Marketing and SEO teams that need to know when ChatGPT names or links their brand
Best tools for tracking ChatGPT citations
Six tools ranked on how cheaply they cover ChatGPT, whether they report the sources an answer cited, and whether they publish sampling depth and response volume.
Karl-Gustav Kallasmaa, Founder & CEOLast updated How this list was made
Six tools were shortlisted because each publishes a product or pricing page describing ChatGPT answer tracking. Each vendor's own page was fetched on 3 September 2026 and read for four things: the cheapest plan on which ChatGPT is covered, whether the product reports the sources an answer cited rather than only a score, whether the vendor states how many times each prompt is sampled, and whether a monthly volume of AI responses is published. No third-party roundup was used and no vendor was contacted. Where a page does not answer a question, the cell says so. Attensira is one of the six and is judged on the same four criteria, losing two of them.
The criteria
Declared before the ranking, with how each one was measured and who wins it.
| Criterion | How it was measured | Best on this |
|---|---|---|
| Cheapest plan on which ChatGPT is coveredChatGPT is the assistant most teams are asked about first, and several vendors put it on the entry tier while others bundle it with engines you do not need yet, which changes the real cost of answering that one question. | Read from the vendor's own pricing page: the cheapest plan, free tiers included, on which ChatGPT appears as covered without a paid add-on. | Athena |
| Reports the sources an answer citedA visibility score tells you that you are absent; the list of pages the assistant cited instead tells you where to go next, and it is the difference between a metric and a task list. | Read from the vendor's own product or pricing page: whether citation or source analysis is described as a named capability rather than implied by a screenshot. | Scrunch AI |
| Sampling depth stated per planChatGPT answers the same prompt differently on different days, so a citation counted once is a coin flip and a citation counted three times is evidence, and only a vendor that publishes the number lets you tell which you bought. | Read from the vendor's own pricing page: whether it states, as a number, how many times each tracked prompt is asked per reading. | Attensira |
| Monthly AI response volume publishedPrompt count is a plan limit; response count is the amount of evidence you actually get, and two tools with the same prompt allowance can differ by an order of magnitude in how many answers they collect. | Read from the vendor's own pricing page: whether a monthly number of AI responses, credits or answers is published per plan. | Attensira |
The tools
- 1
Athena
A credit-metered AI visibility tool whose free Essential tier names ChatGPT among the five assistants listed against that plan, with a wider ten-model list on the paid Starter tier. Its pricing page defines the unit precisely, as one credit per AI response.
- Best for:
- Anyone who wants to see how ChatGPT currently answers questions about their brand without spending anything first, and can work from a fixed number of responses.
- Where it falls short:
- The free allocation is a sample rather than a monitor, the step to the next published plan is large, and the pricing page does not state how many responses go into one tracked prompt.
- Cheapest plan on which ChatGPT is covered: Free Essential tier; ChatGPT among the five assistants named against it source
- Reports the sources an answer cited: Not stated on the pricing page source
- Sampling depth stated per plan: One credit is one AI response; repeats per prompt not stated source
- Monthly AI response volume published: 300 credits free, 3,600 per month on Starter source
Athena · checked on
- 2
Otterly.AI
An AI search monitoring tool that includes ChatGPT on every plan, starting at a low published entry price, and states its prompt allowance per tier. Three further assistants are sold as separately priced add-ons.
- Best for:
- A small team that wants continuous ChatGPT tracking at the lowest published paid price and does not need Claude or Gemini alongside it.
- Where it falls short:
- The entry plan tracks fifteen prompts, which is one topic rather than a market, and neither sampling depth nor a monthly response volume is published.
- Cheapest plan on which ChatGPT is covered: ChatGPT included on Lite at $29/month source
- Reports the sources an answer cited: Not stated on the pricing page source
- Sampling depth stated per plan: Not stated source
- Monthly AI response volume published: 15 prompts on Lite; response volume not published source
Otterly.AI · checked on
- 3
Profound
An AI search visibility platform whose entry tier is explicitly and only ChatGPT tracking, with fifty prompts and three seats, rising to three answer engines on the middle tier. Few vendors are this direct about what an entry plan does not include.
- Best for:
- A team whose question is specifically about ChatGPT and who would rather buy a plan that says so than pay for engines it will not look at.
- Where it falls short:
- The step from ChatGPT-only to three engines is a four-fold price increase, and the pricing page publishes neither a sampling depth nor a monthly response volume.
- Cheapest plan on which ChatGPT is covered: ChatGPT tracking only on Starter at $99/month source
- Reports the sources an answer cited: Not stated on the pricing page source
- Sampling depth stated per plan: Not stated source
- Monthly AI response volume published: 50 prompts tracked; response volume not published source
Profound · checked on
- 4
Semrush
A general SEO suite that includes AI visibility in every plan, with a daily prompt allowance rising by tier. For a team already on the suite, ChatGPT tracking arrives without a second subscription or a second tool to learn.
- Best for:
- A team that already pays for Semrush and wants ChatGPT answer tracking beside its existing organic search reporting.
- Where it falls short:
- The AI reporting is one module in a broad suite, extra users are a paid add-on, and neither sampling depth nor a monthly response volume is published.
- Cheapest plan on which ChatGPT is covered: AI visibility in every plan, cheapest at $117.33/month annual source
- Reports the sources an answer cited: Not stated on the pricing page source
- Sampling depth stated per plan: Not stated source
- Monthly AI response volume published: 50 to 200 prompts per day; response volume not published source
Semrush · checked on
- 5
Scrunch AI
A platform whose product description names citation analysis as one of three modules, describing it as uncovering the sources AI relies on, alongside prompt analytics covering citations, competitors and rankings across ChatGPT and four other assistants.
- Best for:
- A team whose next action depends on knowing which third-party pages ChatGPT cites instead of theirs, rather than on the score itself.
- Where it falls short:
- It is the most expensive published entry plan among these six, its self-serve trial is capped at a single brand, and neither sampling depth nor a monthly response volume is published.
- Cheapest plan on which ChatGPT is covered: ChatGPT covered; cheapest published plan $250/month annual source
- Reports the sources an answer cited: Citation analysis named as a product module source
- Sampling depth stated per plan: Not stated source
- Monthly AI response volume published: Prompt allowance stated; response volume not published source
Scrunch AI · checked on
- 6
Attensira
That is usAn AI visibility platform that publishes both a sampling depth and a monthly response budget against every plan, and whose product page describes showing which pages the models quote. ChatGPT is covered on the entry plan alongside three other models.
- Best for:
- A team that has to defend a ChatGPT citation trend to somebody sceptical and needs to be able to say how many answers the trend is built from.
- Where it falls short:
- There is no free tier and no trial stated, so there is no way to see it against your own brand without buying a month, and the entry plan samples each prompt once, which is the same noise floor as the cheaper tools here.
- Cheapest plan on which ChatGPT is covered: ChatGPT on Starter at $129/month, with three other models source
- Reports the sources an answer cited: Shows which pages the models quote source
- Sampling depth stated per plan: n=1 on Starter, n=3 on Growth and Business source
- Monthly AI response volume published: 6,000 responses per month on Starter, 67,500 on Growth source
Attensira · checked on
Which tool tracks ChatGPT citations best?
Athena is the cheapest, because its free tier covers ChatGPT with an included allocation of 300 responses. Otterly.AI is the cheapest paid option that includes ChatGPT continuously, at $29 a month. Profound is the only vendor that sells a plan explicitly scoped to ChatGPT and prices it as such. Scrunch names citation analysis as a product module and describes it as uncovering the sources AI relies on. Attensira publishes the sampling depth and the monthly response volume behind its numbers, and wins those two criteria while losing the other two.
Mention, citation, and why the distinction decides the tool
Before comparing products, settle what you are counting, because the word "citation" is doing three jobs in this category.
A mention is your brand appearing in the text of an answer. It happens whether or not the assistant browsed the web, and it is the thing most buyers actually care about, because a named brand enters a shortlist.
A citation is a linked source attached to an answer, which only happens when the assistant retrieved something. ChatGPT links sources when it browses and does not when it answers from what the model already contains, so a tool that counts only links will systematically under-report how often you are named.
A cited competitor source is the third and most actionable: the third-party page an assistant used instead of yours. For most brands the practical finding is not that they scored low but that the same handful of roundups and review pages keep supplying the answer, and none of them list you.
That third category is why the citation-source criterion is on this page at all. A visibility score tells you that you are absent. The list of pages cited instead tells you what to do on Monday.
Reading the ranking
Athena wins the cheapest-coverage criterion on a technicality that is also completely real: its Essential tier is free, ChatGPT is on the monitored list, and the free allocation is defined precisely as 300 AI responses. Nothing else here gives you ChatGPT answer data for nothing. The limits are equally real — 300 responses is a snapshot rather than a trend, and the pricing page does not say how many responses go into one tracked prompt, so you cannot tell how much of a sample each prompt got.
Otterly.AI is the cheapest continuous option, including ChatGPT on every plan from $29 a month. Its pricing page is the plainest on this list about what is included and what is an add-on. Fifteen prompts on the entry tier is the constraint, and it is a real one: that is a topic, not a market.
Profound is the only vendor here that treats ChatGPT as a product boundary rather than a feature, writing "ChatGPT tracking only" against its $99 Starter tier. If your question is genuinely about ChatGPT alone, that clarity is worth something, and fifty prompts with three seats is a workable configuration. The jump to three answer engines is a four-fold price increase, which is the price of that clarity.
Scrunch wins the citation criterion. It names citation analysis as one of three product modules and describes it as uncovering the sources AI relies on, with prompt analytics covering trends, citations, competitors and rankings. That is the capability that turns a score into a plan, and Scrunch is the entry here that puts it front and centre. It is also the most expensive published entry plan of the six, and its trial is capped at one brand.
Semrush wins nothing here and remains the sensible default for a team already on the suite, because AI visibility is inside every plan rather than sold beside it. What it does not do is tell you anything about how its numbers were produced, which on this particular question matters more than usual.
Attensira wins sampling depth and published response volume. We state n=1 on Starter and n=3 above it, with 6,000 monthly responses on Starter and 67,500 on Growth, because a ChatGPT citation trend is only interpretable if you know how many answers it is built from. We also show which pages the models quote where you are not present, which is the same actionable finding Scrunch leads with.
We lose two criteria and they are not small ones. We are not the cheapest way in by any measure — Athena's free tier and Otterly's $29 plan both beat our $129 Starter, and we have no free tier and no stated trial, so evaluating us against your own brand means buying a month. And on citation analysis specifically, Scrunch has built more of the surface than we have. If your budget is zero, start with Athena. If your only question is ChatGPT and you want it cheap and continuous, start with Otterly.
Why the numbers disagree, and which disagreement matters
Run two of these tools on the same brand in the same week and they will report different ChatGPT visibility. That is expected, and the reasons divide into ones worth worrying about and ones that are not.
Different prompt sets. This is the largest source of disagreement and the least interesting, because it is entirely under your control. Two tools asking two different sets of forty questions are measuring two different things. If you paste the same prompts into both, most of the gap closes. Any comparison run on vendor-generated prompt lists is comparing the generators, not the products.
Different sampling depth. A tool that asks once and a tool that asks three times will disagree on any prompt whose answer is unstable, which in practice is most of the interesting ones. The single-sample tool is not wrong; it is reporting a noisier estimate of the same underlying quantity. This gap does not close by pasting in the same prompts, and it is the reason the sampling criterion is on this page.
Different definitions of a hit. A brand named in prose, a brand linked in a source list, and a brand named inside a page the assistant cited are three events. Vendors count them differently and rarely say which. This gap is the one worth investigating before you buy, because it changes what the number means rather than how precise it is.
Different retrieval behaviour. ChatGPT browses for some questions and not others, and whether it browsed changes whether links exist to count at all. Two tools asking at different times of day can catch different behaviour. There is nothing to fix here; it is an argument for repeat sampling rather than for switching vendors.
The practical implication is that a single visibility percentage from any of these products, quoted without its prompt set and its sampling depth, is not a number anyone should act on. What the good tools give you is not a truer percentage — it is a percentage that comes with enough context to be compared against itself next month.
What to do with the answer once you have it
Tracking is not the work. The finding that matters from any of these tools is almost never "our score is low" — it is the list of pages the assistant used instead of yours, and that list is usually short and usually the same across questions.
In most categories it is a handful of third-party roundups, review sites and comparison pages, none of which you control. The response is not to rewrite your own product page harder. It is to become present, accurately, in the sources that are already supplying the answer: correct listings on the review sites that get cited, participation in the communities that get quoted, and comparison content of your own that is fair enough to be worth citing.
That is slower than changing a title tag and it is the only intervention that reliably moves an assistant's answer. A tool that surfaces those sources turns a monitoring subscription into a work queue. A tool that reports only a percentage leaves you optimising something you cannot see.
How to check a vendor's ChatGPT number in ten minutes
Ask three questions in a trial or a demo, and treat a vague answer as an answer.
"Does this count mentions, linked citations, or both, and can I see them separately?" If the product cannot separate them, its number will move for reasons you cannot explain, because ChatGPT's browsing behaviour varies by question and by day.
"How many times was each prompt asked to produce this figure?" One is a legitimate answer. Refusing to say is not. This single number sets how much of a week-on-week change is real, and it is the cheapest thing a vendor could publish and mostly does not.
"Show me the sources behind one answer where I am absent." If the product can put the three pages ChatGPT used in front of you within a click, it will generate work you can do. If it can only show a percentage, you have bought a chart.
"What happens when I am absent from an answer?" The tools that earn their subscription turn an absence into a list of pages to go and be present on. The ones that do not leave you with a chart and a meeting. Ask to see the workflow, not the dashboard.
Then run your own control. Ask five of your prompts yourself, twice, on separate days, and compare what you saw against what the tool reported for the same period. Any tool worth its subscription will survive that comparison. Every price and capability above was read on 3 September 2026, so re-check the vendor pages before acting on them.
Questions people ask
- What is the cheapest way to track ChatGPT citations?
- Athena's free Essential tier, which includes a fixed credit allocation covering 300 AI responses at no cost. Among paid plans, Otterly.AI includes ChatGPT on its $29 Lite plan and Profound sells a ChatGPT-only Starter plan at $99. All three were read from the vendors' own pricing pages on 3 September 2026.
- Does ChatGPT actually cite sources?
- It links sources when it browses the web to answer, and does not when it answers from what the model already contains. That is why brand mentions and linked citations are different measurements, and why a tool that only counts links will under-report how often you are named. Ask any vendor which of the two their number counts.
- Why do two tools report different ChatGPT visibility for the same brand?
- Usually because they asked different questions, asked a different number of times, or count a mention differently from a linked citation. The prompt set and the sampling depth explain most of the gap. Only one of the six pricing pages read for this list publishes the sampling number at all.
- How many prompts do I need to track ChatGPT usefully?
- Fewer than most plans sell, chosen more carefully than most vendors suggest. Fifteen to fifty questions taken from real sales calls, support tickets and search terms will produce a clearer picture than several hundred generated from your own domain, because generated prompts inherit your vocabulary and flatter the result.
- When were these plans checked?
- Every price, allowance and capability on this page was read from the named vendor page on 3 September 2026. Tiers in this category are re-cut within a quarter, so treat each cell as a prompt to open the vendor's page rather than a current fact.
Sources
Every cell above, with the page it came from and the date that page was read.
Athena's pricing page lists Essential as free with a $25 free credit covering 300 credits and Starter at $295/month covering 3,600 credits, and defines the unit as one credit equals one AI response.
athenahq.ai · retrieved · changes often, check the source
“1 credit = 1 AI response”
Athena's pricing page names ChatGPT, Perplexity, Google AI Overviews, Gemini and Copilot against the free Essential plan and states 10 models on Starter, adding Claude, Grok, DeepSeek and Meta AI, and does not describe citation source analysis on that page.
athenahq.ai · retrieved · changes often, check the source
Otterly.AI's pricing page includes ChatGPT, Google AI Overviews, Perplexity and Microsoft Copilot on all plans starting at $29/month with 15 search prompts on Lite, and sells Claude, Google AI Mode and Gemini as add-ons.
otterly.ai · retrieved · changes often, check the source
“ChatGPT, Google AI Overviews, Perplexity, MS Copilot”
Otterly.AI's pricing page states a search prompt allowance per plan but does not state how many times each prompt is asked per reading or a monthly response volume.
otterly.ai · retrieved · changes often, check the source
Profound's pricing page lists Starter at $99/month with 50 prompts tracked and ChatGPT tracking only, Growth at $399/month with 3 answer engines, and does not publish a monthly response volume or citation analysis feature.
tryprofound.com · retrieved · changes often, check the source
“ChatGPT tracking only”
Profound's pricing page states prompt counts and answer-engine counts per plan but does not state how many times each prompt is sampled per reading.
tryprofound.com · retrieved · changes often, check the source
Semrush's pricing page presents AI visibility as part of every plan, with the daily prompt allowance rising from 50 on Starter to 200 on Advanced, the cheapest plan listed at $117.33/mo billed annually, and no citation analysis or response volume stated.
semrush.com · retrieved · changes often, check the source
“$117.33/mo billed annually”
Semrush's pricing page states a daily prompt allowance per plan but does not state how many times each prompt is asked per reading.
semrush.com · retrieved · changes often, check the source
Scrunch's site describes citation analysis to uncover the sources AI relies on as part of its Insights module, and prompt analytics with trends, citations, competitors and rankings as part of Monitoring, across ChatGPT, Perplexity, Claude, Gemini and Copilot.
scrunch.com · retrieved · changes often, check the source
“Citation analysis to uncover the sources AI relies on”
Scrunch's pricing page lists Starter at $250 per month billed annually or $300 month-to-month and Growth at $417 billed annually, without publishing a monthly response volume.
scrunch.com · retrieved · changes often, check the source
“$250 per month (billed annually) or $300 month-to-month”
Scrunch's pricing page states a custom prompt allowance but does not state how many times each prompt is asked per reading.
scrunch.com · retrieved · changes often, check the source
Attensira's pricing page lists 4 models on Starter at $129/month (ChatGPT, Perplexity, Claude and Google) and 5 models on Growth and Business, which add Gemini.
attensira.com · retrieved · changes often, check the source
Attensira's pricing page states 50 prompts and 6,000 responses per month on Starter sampled once per reading, and 150 prompts and 67,500 responses on Growth sampled three times per reading.
attensira.com · retrieved · changes often, check the source
“Sampled 3x per reading (n=3)”
Attensira's home page describes showing the pages the models quote where the brand is not present, naming the listicles the models quote most.
attensira.com · retrieved · changes often, check the source
“pages the models quote”