How your brand stands in AI answers
One score, the decisions under it, and the evidence on demand. Everything below comes from your own audit and monitoring runs.
—
Labelled and auditable: every component links to the evidence behind it.
of the site for AI
Citability layers
Four-layer model. The weakest layer is usually "Quotable": content an engine can quote verbatim.
What now
the moves the data points toAction Center
done → measured, with the noise bandObservations, not proofs. We show what changed after your changes, together with how much the engines move on their own. We cannot confirm that your changes were the reason. This list is about pages the engines already fetch; if your page is not among the sources for a question, start with third-party pages.
I changed something else
Record a change we did not propose. We take a snapshot of the page and compare the next runs against it.
The legend loads from live data (noise band per engine, NOISE-1.0).
Visibility trend
Score per engine
you vs categoryPosition in your industry
among companies audited by CitePulseThe position indicator loads live.
What AI really says about your brand
Evidence, not only a score: verbatim quotes, hits per question, the matrix, hallucinations and the buying funnel.
Verbatim quotes
fragments of AI answers about the brandBrand sentiment
how AI speaks about you in its answersBuying questions
the questions we ask the enginesEach question is checked against every engine; "cited" means at least one engine cited you for it.
Brand risk
hallucinationsStatements about you that contradict your own site. Each one is a signal for the fact registry.
Your buying questions
the 5 grounded audit questions plus your ownEach own question is asked 3 times per engine in every run, like the audit series. A question added today shows up in the matrix and "seen in k of n" from the next run; n counts only the runs that asked it.
Question × engine matrix
| Question | Perp | Gem | GPT | Claude | Copilot | Rate |
|---|---|---|---|---|---|---|
| Loads from live data. | ||||||
Buying funnel
citations per stageWhere the weakest stage is the decision stage, you lose exactly where people buy.
Who is the source when AI answers about your category
Not how many citations, but which sources produce the visibility and where a competitor wins.
Citation share
Your share of all the sources the engines cited for your buying questions. Not "share of voice" in the answer prose, which we do not measure.
Source gaps
where you are absent and it mattersCompetitors named by the audit
rivals AI names instead of you on buying questions| Competitor | Named | Sample buying questions |
|---|---|---|
| Loads from live data. | ||
"Named" = on how many of the tested questions AI mentioned the competitor. The same rivals your report names.
Where a competitor wins
competitor domains AI cites as a source| Competitor domain | Citations | Engines |
|---|---|---|
| Loads from live data. | ||
Category share
you vs cited competitorsYour pages AI cites
per page: how many questions, which engines, in how many of n runs| Your page | Questions | Engines | Seen |
|---|---|---|---|
| Loads from live data. | |||
One run is a snapshot, not a state: cited sources change from day to day. Monitoring repeats the same questions and turns this into "seen in k of n".
Where to play
door A: your own page can win · door B: lists and directories win| Question | Door | What it means |
|---|
DOORS-1.0. The verdict is per phrasing; draw conclusions from the whole set. We never say "you are not on X", we show which pages win.
Is the site ready for AI to read and cite it
Readiness, extractability, agent readiness and brand defence. Fixes live in the Auto-Fix tab.
Site-Readiness
Robots access for AI bots, structured data, server-rendered content, FAQ and product data. The number is the same one your report shows.
Extractability signals
can AI pull out a fragmentAgent readiness
can a buying agent copeBrand defence
resistance to prompt injectionHidden instructions aimed at AI assistants inside your pages would put the brand at risk.
Ready fixes, one click
Each fix is a ready file or snippet. Download it, put it in place, and record it with "Done" in the Action Center.
Draft Studio (structural)
blocks built only from your approved facts, no language modelThree ready blocks for the page that should win a question: an FAQ answered from your approved facts, a comparison table with your column filled and competitor cells left for you, and a visible last-updated date. Every number has a fact behind it; gaps are marked, never guessed.
Structural layer only (owner decision 2026-09-06). A short model-written answer paragraph, with a verifier and your approval, is a separate decision after the doors A/B read.
Fix queue
ready fix filesDo AI engines agree on the facts about you
Registered facts checked against AI answers, your own site and third-party listings, with drift and stale echoes per engine.
SEO × AI, AI traffic and the topical map
Where Google ranks you but AI does not cite you, where AI cites you but Google does not, and which sessions already come from assistants.
SEO vs AI
Google ranks, AI does not citeAI traffic
GA4, sessions per engineTopical authority
content coherenceWhich pages the AI crawlers read
Requests from GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Googlebot, Bingbot and others, from your own logs. Crossed with the pages the engines cite. Observations, not causes.
Connect Cloudflare
read-only analytics token, encrypted at restIn Cloudflare: My Profile → API Tokens → Create Token → Custom token → Permissions: Zone · Analytics · Read → Zone Resources: only your site. Paste the token here. We verify it, store it encrypted and can delete it in one click.
Or upload a log file
nginx / Apache combined or JSON lines (Cloudflare Logpush), up to 20 MBOnly aggregates are kept (day × crawler × path). IP addresses and raw lines are discarded during parsing.
Crawlers, last 30 days
requests per crawler| Crawler | Why it reads | 7 days | 30 days | Last seen |
|---|---|---|---|---|
| Loads from live data. | ||||
Top pages per crawler
Read × cited
your pages, last 30 daysCorrelation only. A page that is read and cited is not proof of a link between the two.
robots.txt vs reality
Badge, alerts and domains
A public score badge to embed, the full alert feed and the domain switcher.
CitePulse badge
embed your score on your siteThe badge unlocks once your score qualifies (active monitoring and a score above the threshold).
Domains
click to switchAll alerts
full feedPlan, usage and invoices
Your current plan, payment method and invoice history.
Plan and subscriptions
Payment
Plan changes, card updates and cancellation go through the Stripe portal. We never store card data.
Invoices
history| Number | Date | Amount | Status | |
|---|---|---|---|---|
| Loading invoices… | ||||