How scoring works
What the AI-visibility score actually predicts
Cite Me reads your public pages and rates how easily an AI answer engine — ChatGPT, Perplexity, Google AI Overviews — could quote them. It runs entirely on-page; no orders or customer data are touched.
PREDICTED (heuristic) Every score is a prediction from the 9 signals below — not a measured citation count and not a ranking. A scan reads up to 100 of your URLs per run.
Two scores, not one
Cite Me reports a page score and a store-readiness score, because the two are fixed in different places. The 4 signals in the first group are what you change on an individual page, so they — and only they — make up that page’s 0–100 score. The 4 in the second group are store-level rather than per-page (the theme app embed, your brand’s Organization data, your robots.txt), so they are scored once for the store instead of dragging down a product page that could never fix them alone.
Page score — the 4 on-page signals
Each percentage is that signal’s real share of the 0–100 page score. Each signal is rated 0–100 on the page, then combined in these proportions.
Citable passages
34%Self-contained answers about 130–170 words long — the kind of complete chunk an AI can quote whole.
Improve it: Write one complete answer paragraph per question heading.
Front-loaded answer
26%Whether the answer or definition appears in the first ~120 words, not buried at the bottom.
Improve it: Lead with the answer in the first one or two sentences.
Question headings
21%How many headings are phrased as the real questions customers ask.
Improve it: “How long does it last?” works better than “Durability.”
Meta title & description
19%The page’s meta title (30–60 chars) and description (70–160) — the summary AI search and Google read to understand and cite the page.
Improve it: Write a focused title and a descriptive meta description (the “SEO meta” button does this from your own content).
Store readiness — the 4 store-wide signals
Each percentage is that signal’s share of the separate store-readiness score. Fixing one of these lifts the whole store at once.
Structured data
33%AEO-useful JSON-LD on the page (Product, Article, FAQPage, HowTo, BreadcrumbList).
Improve it: Turn on the Cite Me theme app embed once — it adds the right schema store-wide. (Ratings/reviews only when real on-page reviews exist.)
Freshness
25%A visible or structured “last updated” date.
Improve it: Show an Updated date and set dateModified in JSON-LD — the same theme app embed does this store-wide.
Entity sameAs
22%Organization sameAs links so AI can confidently identify your brand.
Improve it: Add Organization/Brand JSON-LD with sameAs links to your official profiles.
AI crawler access
20%Whether AI crawlers (GPTBot, ClaudeBot, PerplexityBot, Google-Extended…) are allowed to read your store.
Improve it: Allow them in robots.txt and per-bot directives.
Reported, but counted in neither score
Cite Me rates this signal and tells you what to add, but deliberately does not score you down for it: it depends on real figures only you have, which no button can generate honestly. It contributes 0% to the page score and 0% to store readiness.
Evidence density
0%Concrete numbers, units and figures that make claims quotable rather than vague.
Improve it: Replace “long-lasting” with specifics — hours, ml, %, named attributes.
How the fix list is ordered
The dashboard shows your store average, all 9 signals, and a prioritized list of fixes — ordered by predicted impact (a signal’s weight × how far below target it is), so the biggest wins come first. The full weight table lives in app/lib/signal-weights.ts; the percentages above are derived from it, never typed by hand.
What this score is — and isn’t
- It is a heuristic prediction of how quotable your pages are to AI answer engines, based on what those engines tend to reward.
- It is not a measured citation count, a search ranking, or a “rich result”. Acting on the fixes is expected to help, but we do not guarantee that any AI will cite you.
- llms.txt is treated as hygiene only — its presence does not boost the score.