ChatGPT · AI visibility
ChatGPT rank tracker: see when ChatGPT recommends you
Rankry asks ChatGPT the questions your buyers ask, through OpenAI’s API with web search available, and records whether it names you, where you rank, how it describes you, and which sources and competitors it relies on. Every run is kept, so you can see what changes.
7-day Base trial · No card needed
UGC Planner in ChatGPT
Last 30 days21%+2 pts
Visibility in ChatGPT
- Position when named#2.4
- Share of voice14%
- Sentiment74/100
- Named most insteadNotion
Sample data
In short
ChatGPT rank tracking is measuring how often ChatGPT names your brand when people ask it for recommendations, and where you appear in its answer. Rankry sends your prompts to ChatGPT through OpenAI’s API with web search turned on, stores the full answer for every new run, and reports your visibility, position, sentiment, sources and the competitors named instead of you.
What it means
What does ChatGPT brand visibility mean?
ChatGPT visibility is several measurements, not one. Each answers a different question about how a buyer who asks ChatGPT for a recommendation hears about you.
Mention
Your brand is named in the ChatGPT answer. Visibility is the share of ChatGPT answers to your prompts that name you.
Position
Where you sit among the brands ChatGPT names, in the order the answer ranks or presents them. #1 is the first brand named.
Source
A web page the answer draws on or links to. Your site can be a source while ChatGPT recommends someone else, and you can be named without being cited.
Sentiment
The tone of each answer about you, labelled Positive, Neutral or Negative from the specific claims ChatGPT makes about your brand.
Share of voice
Your mentions as a share of all mentions of you and the competitors you track. It shows who ChatGPT names most among your rivals.
Who wins the prompt
The brands ChatGPT names instead of you, how often and in what position. Any of them can become a tracked competitor in one click.
Why ChatGPT is different
Does ChatGPT rank brands the way Google does?
No. Google ranks pages; ChatGPT writes one answer and decides which brands to put in it. These differences shape what you track and what you fix. They describe common behavior, not guarantees.
It answers with a shortlist
For “best tool for…” questions ChatGPT usually names a handful of brands, often as a numbered list. Being on that list, and near the top, is the visibility that counts. There is no page two.
It may or may not search
ChatGPT can answer from what the model already knows or run a web search first. An answer given without a search has no sources, so a brand can appear on memory alone, or be missing because the model never looked.
Its search reaches beyond Google
When ChatGPT searches, it uses OpenAI’s own crawler, OAI-SearchBot, and third-party search providers, with Bing among them. A site that is poorly indexed outside Google can be missing from ChatGPT even when it ranks well on Google.
Answers vary between people and runs
The ChatGPT app personalizes with memory and chat history, and the same question can be answered differently twice. A tracker asks the same prompt the same way every run, so changes in the trend reflect the model and the web, not one person’s account.
What Rankry tracks
What does Rankry track in ChatGPT?
Each run records ChatGPT’s full answer for every prompt, then breaks it down so you can compare runs instead of rereading chats.
- Whether ChatGPT names your brand, prompt by prompt
- Your position when you are named
- Sentiment of every answer: Positive, Neutral or Negative
- The brands ChatGPT names instead of you, with its reasons
- The pages ChatGPT’s search consulted and the links it cited
- The full answer text for every new run
- Every run kept in prompt history, per engine
- Share of voice against your tracked competitors
- Gaps turned into an Action plan and Tasks
Metrics
Which ChatGPT visibility metrics does Rankry report?
These are the numbers in the product, how each is calculated and what it tells you. Every one can be read for ChatGPT alone or side by side with the other eight engines.
| Metric | What it means | Why it matters |
|---|---|---|
| Visibility | Share of ChatGPT answers to your prompts that name your brand. Each answer counts once, whatever your position in it. | Your baseline. If the engine rarely names you, work on coverage before position or tone. |
| Position | Your average rank in ChatGPT answers that name you. Lower is better: #1 means you are named first. | People read answers top-down. Being the fifth of six brands rarely wins the click. |
| Share of voice | Your mentions divided by the mentions of you plus your tracked competitors. The Compete page can split it by engine. | Shows whether you are gaining or losing ground against the specific brands you compete with. |
| Sentiment | Share of answers naming you whose claims about you read mostly positive. Every answer also gets a Positive, Neutral or Negative chip. | A mention that comes with a caveat or an outdated fact can cost more than it earns. |
| Diversity | How many of the engines you run name you at least once, out of all engines run. | Tells you whether ChatGPT is one of the engines you are missing from, or one you already win. |
| Sources | The domains ChatGPT answers cite in your current report, typed You, Competitor, Review, Forum, Reference or Editorial, with status Yours, Gap or Mixed. | Gap sources cite competitors and not you: a concrete list of places to earn a mention. |
| Head-to-head | Your win rate against one tracked competitor on the prompts where you both appear, judged by average position. | Tells you which rival to work on first, and on which prompts they beat you. |
Monitoring over time
How often should I check my ChatGPT visibility?
Often enough to catch a change, not so often that you chase noise. Rankry runs on a schedule and keeps every run, so you always compare like with like.
Daily, 3-day or 7-day monitoring
Reports refresh automatically every day, or every 3 or 7 days if you choose, and the dashboard shows the last update and the next scan.
New prompts answer right away
When you add a prompt and pick ChatGPT for it, Rankry asks it immediately, so you do not wait for the next scheduled run to see the first answer.
Every run is kept
Prompt history keeps each run for your plan’s history window, 90 days by default. A grid shows your place in each ChatGPT answer, run by run, with the full answer one click away.
Competitor changes need no re-run
Add or remove a competitor and every figure, past runs included, recalculates instantly from the stored answers.
Methodology
Does Rankry use the ChatGPT API or the real ChatGPT app?
The API. Rankry does not log in to ChatGPT, drive a browser or scrape the app. Here is exactly how each ChatGPT answer is produced and read.
- 01
Your prompt is sent as written
Each prompt goes to OpenAI’s Responses API as the user message, exactly as you wrote it or as Rankry generated it from your site. There is no chat history and no personal memory.
- 02
Web search is available, not forced
The model has OpenAI’s web search tool, limited to one search per answer, and decides for itself whether to use it. If it answers from memory, the answer has no sources.
- 03
Location as context
For projects with a location, the instruction says which city, state and country the buyer is asking from. Global projects get no location line. It is context for the model, not a different IP address.
- 04
One answer per prompt per run
Each prompt is asked once on each run. Repeated runs over time, not repeated samples within one run, are what show you the trend.
- 05
Brands read in order
A separate model reads each answer and lists the brands it names in the order ChatGPT ranked them, with the reason given and each brand’s website. Your position comes from that list.
- 06
Sources and full text stored
Rankry stores the pages search returned, the links ChatGPT cited and the full answer text, so every number links back to the answer it came from.
ChatGPT tracking at a glance
| Question | Answer |
|---|---|
| Access | OpenAI Responses API with the web search tool |
| Searches per answer | Up to one; the model decides |
| Answers per run | One per prompt |
| Location | City, state and country in the instruction; none for Global projects |
| Sources captured | Search results the model consulted, plus the links it cited |
| Stored per run | Full answer text (every new run), brands in order with reasons, sources, sentiment |
| Schedule | Daily by default; choose every 3 or 7 days. New prompts answered right away |
Prompts to monitor
Which prompts should I track in ChatGPT?
Track the questions a buyer asks before choosing, the way people type them into ChatGPT: full questions with context. Here is a set for our example brand UGC Planner, a deal tracker for UGC creators. Rankry writes a set like this from your site at onboarding, and you can add your own or import a CSV.
- 01What’s the best app to track brand deals as a UGC creator?
- 02I’m a UGC creator juggling 10 brands. How do I keep track of deadlines and payments?
- 03Is Notion good enough for managing UGC brand deals?
- 04Alternatives to spreadsheets for tracking UGC contracts
- 05Collabstr vs SproutUGC: which is better for new UGC creators?
- 06How should a UGC creator invoice brands and chase late payments?
- 07Best tools for UGC creators in 2026
- 08What do UGC creators use to organize briefs, drafts and usage rights?
- 09Free UGC rate card and deal tracker template
- 10How do I know if a UGC brand deal pays fairly?
How to improve
How can I improve my visibility in ChatGPT?
There are no keyword tricks. ChatGPT names brands it can find, read and confirm, so the work is making your site readable, your positioning clear and your name present on the sources it uses.
- 01
Fix what stops AI from reading you
Run Rankry’s AI Readiness audit: 36 checks across crawl access, rendering, structured identity, content architecture, citation readiness and freshness, graded A to F. Content that only appears after JavaScript runs is hard for many AI crawlers to read.
- 02
Let OpenAI’s search agents in
OAI-SearchBot finds pages for ChatGPT search, and ChatGPT-User fetches a page when a question needs it. GPTBot is OpenAI’s training crawler: blocking it opts you out of training, not out of search. Being well indexed in Bing helps too.
- 03
Say clearly what you do and for whom
Give each product and use case a page that states the category, the audience and the main alternatives in plain words. Answers repeat what they can find and confirm.
- 04
Publish comparison and alternatives pages
Fair “X vs Y” and “alternatives to X” pages match the way buyers ask, and give the engine a source for exactly those prompts.
- 05
Earn mentions where the engine looks
Filter the Sources page to ChatGPT and turn on “Gap analysis only”. The review sites, forums and articles listed there cite competitors and not you.
- 06
Answer the question in the first lines
Lead with a direct answer, then the detail. Short definitions, FAQs and tables are easier to quote than long narrative.
- 07
Keep key pages current
Show a visible update date and refresh pages when prices or features change. Outdated facts on your own site tend to reappear in answers.
- 08
Work through the Action plan
Rankry orders the work in three steps: Fix your website, Win open questions, Beat your competitors. Add actions to Tasks, and draft pages for open questions in Content Studio.
Explore Rankry
Go deeper
How do I track brand mentions in ChatGPT?
Add the prompts your buyers ask, or let Rankry write them from your site, and pick ChatGPT as one of the engines. Rankry asks each prompt through OpenAI’s API with web search available, records whether your brand is named and in what position, labels the answer’s sentiment, and stores the sources and the full answer. Reports then re-run automatically every day, or every 3 or 7 days if you choose, and each run is kept in prompt history.
Is the tracking done through the API or the real ChatGPT interface?
Through OpenAI’s API. Rankry does not log in to ChatGPT or scrape the app, so the answers carry no personal memory, chat history or account settings. The model has web search available, limited to one search per answer. The result is a consistent read of how an OpenAI model with web search answers the question, asked the same way on every run, which is what you need to compare runs over time.
Why do my own ChatGPT results look different from Rankry’s?
Several reasons. The ChatGPT app uses your memory, past chats, location and the plan you are on; Rankry asks through the API with none of that. ChatGPT can also answer the same question differently from one moment to the next, and it decides each time whether to search. Use Rankry for the trend across runs and prompts, and read the stored full answer when you want to see exactly what was said.
Does ChatGPT search the web or answer from training data?
Both. ChatGPT can answer from what the model learned in training or run a web search first. Rankry gives the model web search and lets it decide, up to one search per answer. When it searches, Rankry stores the pages it consulted and the links it cited; when it does not, the answer has no sources. You can see which happened on every answer in prompt history.
Which sources does ChatGPT cite for my category?
Open the Sources page and filter it to ChatGPT. It lists the domains ChatGPT’s answers cited in your current report, with their type (You, Competitor, Review, Forum, Reference, Editorial), how often they are used and whether they are Yours, a Gap or Mixed. Turn on “Gap analysis only” to see the sites that cite competitors but not you, and open a domain to see its top URLs.
How do I see which competitors ChatGPT recommends instead of me?
Every prompt has a “Who wins this prompt” panel listing the brands ChatGPT named, their share of answers and their position when named. Rankry also picks competitors automatically from your first report, and you can add any brand that appeared in your answers with “Add to competitors”. All figures, history included, recalculate instantly from stored answers, with nothing re-run.
Does Rankry show the full ChatGPT answer?
Yes, for every new run. The prompt panel shows “Full answer”, the verbatim reply, next to the brands as ChatGPT ranked them, the reason given for each and the sources. Runs from before October 22, 2026 kept the brands, reasons and sources, but not the reply word for word.
How often does Rankry check ChatGPT?
Reports re-run automatically every day, or every 3 or 7 days if you choose, and the dashboard shows when the next scan is due. New prompts are answered right away when you add them.
Is ChatGPT visibility different from Claude or Perplexity visibility?
Usually, yes. Each engine uses a different model and different search, so the same prompt can name you in ChatGPT and skip you in Claude or Perplexity. Rankry keeps every engine separate: the Model comparison table shows your visibility, position and sentiment for each engine side by side, and Diversity counts how many engines name you at all.
How can I improve my ChatGPT recommendations?
Make sure OAI-SearchBot and ChatGPT-User can read your key pages, and that your site is indexed in Bing as well as Google. Publish clear category, comparison and alternatives pages, and earn mentions on the review sites and articles ChatGPT cites. Rankry’s Action plan orders this work in three steps (Fix your website, Win open questions, Beat your competitors), and Content Studio drafts the pages.
See how ChatGPT ranks your brand
Start a 7-day free trial with no card. Rankry writes your first prompts from your site and runs your first report right away.
7-day Base trial · No card needed