engines learn about you beyond your own site? In this glossary it sits next to competitor in the answer, marketplace and booking platform and NAP.

Also known as marketplace · booking platform · aggregator

A marketplace or booking platform is a site that gathers products, services or appointments from many businesses in one place and handles the sale or booking, such as hotel booking sites, medical appointment platforms, food delivery apps and stores where many merchants sell.

These platforms can reach the citation in buying questions ahead of the business's own site: in a real report for an electronics store in Egypt, a classifieds platform was among the sites cited most often instead of the store's own. So claim your listing on the platform and complete it, as in how do you claim and complete your listings on directories and platforms?

Also known as UGC · user content

User-generated content is text written by the public rather than the site's owner: reviews, comments, forum posts, and visitors' questions and answers.

Google notes that its generative AI features can show what is being said about products and services across the web, including blogs, videos and forum discussions, but warns that seeking inauthentic mentions is not as helpful as it might seem. It counts spammy forum posts and blog comment spam as user-generated spam, and recommends marking links users add with the ugc value.

If someone asks on a forum about the best AC maintenance company in Dammam, and real customers answer from their own experience and name your company, that is a mention you did not write. Replies you post under made-up names are what Google's policies warn against.

The clean way to gather reviews is in how do you grow your reviews without breaking Google's policies? Close to it: online reviews, online directory and backlinks.

Also known as AI citation · source citation · cited source

A citation is a link an engine places beside a sentence in its answer to say the information came from that page.

It is the only route a reader has from the answer to your site. How each engine shows one, and how it differs from a mention and a rank, is in what is a citation in an AI answer?

Also known as Search Console · GSC · Google Webmaster Tools

Google Search Console is a Google service that shows a site's owner which of its pages are indexed, which searches its links were shown for, and how often they were clicked.

Some owners write it «سيرش كونسول». It holds the Page indexing report and the Performance report, and since June 2026 the Search generative AI performance report. How to read them with Bing's report is in what Google's and Bing's reports show about AI answers.

Also known as impressions · search impressions

Impressions are how often someone saw a link to your site on Google, as Search Console defines them in its Performance report.

A click is something else: how often someone clicked a link from Google to your site, and the click-through rate is clicks divided by impressions. In the generative AI report, impressions count your site's links in AI Overviews and AI Mode, as in Google's and Bing's reports. In Arabic we write «مرات العرض».

Also known as Search generative AI performance report · generative AI performance report

The Search generative AI performance report is a Google Search Console report, announced on 3 June 2026, that counts impressions of your site's links in generative AI features in Search, such as AI Overviews and AI Mode, and in Discover.

It breaks them down by page, country, device and date, and does not mention the Gemini app, so Gemini citations are found by asking the question itself, as in how can I tell if I'm being cited in Gemini?

Also known as Page indexing report

Indexing statuses are the reasons Search Console's Page indexing report gives for each page Google has not indexed, including these three:

• Discovered - currently not indexed. Google found the page but has not crawled it yet, usually because it rescheduled the crawl so as not to overload the site.
• Crawled - currently not indexed. Google crawled the page but did not index it. It may be indexed later, and there is no need to resubmit it.
• Duplicate, Google chose different canonical than user. Google thinks another URL makes a better canonical, so it indexed that one instead of your page.

Also known as Bing AI Performance

AI Performance is a Bing Webmaster Tools report, launched on 10 February 2026, that counts how often your site is cited as a source in Copilot answers, Bing's AI summaries and some partner integrations.

It shows total citations, the average number of your pages cited per day, citations per page, and the phrases the AI searched with when it retrieved your content, which it calls grounding queries. More in Google's and Bing's reports.

Also known as purchase question · commercial question

A buying question is a question that comes before a decision to pay: it names a service, and usually a city and a condition such as price or timing.

For example: which florist in Riyadh delivers the same day? That question, not a keyword, is the unit Tasadr measures, because an engine answers a question and does not rank words. It is written in your customer's words, not in the language of your internal deck.

How to gather ten worth measuring is in how to choose the questions you measure on.

Also known as informational query

An informational question asks for understanding rather than a decision, such as "what is the difference between clear aligners and metal braces?", and usually names no city and no price.

It helps build trust, but it sits further from a purchase than a buying question, which is why we start measuring with buying questions. The two are compared, with examples, in what is the difference between a buying and an informational question?

Also known as question gap · missed questions

A question gap is the questions your customers ask that someone else leads the answers to, which you neither measure nor answer on your pages.

Tasadr shows them beside your questions, each with the kind of site that leads it, as in the sample report, which shows under each question the kind of other site cited most on it. How we build it is in how we measure your score.

Also known as AI referrals · LLM referral traffic

AI referral traffic is the visits your site gets when a reader clicks a link in an AI engine's answer, which analytics services record as coming from the engine's domain, such as chatgpt.com, perplexity.ai or gemini.google.com.

It is part of the picture, not all of it: your customer may read the answer and stop there, and a visit that arrives without a source may be counted as direct traffic. Clicks from AI Overviews and AI Mode are counted by Google within search traffic in Google Search Console, under the Web search type. How to read it step by step in GA4 is in how do you see ChatGPT and Perplexity visits in GA4?

Also known as UTM tags · UTM codes · campaign parameters

UTM parameters are words added to the end of a link to your site, chiefly utm_source, utm_medium and utm_campaign, that tell your analytics where the visit came from.

Google's Analytics Help recommends always using those three, and warns that their values are case-sensitive. Google Analytics 4 puts a visit in its "AI Assistant" channel when utm_medium is exactly ai-assistant, or when the site it came from is on its list of AI assistants.

An illustrative example: a fictional clinic's link in its Google Business Profile reads falak-dental.sa/?utm_source=google&utm_medium=organic&utm_campaign=gbp, so its reports show how many visits came from the profile alone. The links an engine puts in its answer are not yours to tag, and parameters change nothing in the answer itself: they only name the visit once it arrives.

How to read these visits in full is in how do you see ChatGPT and Perplexity visits in GA4? Next to it in this glossary: AI referral traffic, Google Analytics 4 and direct traffic.

Also known as GA4 · Google Analytics

Google Analytics 4 is Google's service for measuring visits to websites and apps, recording visits, where they came from and what visitors did, and it replaced Universal Analytics, whose standard properties stopped processing data in July 2023.

In it, AI visits count as referrals from the engines' domains, such as chatgpt.com or perplexity.ai, so you can see which page received them and what their visitors did next. The method is in how do you see ChatGPT and Perplexity visits in GA4? See also AI referral traffic and UTM parameters.

Also known as direct visits · Direct channel · (direct) / (none)

Direct traffic is visits Google Analytics cannot trace to a source, so it files them under its "Direct" channel; they include visits from people who type your address or open it from a saved link.

Google says a session is processed as direct when no information about the referral source is available. Among the causes it lists are links without UTM parameters, redirects that strip them, URL shorteners, links inside PDF or Word files, and ad blockers that interfere with cookies.

This matters for AI answers: a customer who read your shop's name in an answer with no link, then typed your address into the browser, is counted as direct, not as a visit from AI. So do not judge what answers do for you from the referral channel alone; look at the answer itself, and whether it cited you or only mentioned you without a link.

The full method is in how do you see ChatGPT and Perplexity visits in GA4? Related terms: UTM parameters, AI referral traffic and zero-click search.

Also known as CTR · clickthrough rate

Click-through rate is clicks divided by impressions, as Google Search Console calculates it in its Performance report: if your link was shown 1,000 times and clicked 30 times, your click-through rate is 3%.

When Google writes an AI summary above the results, a searcher may stop at what they read. In a Pew Research Center study published on 22 July 2025, based on the browsing of 900 US adults in March 2025, users clicked a traditional result link in 8% of visits to Google pages with an AI summary, against 15% on pages without one, and clicked a link inside the summary itself in 1% of them.

So your click-through rate can fall while your position holds, and on its own it cannot tell you where you stand in AI answers, as in what do Google's and Bing's reports show about AI answers? Read it with impressions and average position, and see also zero-click search.

Also known as average ranking position · position

Average position is a figure in Google Search Console's Performance report that says where your site's topmost link sat in search results, on average, with position 1 at the top.

Google takes the topmost position a link to your site held in each search, then averages: if your links held positions 2, 4 and 6 in one search and 3, 5 and 9 in another, the average is 2.5. It says an AI Overview occupies a single position, which every link inside it shares, and that AI Mode follows the same method as a regular results page.

Read it for what it is: it says where your link sat in a list, not whether an engine's answer carried your sentence or credited you. A Kuwaiti perfume shop's page may average position 3 for "oud perfume" and still not be cited in ChatGPT's answer to which shop in Kuwait is best for oud.

How to read Google's and Bing's reports together is in what do Google's and Bing's reports show about AI answers? Read it with impressions, click-through rate and rank tracking.

Also known as branded search · brand query

A branded query is a search in which the searcher types the name of your business or of a product that is yours alone, in any language or spelling, rather than the name of the service.

Google Search Console separates it from other queries with a branded queries filter in its Performance report, announced on 20 November 2025 and opened to every eligible site in March 2026. Google says the filter counts your brand name in all languages, with its typos, and that branded queries typically bring higher-ranking pages and higher click-through rates.

An illustrative example: for a chain called Rukab Restaurants at rukab-restaurants.com, a search for its name in Arabic and "rukab restaurants menu" are both branded queries, while "best mandi restaurant in Riyadh" is not. Asking an AI engine about your own name shows you what it knows about you, but the real contest is in the questions without your name, as in what are the common mistakes when you check yourself? In measurement the buying question matters more, and what lets an engine tie your name to your business is the entity.

Also known as keyword rank tracking · keyword tracking · SERP tracking

Rank tracking is following where your pages rank in search results for chosen phrases, day after day, often with a separate rank for each city and device.

It is a familiar part of SEO, but it measures your place in a list of links, not your place in the answer: first place in Google does not mean an engine will cite your page. A chat engine's answer also moves, and OpenAI says the same question may yield different answers across queries. So where you stand in answers cannot be read from a single snapshot.

An abaya shop in Riyadh may rank first for "soft black abayas", yet when a customer asks ChatGPT which shop in Riyadh has the nicest abayas, it may cite a list on a fashion site and not name the shop. The unit here is a whole question, not a keyword, and the result is cited, mentioned or absent, not a rank number.

The full difference is in a rank or a citation: what changes in the result? Related terms: average position, answer variability and the Tasadr score.

Also known as AI answer volatility · answer drift

Answer variability is an AI engine giving a different answer to the same question from one time to the next, in its text, its sources or the businesses it recommends.

So never judge where you stand from one answer; at Tasadr a movement is not called a change until it holds, as in confirmed change. What usually changes and what holds are in why do ChatGPT's answers change, and when is a change real?

Also known as cited

Cited means your site's link is among the sources the engine relied on in its answer, whether or not it named you in the text; it is the best of the four outcomes we record for every answer.

It is the possible visit, and a sign that the engine treated your page as a source it read rather than a name it remembered.

Also known as mentioned without a link · brand mention

Mentioned means the engine wrote your brand's name in the answer with no link to your site.

The reader learns you exist but finds no route to you in the answer, and goes back to searching. It is better than being absent and short of being cited, and it can mean the engine knows your name but found no page on your site that answers the same question.

Also known as absent

Absent means the engine wrote an answer that neither cites your site nor names you.

It may have named a competitor, or nobody at all. When it names someone else, we record who was cited instead of you and in what words, then put their page beside yours to show what theirs answers and yours does not.

Also known as no answer

No answer means the engine wrote no answer to that question in that measurement, and it is never counted against your site.

The difference from absent is the difference between an engine that chose someone else and an engine that chose nobody. Confusing the two makes the score lie, because charging you for the engine's silence lowers your number over an answer that was never written.

All four outcomes together are in the four outcomes per answer.

Also known as tasaddur · leading AI answers

Tasaddur is being the source an AI engine builds its answer on, not a name passing in a line.

It is the result GEO works towards and the word we measure by. The Arabic word means taking the lead, and it gives Tasadr its name. A site that leads a question is the site cited in its answer; a site mentioned without a link has not led it yet, even though its name was written.

Also known as Tasadr score · tasaddur score

The Tasadr score is a number out of 100 that says where your site stands inside AI engines' answers to your customers' questions.

It is built from every answer to every question on every engine. “No answer” is never counted against you. When a measurement spans a range, we show the range beside the score, so you know how settled the number you are reading is.

Under every verdict in it sit the answer's excerpt, word for word as the engine wrote it, its sources and its date, so you can check it yourself. What the number says and does not say is in what the score out of 100 tells you.

Also known as confirmed change

A confirmed change is a movement that passes our fixed rule rather than one odd answer, and it is the only thing that reaches you as an alert in your dashboard.

The rule is written in advance and is the same for every account, so headlines and alerts do not flicker with every answer; what does not pass it stays in your answer history to read whenever you like. The detail is in when we say something has changed, and why answers change at all in why do ChatGPT's answers change, and when is a change real?:

• A new competitor: we do not call it new on a single answer; the change has to hold.
• A lost citation: we do not report it lost on a single answer without it; the absence has to hold.
• A new gain: counts at once, because a citation that happened has happened.

Also known as answer competitor · competitor in AI answers

A competitor in the answer is a site the engine cited for your question instead of you, and it may not be your competitor in the market or in the search rankings.

It may be a directory, a comparison site, or a small business that wrote one precise page. We put its page beside yours to show what it answers and yours does not, and the recurring gaps become an action plan ranked by priority.

How many of these analyses you get a month follows your plan, as set out in Tasadr plans and prices. How to find and follow one is in which competitor ChatGPT recommends instead of you.

Also known as task list · prioritised action plan

Priority Fixes in Tasadr is your list of tasks ranked by priority; each task is a question your customers ask the AI engines, and a competitor leads it instead of you.

Each task has three steps, in order: "Find out why they lead", then "Edit your page", then the "Implementation check". The work itself is set out in numbered steps that say what to do on your page and who does it: you, your developer or your content writer. The task reaches whoever applies it by copy, on WhatsApp or through a private link.

For example, in a real report for an electronics store in Egypt the price questions were the store's strongest: 2 of the 3 engines relied on its site for “How much does a 1.5-horsepower air conditioner cost on Egyptian shopping sites?”. So the first step we suggested was pages that answer the price of each of its product categories directly.

All the stages are in execution: the opportunity becomes a task for whoever applies it. Next to it in this glossary: why-they-lead analysis, implementation check and impact.

Also known as competitor analysis in AI answers · AI competitor analysis

A why-they-lead analysis is a comparison between your page and the page of the competitor an engine cited instead of you for one question, explaining why the engine chose that page and what to change on yours first.

It differs from the classic kind in that you do not write the list of competitors: the engine writes it in every answer. At Tasadr every why-they-lead analysis comes with on-page fixes ready to publish, and how many you get each month depends on your plan, as in what is a why-they-lead analysis? How to find that competitor and track it is in which competitor does ChatGPT recommend instead of you? It rests on the competitor in the answer, and its findings build Priority Fixes.

Also known as edit verification · implementation verification

The implementation check is the step in Tasadr where we check your page after you say you have edited it, to confirm the edit is really live before we measure its impact.

A task does not become "Done" because you pressed a button. If we find no change on the page, we tell you and the task stays open; if we find part of it, we tell you what is missing; when we find the whole edit, the task becomes "Done" with the date of the check, and anyone who opens the task's link sees that we verified it.

Say your developer wrote the new price paragraph on the service page and forgot to publish it, so it sits in a draft. The check tells you before you wait for a result that will not come, which is why every impact you see in Tasadr follows an edit we verified, not a "Done" mark alone.

More in verification: was the edit really made? Related terms: Priority Fixes, impact and confirmed change.

Also known as impact measurement · measuring the impact of a change

Impact is the result of a question whose page you edited, read in the measurements that follow the edit, to see whether your site is now cited, whether your score moved, and which competitors gained or lost ground.

At Tasadr it comes after the implementation check, so no impact is counted for an edit we did not find on the page. We compare the question you worked on with the ones you left untouched, over the same measurements before and after your edit; it is a comparison, not a guarantee, because answers change for many reasons, as in answer variability.

If you connect Google Search Console, you see your site's clicks beside your Tasadr score week by week. The two lines moving together shows timing, not cause: one does not prove the other. More in impact: did the edit change your standing?

Also known as AI readiness audit · site readiness check

An AI readiness check reads your site itself, not the engines' answers, to say whether crawlers can reach it, whether its pages are indexed and whether its text is written in a way that may make it easier for AI engines to understand and quote.

Google says its generative AI features require only an indexed page eligible to be shown with a snippet, with no special file or markup, so a serious check is short. A readiness score does not say whether an engine cited you, or whom it cited instead; only measuring the answers themselves does, as in what is the difference between a readiness check and measuring the answers?

An illustrative example: a Riyadh law firm's site passes every check, then is absent from the answer to which commercial litigation lawyer in Riyadh is best, perhaps because a lawyers' directory answers the question in its first line. Close to it: crawler, indexing and server-side rendering.