ChatGPT Visibility Checker: How to Test Your Brand for Free
A ChatGPT visibility checker tells you whether ChatGPT mentions, recommends or cites your brand when people ask about your category. Free tools from HubSpot, Ahrefs and Mangools give you a quick snapshot, but each measures something slightly different from what real users see. This guide compares them honestly and gives you a manual prompt protocol with a tracking sheet you can rerun every month.
What a ChatGPT visibility check actually measures
There is no official "ChatGPT ranking." OpenAI does not publish a position report for brands, so every checker builds its own proxy. Before trusting any score, know which of these four things it is measuring:
- Mention: does the answer name your brand at all?
- Recommendation: is your brand suggested as an option, and where in the list?
- Citation: does the answer link to a page on your site as a source?
- Accuracy: is what ChatGPT says about you correct (pricing, features, location, what you do)?
The split matters because ChatGPT answers in two modes. Without web search, it answers from what the model learned in training. With search, it pulls fresh pages and shows sources. OpenAI's crawler documentation says OAI-SearchBot is the bot that surfaces websites in ChatGPT's search features, and that sites opted out of it will not be shown in ChatGPT search answers. So a brand can be well known to the model and still never get cited, or the other way around.
Free ChatGPT visibility checkers compared
We opened each tool's page on October 17, 2026. Features and limits change, so check the page before relying on them.
| Tool | What it queries | What you get free | Watch out for |
|---|---|---|---|
| HubSpot AI Search Grader | ChatGPT (an API model named in its FAQ), Perplexity, Gemini | A score out of 100 across sentiment, presence quality, brand recognition, share of voice and competition | One-time snapshot based on what models know; a form stands between you and the full report |
| Ahrefs Free AI Visibility Checker | ChatGPT, Gemini, Perplexity, Copilot, Google AI Overviews and AI Mode, using prompts derived from real searches | Total mentions, mentions by platform, top topics, top cited pages; no signup | Preview only (top items per section), no history, no competitor comparison, no custom prompts |
| Mangools AI Search Grader | Several models, including a small GPT model for everyone and more models after a free login | Visibility, average ranking in answers, a 0–100 score, competitor view | Uses lightweight API models, not the ChatGPT app with search; usage limits are not stated |
| Bing Webmaster Tools AI Performance | Microsoft Copilot, AI summaries in Bing, select partner integrations | Citation counts, cited pages, grounding queries for your own verified site | Not a ChatGPT report; Bing says the data is a sample of citation activity |
| Manual protocol (below) | The real ChatGPT app, with and without search | Exactly what a user sees for prompts you choose | Takes about an hour a month; answers vary run to run |
The honest takeaway: the free graders are good for a first look and for spotting obvious gaps, like a competitor named everywhere while you are named nowhere. They are weak for tracking change over time, because the prompts, models and scoring are theirs, not yours. A 62 that becomes a 58 next month tells you little about why.
Why one-click scores disagree with what you see
- Different model. Some graders call a small, cheap API model. Users in the ChatGPT app often get a larger model with search turned on. The answers can differ a lot.
- No search vs search. A model-memory check says nothing about whether your pages get cited when ChatGPT searches the web.
- Different prompts. Tools generate prompts from your category description. If they phrase your niche differently from your customers, the result reflects their wording.
- Randomness. The same prompt can return different lists on different runs. A single run is an anecdote, not a measurement.
- Composite scores. Weighted 0–100 scores mix sentiment, presence and competition using the vendor's weights. Useful as a headline, but you cannot act on a number you cannot decompose.
The manual ChatGPT visibility protocol
This is the method we recommend when you want a result you can trust and repeat. It needs a ChatGPT account, a spreadsheet and about an hour.
Step 1. Write 20 prompts in four buckets
| Bucket | What it tests | Example prompts (illustrative, for a small bakery supply shop) |
|---|---|---|
| Category (6) | Are you in the shortlist when people don't know you? | "best online shop for cake decorating supplies in the US" |
| Problem (6) | Are you suggested as a fix for a job to be done? | "where can I buy food-safe gel colors in small quantities" |
| Comparison (4) | How you are framed next to competitors | "[your brand] vs [competitor] for home bakers" |
| Brand (4) | Is what ChatGPT says about you correct? | "what is [your brand]", "is [your brand] legit", "[your brand] shipping costs" |
Use the words your customers use. Your Search Console queries and support emails are the best source for phrasing.
Step 2. Remove personalization
Run prompts in a Temporary Chat or a logged-out browser window, so memory and past conversations about your brand don't color the answer. Never run them from the account you use to work on your own site.
Step 3. Run each prompt twice: without and with search
First ask normally and note whether the answer used web sources. Then ask the same prompt with search explicitly turned on. The first run shows model knowledge; the second shows citation visibility.
Step 4. Repeat three times
Run each prompt three times in fresh chats. Record a mention only if it appears in at least two of three runs. This filters out lucky one-offs.
Step 5. Log the same columns every time
| Column | What to write |
|---|---|
| Date, prompt, bucket, mode | 2026-10-26, prompt text, Category, Search on |
| Mentioned (runs out of 3) | 0–3 |
| Position in list | 1, 2, 3… or "not listed" |
| Cited URL | The page from your site, if any |
| Competitors named | Comma-separated |
| Accuracy issue | Wrong price, outdated feature, wrong city… |
| Sources cited instead of you | Domains ChatGPT trusted for this answer |
Step 6. Calculate three numbers
- Mention rate = prompts where you appear in 2+ runs ÷ total prompts.
- Citation rate = search-on prompts where your URL is cited ÷ search-on prompts.
- Accuracy issues = count of wrong facts in the Brand bucket.
Worked example (illustrative, not real data): you run 20 prompts. You appear in 2+ runs for 5 of them, so your mention rate is 25%. With search on, your site is cited for 3 of 20 prompts, a 15% citation rate. In the Brand bucket ChatGPT gets your shipping policy wrong twice. That tells you exactly where to work: the shipping page, and the category pages that compete for those 15 prompts where you are missing.
Reading the results: what to fix first
| What you see | Likely cause | First fix |
|---|---|---|
| Never cited with search on, even for brand prompts | OAI-SearchBot blocked, or pages not indexed by the search providers ChatGPT uses | Check robots.txt with the AI crawler checker; follow this indexing guide |
| Known without search, but facts are outdated | Old information about you dominates the web | Update your About, pricing and FAQ pages; correct old directory listings |
| Competitors listed, you are not | Few independent mentions of you in your category | Get mentioned in roundups, communities and partner articles; see how to get cited by ChatGPT |
| Cited, but for the wrong page | Your best answer is buried or split across pages | Make one clear page per question with the answer near the top |
| Other domains cited for your own brand prompts | Third-party reviews explain you better than you do | Add a plain, factual page that answers the same questions |
Measure the outcome, not just the visibility
Visibility is a leading indicator. The result you actually want is visitors. ChatGPT referrals show up in analytics as their own source, so set up a channel for AI assistants and watch it monthly. The step-by-step setup is in how to track AI traffic in GA4. If the mention rate rises but AI referrals stay flat, your mentions are probably happening in answers with no link, which is still useful for brand recall but not for traffic.
For Microsoft's side of AI search, Bing Webmaster Tools now has a free AI Performance report with citation counts and the "grounding queries" behind them for your verified site. In June 2026 Bing added Intents, Topics, Citation Share and period comparison in preview. It covers Copilot and Bing, not ChatGPT, but it is the only free first-party citation data available right now, and it is worth setting up next to your manual sheet.
A monthly routine that takes one hour
- Rerun the same 20 prompts, same modes, same three runs.
- Update mention rate, citation rate and accuracy issues in a new row.
- Check the Bing AI Performance report and your AI referral channel.
- Pick one fix from the table above and ship it before next month.
- Run one free grader as a sanity check, but don't chase its score.
Keep the prompt list stable. If you change prompts every month, you are measuring your prompt-writing, not your visibility. Add new prompts in a separate tab and fold them in each quarter.
What to do next
Start with the technical floor: if OAI-SearchBot can't reach your pages, no amount of content will get you cited in ChatGPT search. Run your domain through the free AI crawler checker to see which AI bots your robots.txt allows, in under a minute. Then build your 20-prompt sheet and run the first baseline today.
Independent mentions are the slowest part to fix. Satelo is a small public-beta network where similar sites publish partner articles about each other in a circle, with links marked rel="sponsored". If you want to see where your site stands first, check its Quality Score: you get the number and the mostly-open data behind it in 30 seconds.
FAQ
Is there a free ChatGPT visibility checker?
Yes. HubSpot AI Search Grader, the Ahrefs Free AI Visibility Checker and Mangools AI Search Grader all offer free checks. They are snapshots with their own prompts and scoring, so use them for a first look and a manual prompt protocol for tracking.
How do I check if ChatGPT mentions my brand?
Write 20 prompts your customers would ask, run each in a Temporary Chat three times with and without web search, and log whether you are mentioned, your position, and whether your site is cited. Repeat monthly with the same prompts.
Why does ChatGPT not show my website as a source?
Common causes are robots.txt blocking OAI-SearchBot, pages that search engines have not indexed, or simply stronger pages from other sites answering the same question. OpenAI says sites opted out of OAI-SearchBot are not shown in ChatGPT search answers.
Are AI visibility scores accurate?
They are consistent within one tool but not comparable across tools. Many use small API models without web search, their own prompts, and weighted formulas. Treat a score as a hint and look at which prompts and competitors drive it.
Can I track ChatGPT traffic in Google Analytics?
Yes. Visits from ChatGPT links appear as a referral source, and you can group AI assistants into a custom channel in GA4. That shows whether visibility is turning into visitors.
Does Bing Webmaster Tools show ChatGPT citations?
No. Its AI Performance report covers Microsoft Copilot, AI summaries in Bing and select partner integrations. It is still useful as free first-party citation data for Microsoft's AI surfaces.
Read next
Check your site's Quality Score
Free, no sign-up: a 0–100 score from mostly open data. Then add your site to Satelo and get partner articles from sites of your level.
Enter a domain like example.com
Add my site