Skip to content
Built for AI agentsConnect yours
ChatGPT logo

ChatGPT

ChatGPT visibility tracking: get recommended by ChatGPT

ChatGPT is where buyers now ask the question your category page was written to answer. See which of its answers name a competitor instead of you, the live pages it read to decide, and whether a change you made moved it.

Start 7-day free trialAll platforms

Card required, nothing charged for 7 days

  • Search product surface
  • Six days a week
  • Asked from your market

Search product surface

ChatGPT tracking in the search product your buyers open

We ask ChatGPT where your buyers use it: the search product, which fetches live web pages before it answers. The brands it recommends there are the ones your buyers hear about. The API is a different product with its own retrieval, so its answers would describe something else. Every prompt runs six days a week, asked from the market your workspace measures.

Six days a week.
The same prompts on a fixed schedule, so a change in the answer is a change in ChatGPT, not in what we asked.
From your market.
Asked the way a buyer in the country your workspace measures would ask it.
Beside the other five.
ChatGPT's answer sits in the same list as every other engine's answer to that prompt, so who named you and who did not takes one glance.

Asked from your country

ChatGPT's country list decides which markets you can measure

Every ChatGPT question goes out with your workspace’s country and language, so the answer is the one a buyer there would read. The markets on offer are generated from the countries our ChatGPT collection can ask from. A market has to work on every engine that takes one, and ChatGPT’s list is the narrowest of the four, so it sets the menu for all of them.

Country and language, every time.
Both are sent with each ChatGPT question, not inferred from where our servers sit.
Four engines move with it.
ChatGPT, Gemini and Google's two surfaces follow your market. Claude and Perplexity answer the same everywhere, and the screen says so.
A new market starts a new series.
Answers asked from another country are a different measurement, so the trend restarts rather than pretending to continue.

Answer and sources

The live pages ChatGPT read before it named you

ChatGPT builds each answer from pages it fetches when the question is asked, so its sources show where to work. We keep every answer word for word, the brands in the order it named them, and every page it cited, with your own pages marked. Sites it keeps reading that name a competitor and never you are the places to get onto.

Where it placed you.
Second place is a different outcome from first, so the position of every brand it named is kept.
The pages it read.
Every source, in the order ChatGPT listed them, yours and your competitors' alike.
The answer itself.
Stored verbatim, so any rate on your Overview traces back to the sentence behind it.

What ChatGPT's search product reads that its API does not

Two products share the name. Only one of them is the answer your buyers see.

Calling ChatGPT through its API is the easy way to measure it, and it measures a different product. We ran both routes side by side on the same questions before choosing, and collect from the consumer search product, the ChatGPT your buyers use, keeping every live page it cited with the same citation tracking the other engines get. The method is stored beside every answer, so if the route ever changed, a trend across the switch would read as a change of route rather than a change in ChatGPT.

How ChatGPT answers through its API and through its search product, as RankHound collects it
Who reads itAPI:Developers, inside apps they build. It is not the chatgpt.com your buyers open.Search product:The ChatGPT people open at chatgpt.com to ask which tool to buy.
Searching the webAPI:Only when the caller switches the search tool on.Search product:Every time. The search product fetches live pages before it answers.
RetrievalAPI:Its own, so it can name different brands to the same question.Search product:The product's own, which is the retrieval your buyers get.
LocationAPI:Whatever the caller sends, if anything.Search product:Your workspace's country and language, on every question.

Per-engine visibility

ChatGPT visibility moves first, so it gets its own line

Because ChatGPT reads live pages, it is usually the first engine to reflect a change you made, which makes it the first place to look for whether that change worked. Its rate sits on its own card with the count behind it and is never blended with Claude, Gemini, Perplexity or Google's two surfaces, so engines that have not caught up yet cannot dilute a move here.

When the line moves

When ChatGPT starts naming you more, check what you just changed

ChatGPT reads live pages, so a page you rewrote last week can be in its answers this week, which makes it the quickest place to see whether a change got you recommended more often. When its rate rises, you get an email that says so, with both weeks’ counts and a nudge to look at what changed on your side around then. A fall sends the same email the other way. It is the same AI brand monitoring every engine has, read on ChatGPT’s answers alone.

This week against last.
Two seven-day windows of ChatGPT answers, never a single day and never pooled with the other engines.
Enough answers both weeks.
At least 10 ChatGPT answers in each window by default, so a quiet week cannot pass for a drop. You can raise or lower it.
Bigger than the noise.
The move must beat 15 points by default, and the variance expected at that sample size. You set the first in settings; the second is arithmetic.

Getting recommended by ChatGPT, question by question

If something here is unclear, ask us before you sign up rather than after.

Start 7-day free trial

Card required, nothing charged for 7 days

How do I get ChatGPT to recommend my brand?
Start with the answers where it named a competitor instead of you, and the pages it read for them. ChatGPT builds each answer from live pages, so the work is getting your brand onto the pages it reads and making your own pages say plainly what you are the right choice for. The same prompts keep running six days a week, so ChatGPT's own rate shows whether a change moved its answers. No one can promise that ChatGPT will name you, and we do not.
How do I track my brand in ChatGPT?
Paste your domain and we propose the competitors and the buyer questions to track; nothing runs until you confirm them. Each question is then put to ChatGPT's search product six days a week, from the market your workspace measures, and every answer is kept verbatim with the brands it named, the order it named them in and the live pages it cited. Collected from the search product surface, which retrieves live web pages.
Is RankHound a ChatGPT rank tracker?
It records what people usually want from one: where ChatGPT placed your brand among the brands it named, prompt by prompt, six days a week. ChatGPT holds no fixed ranking, though. Each answer is generated afresh, so the numbers worth reading are the share of answers that named you, your place when it did and the pages it read, each with the count of answers behind it.
Why does ChatGPT's answer change from day to day?
Its answers are generated rather than looked up, and the pages it fetches change from day to day, so wording, brand order and citations all vary between runs. That is why we report the share of ChatGPT answers that named you, with the count behind it, instead of a single yes or no, and show a fraction rather than a percentage below five answers.
What makes ChatGPT different to measure?
ChatGPT's search surface retrieves live pages, so its answers move with your content rather than with a training cut-off. That makes it the engine where changes you make show up soonest, and the one worth watching weekly rather than quarterly.
Why not track ChatGPT through its API?
Because the API is not what your buyers use. It is a separate product with its own retrieval, so it can name different brands to the ChatGPT people actually open. We read the search product, and record the method beside every answer, so a change in how we collect can never pass for a change in ChatGPT.
Is ChatGPT averaged into one visibility score?
Never averaged. ChatGPT reads live pages through its search product, Claude and Perplexity answer through an API, and Google's two surfaces come from the results page. A blended score would hide which one you are losing, so ChatGPT keeps its own card and its own count, and any figure that pools the engines, like your headline, says so on the screen.
Which countries can I track ChatGPT mentions in?
Any market on the list in your settings. Each ChatGPT question is asked with your workspace's country and language, and the list itself is generated from the countries our ChatGPT collection can ask from, which is the narrowest list of any engine we track. One brand is measured in one market; to track the same brand in another country, add a brand for that market.
Will I be told when ChatGPT starts or stops naming my brand?
Yes, by email. We compare ChatGPT's answers over the last seven days with the seven before, and write to you when the rate moves by more than both your threshold (15 points by default) and the variance expected at that sample size, with at least 10 answers in each window by default. Both thresholds are adjustable in settings, and a complete stop on an engine that used to name you is reported on its own.

Related reading

7-day trial

Find what it takes to get recommended by ChatGPT

Start with your domain and the questions your buyers ask. ChatGPT's first answers are readable in full within minutes, with every page it cited, and each week after shows whether your changes moved them.

Start 7-day free trialHow to measure AI visibility

Card required, nothing charged for 7 days.

On every plan

  • Every answer stored in full, with its citations.
  • Every rate shown with the number of checks behind it.
  • Cited-but-not-named reported as its own state.