How to know if your site shows up in ChatGPT and Gemini

Your site leads the answer when the engine picks it as a source and puts its link in the answer. When your customer asks ChatGPT or Gemini about a service in their city, the engine writes one paragraph that points to two or three sources, and any site that is not among them has missed that answer.

What you look for in each answer is one of four outcomes, which every page of Tasadr (تصدَّر) calls by the same names:

• Cited: your site's link is among the sources the engine relied on, even if it did not write your name. That is leading the answer.
• Mentioned without a link: it wrote your name with no route back to you.
• Absent: it wrote an answer that neither links to your site nor names you.
• No answer: the engine wrote no answer to the question that time.

You need a table, the questions your customers type, and seven steps in order. You do not need any tool to start.

Ask Gemini the questions your customers ask, in their words, then open the sources under its answer: a link to your page means Gemini cited you; your name alone is a mention.

Five things to know before you judge:

• Where Gemini puts its sources: Google says the Sources button sits at the bottom of the response or inline in it, and opens a side panel with the links. If there is no Sources button under a response, Gemini gave no links for it, so record it as mentioned without a link or absent, depending on what it wrote.
• Search Console does not count the Gemini app: Google's Search generative AI performance report, announced on 3 June 2026, counts impressions in AI Overviews, AI Mode and Discover's generative AI features. It does not mention the Gemini app, so do not look for Gemini citations there.
• Visits from Gemini: in GA4, visits referred by gemini.google.com tell you a reader opened your link from an answer, not how often you were cited and nobody clicked.
• Repeat before you judge: Gemini's answer can change from one day to the next, so ask the same question on different days. Why answers change, and how to tell a real change, is in why do ChatGPT's answers change, and when is a change real?
• In Tasadr: we measure Gemini on every plan, the free trial included, and keep each verdict with the answer's excerpt, its sources and its date.

Each pattern in your table says something different about your site:

• Absent on most questions: you have no page answering that question, or the engine cannot reach it. Start with one page.
• Mentioned without a link: the engine knows your name but finds no page on your site to cite. Write a page that answers the question in plain text.
• Cited on ChatGPT and absent on Gemini, or the other way round: each engine picks its sources its own way. Read the pages cited by the engine you are absent on.
• No answer: a question nobody has won yet, neither you nor your competitor. An open opportunity for a page that answers it.
• One competitor keeps coming back: that is your competitor in the answer, even if they are no competitor of yours in the market. Start with the question where they are ahead of you.

ChatGPT and Gemini are a start, not the whole picture. Your customer may read the Google AI Overviews summary above the search results before opening any chat, or ask Google AI Mode, Perplexity, Copilot (Microsoft's answer inside Bing search), Claude or Grok.

The same check works on each of them, with two differences: Google's answers in AI Overviews and AI Mode can differ with the asker's country, and AI Overviews does not write a summary for every Arabic question in the first place. What sets each engine apart is on the eight AI engines we measure in Arabic, and the summary in detail on the guide to Google AI Overviews in Saudi Arabia.

A check by hand answers for today, not for the month: answers change from one day to the next, your account, city and chat history can colour what you see, and ten questions on 8 engines mean 80 answers to read and record every time.

When a change is real rather than chance, and how to test it yourself in three days, is in why do ChatGPT's answers change, and when is a change real?.

Tasadr answers your check's question regularly, keeps every result with its date, and adds what a table cannot give you. How we measure is explained in how we measure the Tasadr score, and you can see all of it in a real report for a clinic group in Riyadh and Jeddah:

• Questions in your customer's words and dialect, written for their own market.
• Google's answers as your customer sees them in their country, not a generic version.
• Repeated measurement of every question, so one lucky answer is never counted as a change.
• An excerpt of the answer as evidence under every verdict, with the links the engine put in it.
• The Tasadr score out of 100, with its range whenever a measurement spans one.
• Who is cited instead of you, why, and what to change first.
• An alert in your dashboard once a change is confirmed, not before.

An action plan in five steps, starting from one question. The work as a whole is called optimising for AI answers (GEO).

Six mistakes that make your check look better or worse than reality:

• Asking about your business by name. The engine will find you, but your customer asks about a need, not your name.
• Asking in English about a market that asks in Arabic, and reading an answer about another market.
• Judging from one answer, or from one day.
• Asking from your usual account, with a long history about your business.
• Counting a mention without a link as a citation. A name alone does not bring the reader to you.
• Checking ChatGPT alone, when your customer may read Google's answer first.