ChatGPT search console

Not a public one, as of 7 November 2026. OpenAI publishes documentation for its crawlers, OAI-SearchBot for ChatGPT search and GPTBot for training, and it tags referral links from ChatGPT search with utm_source=chatgpt.com so you can see visits in your own analytics. It has not published a console where a site owner can see impressions, queries or rankings inside ChatGPT.

That is likely to change. On 6 November 2026 Search Engine Roundtable reported a screenshot, first shared on LinkedIn, of a ChatGPT banner inviting businesses to claim their business in ChatGPT. The reported tool counts how often a verified business appears in ChatGPT responses and how often people click through to its website, charted daily. Verification reportedly runs through connections such as a Shopify store. OpenAI has not announced it, and details may change before any launch.

What it reportedly does not show matters as much. According to the people who saw it, there is no prompt data, no user data and no competitor benchmark. You would learn that you appeared, but not which questions triggered it, what the answer said about you, or who was named instead of you.

Google Search Console reports generative AI performance for Google's own surfaces, AI Overviews and AI Mode, as impressions and clicks. It does not cover ChatGPT, Gemini as a separate app, Claude or Perplexity, and it does not show the answer text.

Bing Webmaster Tools added an AI Performance report in public preview in February 2026. It counts how often your pages are used as sources in Microsoft's own AI answers and partner experiences, grouped by grounding query and by page. It has no click data, and it does not report on ChatGPT, even though ChatGPT search has historically drawn on Bing results.

So each first party console describes its owner's surface, in that owner's terms. None of them puts ChatGPT, Gemini, Claude, Perplexity and Google's AI answers side by side for the same question, and none shows which competitor took the answer.

A Google style console starts from impressions and clicks because a results page is a list of links. An AI answer is different: the answer is the destination, and the links are footnotes. A useful AI search console therefore has to record what was said, not only whether you were there.

You write the questions your buyers actually ask. Mentionry puts each one to the answer engines on your plan once a day, three samples a reading, and keeps every result. ChatGPT is on every paid plan. The self-serve plans read ChatGPT, Gemini and Google AI Overviews; Perplexity and Claude can be added on an Enterprise quote.

Each reading stores the answer text, the URLs cited, the brands named and the queries the model searched along the way. From those records Mentionry computes visibility, share of voice and average position, grouped by engine, prompt, topic, day and competitor.

The Pages screen turns the citations into a site report: every page of yours that an engine has cited, the prompts that led to it, and a separate list of pages that have never been cited. The Agent Analytics check reads your robots.txt and reports which AI crawlers, OAI-SearchBot among them, are allowed in, because a site that blocks the crawler cannot be cited by ChatGPT search however good the page is.

On paid plans you can also connect your own Google Search Console property. Mentionry then reads your Google queries, pages, clicks and impressions on a schedule and keeps them beside the AI answers, so Google search performance and AI answer visibility are reviewed in one place without being mixed into one number.

ChatGPT has no fixed rank, because the same question can get a different answer an hour later. That is why each reading takes three samples and why trends matter more than any single day. Visibility is the share of answers that name you. Share of voice compares how often you are named against the competitors in the same answers. Average position is the order in which brands are named.

Two rules keep those numbers honest. A brand that was never mentioned has no average position, rather than a zero that would sort it first. A day when every run failed is shown as a gap in the chart, not as a drop in visibility.

Tracking mentions shows you where you stand. Changing it usually means earning a place on the pages the engines cite. Mentionry lists those cited pages and the openings around them, such as journalist requests, podcasts and resource pages, and drafts the pitch for you to review and send from your own connected mailbox. Whether a page then mentions or links to you is that publisher's decision, and nothing here is promised.

If OpenAI ships the reported tool, use it. First party impressions and clicks from ChatGPT are numbers only OpenAI can see, and no outside tracker can reproduce them.

The two answer different questions. An OpenAI console would tell you how often you appeared and how many people clicked. A tracker tells you which of your buyers' questions you appear for, what ChatGPT said, which competitor was named instead, and which pages it relied on. The first measures the outcome; the second shows what to change.

It measures your own tracked prompts, not the world's conversations. It cannot see what other people type into ChatGPT, and it does not estimate how many people asked a question. The Search Demand screen shows Google keyword volume and a separately labelled, modelled AI search estimate from DataForSEO; neither is a count of real ChatGPT prompts.

It does not have ChatGPT impressions or click counts. Those exist only inside OpenAI. Visits from ChatGPT appear in your own analytics under utm_source=chatgpt.com.

Recording starts the day a prompt is added. There is no backfill, because the answer text and the competitors inside it exist only if they were captured at the time.

It does not run on Microsoft Copilot, Grok or DeepSeek today, and it does not track them.