An AI assistant answers a buyer · illustrative example
“Which inventory platform should an independent grocer use?”
Competitor A is the one most often recommended1, mainly for how quickly it sets up against existing tills2. Buyers comparing options also look at3
Asked 4 times · 3 assistantsSources cited: 2
What AI tells your buyers, and who it credits
Assistants answer a buying question from a few sources they cite. We check whether yours is one of them, and what to fix if it isn’t.
One page, in real time · no signup, no card
An AI assistant answers a buyer · illustrative example
“Which inventory platform should an independent grocer use?”
Competitor A is the one most often recommended1, mainly for how quickly it sets up against existing tills2. Buyers comparing options also look at3
Asked 4 times · 3 assistantsSources cited: 2
01. The masthead
Assistants we measure · ChatGPT · Perplexity · Gemini · Azure OpenAI · Claude · Gemini with Search grounding
Instrument no. 01. Readiness audit
> awaiting input
Spec. A. What a finding looks like
Two of the eight things we check
Criticalthe page has no main heading
AI engines read a page by its headings. With none, they can't tell what the page is about, so they don't quote it.
Warningnothing on the page answers a question directly
Assistants answer by lifting a passage that already answers the question. If no passage on your site answers one, there is nothing to lift.
Illustrative. Your findings come from your own crawl.
02. Exhibit A: held for examination
“What is the best inventory platform for an independent grocery business?” · illustrative run 4/4
For small grocery teams, Competitor A is a leading choice because of its quick setup and retail integrations. It is frequently recommended in comparison pages and review roundups, and its documentation covers common POS setups. Buyers weighing setup effort also cite yourbrand.com, whose comparison page sets the two side by side.
Omitted
Now cited
the story annotates the exhibit as you read · illustrative view
02·a. What shaped it, and what it left out
02·b. What it takes to be in it
“Best way to measure whether AI assistants cite your brand”
02·c. What the record shows after
03. The record · one question, 238 stored answers
Apr 12
Competitor B
Baseline captured · review coverage
Apr 26
Competitor B
New source entered · comparison guide
May 10
Competitor A
Recommendation changed · integration guide
Today
Competitor A
Change verified · 4/4 runs, 3 engines
Illustrative shape of one tracked question · your grid is built from your own stored runs
04. The timeline · who was cited, fortnight by fortnight
Illustrative · fourteen fortnights of one buyer question. Your timeline is built from your own stored runs.
05. The archive · free, and no account
A.2. Free tool · live
No signup
Can AI crawlers read you?
Reads your robots.txt and llms.txt and tells you which assistants are blocked before they ever get to your content.
A.3. Free tool · live
No signup
What a machine reads first
Before an assistant reads a word of your page it reads its title and description. This shows you both, exactly as they arrive.
B.1. Lost question · worked example
Illustrative
“Best inventory software for independent grocery stores”
A.4. Free tool
In-browser
FAQ schema generator
Turns plain questions and answers into paste-ready FAQPage JSON-LD, entirely in your browser. Nothing is uploaded. That is also why this one is a link rather than a panel.
C.1. Rate card
Per answer check
06. Where to start
One page crawled in real time · no AI prompts run · no account, no card
An AI assistant answers a buyer · illustrative example
“Which inventory platform should an independent grocer use?”
Competitor A is the one most often recommended1, mainly for how quickly it sets up against existing tills2. Buyers comparing options also look at3
Asked 4 times · 3 assistantsSources cited: 2
07. Fair questions, answered plainly
CiteTrail measures how AI assistants answer the buying questions in your category. It records whether you get mentioned, which competitors get recommended instead, which sources the answer leaned on, and what to change. The response itself is kept, so every finding can be opened and read.
How often, and how favourably, your brand appears when an assistant answers a question a buyer would actually ask. It is not a keyword ranking. There is no position one and no single answer everyone sees, which is why it has to be sampled repeatedly rather than looked up.
6 surfaces: ChatGPT via the OpenAI API, Perplexity via its sonar API, Gemini via the Gemini API, Claude via the Anthropic Messages API, Azure OpenAI, and Gemini with Google Search grounding. These are provider APIs, the models behind the assistants, not scrapes of the consumer apps. We do not read Google AI Overviews, and we say so rather than letting the logo imply it.
Write them, import them, or start from suggestions generated for your brand and category. Every prompt is classified by intent and funnel stage and scored out of 100, because a comparison question and a definition question are not worth the same effort.
Ask a model the same question twice and you can get two different answers naming two different brands. A single response is an anecdote. CiteTrail runs each prompt repeatedly, across engines, over time, and derives a confidence label from how much it sampled and how consistent the results were. Fewer than two runs is always labelled Low, whatever the answers said. Every finding shows its sample so you can argue with it.
No. Nobody can, and you should be sceptical of anyone who says otherwise. Providers change models, prompts and grounding without notice. We show what is happening, what changed, and what evidence supports an action. The outcome is not ours to promise.
From the answers themselves — the brands assistants actually name in response to your prompts — combined with the competitor set you define. You find rivals you did not know you had, not just the ones already on your list.
Yes. Each client gets an isolated workspace with its own brands, prompts and competitors. Workspaces are the plan lever: Professional includes 2, Growth includes 5, Enterprise is unlimited. White-label reporting starts at Growth.
Yes. Both connect through Google sign-in, both are read-only, and both refresh daily. Search Console data lags two to three days on Google's side, and branded-query tracking needs your brand aliases set first. Available on Professional and above. The integrations page lists exactly what is stored.
One credit is not one check. A check is one prompt, one engine, one region, and engines are priced by what they cost to run: Gemini 1, ChatGPT 1, Claude 3, Azure OpenAI 7, Perplexity 8, Gemini with Search grounding 14. Monthly allowances are 8,000 on Essentials, 25,000 on Professional and 75,000 on Growth. Plan credits do not roll over. Purchased top-up credits never expire and are only spent once the monthly allowance is gone. The console shows the credit cost of a run before you commit to it.
Yes, with no signup and no card. It fetches your homepage and scores whether it is structured to be understood and cited by an assistant, then shows the buyer questions we would track for you. It is a readiness check on your site. It does not query the assistants. That happens inside a workspace, where runs are billed to your plan.
No. You get an email digest, an in-app brief, on-demand HTML reports and CSV exports. Nothing in the product delivers a report to a recipient list on a schedule, so we do not sell it as a feature.
Authentication runs through Zitadel OIDC with httpOnly session cookies, not tokens in browser storage. Workspaces and brands are isolated and access is role-based. We store the prompts and responses we measured so findings stay auditable. Prompts go to the model providers listed on the engines page in order to run them, and are not used to train models. We hold no security certifications today and do not claim any.
†. Footnote register