How to Read Your CiteCue Visibility Score

CiteCue's visibility score is a 0-100 summary of how often your brand shows up, and how much citation presence it earns, across the prompts and AI engines in a scan. Mention frequency carries about half the weight, citation share roughly a third, and direct citations the remainder. One thing to be clear about up front: this is CiteCue's monitoring metric, not a ranking factor supplied by ChatGPT, Gemini, Claude, Perplexity, or Google.

Step 1: Confirm what the scan covered

Before you read anything into the number, check what actually ran:

• Which prompts?
• Which AI engines?
• Were prompts added, removed, or edited since the comparison scan?
• Did the audience, brand aliases, competitors, or topics change?
• Did every expected result finish successfully?

A score is only comparable when the underlying set is. Add ten branded prompts and mention frequency will jump without your unbranded discovery improving at all. If you materially change the prompt set (worth doing deliberately; here's how to choose prompts for tracking), record the change and treat the next scan as a new baseline.

Prompts Monitoring keeps each result connected to its prompt, engine, brand position, citations, and full answer. Look at that detail before acting on the summary score.

Step 2: Understand the three components

CiteCue builds the score from three related outcomes.

Mention frequency: about half the weight

How often the tracked answers name your brand. It answers one question: across the prompts and engines in this scan, how consistently did we make it into the answer?

A mention doesn't mean the answer recommends you, describes you accurately, or cites your website. Sentiment, position, and the full answer tell you what kind of appearance it was.

Citation share: roughly a third

Citation share compares the sources behind the tracked answers: your domain, competitor domains, and third-party domains. It answers: how much of the cited-source presence belongs to us?

This isn't the same as mention share. An answer can name your brand while citing a review site or a competitor comparison. It can also cite your documentation while discussing a broader topic and barely mention you.

Direct citations: the remainder

How often your own pages get cited directly. Did the engine use your domain as a visible source?

A direct citation is evidence that a first-party page participated in a recorded answer. It doesn't prove permanent authority, a fixed ranking, or that the citation will still be there next scan.

Step 3: Read the score with the underlying metrics

Say the score rises. Open the prompt results and figure out why:

• Mentions increased: did the gain come from relevant unbranded prompts, or only branded questions?
• Citation share increased: which page and topic gained citations? Did a competitor lose ground, or did the total source set just grow?
• Direct citations increased: is one page doing all the work, or several?
• Position changed: are you entering shortlists earlier, or showing up in the caveats at the end?
• Sentiment changed: are more mentions positive, neutral, or negative?

The same score change can describe very different business situations. A five-point gain from direct citations on a high-intent comparison is often worth more than a bigger gain from branded awareness prompts, and the composite number alone can't tell you which one you got. Our guide to reading mention, position, citation, and sentiment results goes deeper on each field.

Step 4: Compare like with like

A few comparison rules:

1. Keep a stable core prompt set for trend reporting.
2. Compare the same engines when evaluating platform change.
3. Separate branded, unbranded, comparison, and task prompts.
4. Read at least a few full answers behind every material change.
5. Annotate launches, page updates, pricing changes, and prompt-set edits.
6. Trust repeated movement across scans over one volatile answer.

AI outputs vary even when the prompt doesn't. The score compresses a pile of results into a trend; repeated scans and answer-level evidence are what make that trend credible.

Step 5: Diagnose a low or falling score

Don't set out to "raise the score" in the abstract. Find the component and the prompts responsible.

• Low mentions: the brand isn't in the relevant consideration set, or isn't associated with the use case
• Low citation share: other domains supply more of the sources behind the tracked answers
• Low direct citations: your own pages aren't being selected as visible sources
• Weak position or sentiment: mentions exist but aren't favorable or prominent
• A fall isolated to one engine: likely a platform-specific retrieval or answer change

Open Citations and Competitors to see which domains show up instead and where the 13-factor comparison says you're losing head-to-head. If access or indexing might be involved, run the checks in AI Readiness before you rewrite anything.

Step 6: Turn the score into a reporting sentence

"Visibility is 42" on its own tells nobody anything. Report the driver and the scope:

Visibility rose from 37 to 42 across the unchanged core prompt set. The gain came mainly from three new direct citations on pricing-comparison prompts in Gemini; ChatGPT mention frequency was flat.

That sentence says what moved, where it moved, and what didn't. End with the next test: improve the cited page, correct a recurring source, close a factor gap, or hold the prompt set stable and see whether the pattern repeats. If you're reporting to clients, our client-report tutorial shows how to fold this sentence into a full reporting cycle.