A single visibility score hides more than it reveals. A brand can score 80% while being cited negatively. Another can score 40% but dominate the highest-value buyer prompts. Here is what to measure instead — and why the number your dashboard shows you is often the least useful thing on the page.
1. A scenario that exposes the problem
Brand A has an AI visibility score of 78%. Brand B has 65%. A stakeholder reviews both scores and concludes Brand A is in better shape. Is that conclusion correct?
You cannot tell. Brand A might be cited at 78% across a mix of informational and commercial queries — but when you separate out the high-intent buyer prompts (the ones that actually drive purchase decisions), Brand A might appear in only 30% of them. The 78% is boosted by easy informational mentions where the brand name appears incidentally, not as a recommendation.
Brand B's 65% might look worse — but if 60% of those appearances are on comparison and alternatives prompts (high-intent, late-funnel queries), Brand B may be generating more actual commercial influence per appearance than Brand A's broader but shallower presence.
This is not a theoretical edge case. It is the regular consequence of single-score reporting — it averages together results that should never be averaged, then presents the average as if it is a meaningful number.
2. Five dimensions that single scores flatten
Prompt intent distribution
A visibility score that mixes informational, navigational, and commercial prompts into one number tells you nothing about where you appear. A brand can score high on easy low-intent queries and be invisible on the high-intent queries that drive revenue. Separating by prompt type is the minimum necessary breakout.
Sentiment of appearances
Named mentions are not all equal. Appearing in an AI response as 'Brand A is a good option' is different from appearing as 'Some users report issues with Brand A — you may want to also consider alternatives.' Both count as a mention. Only one is a positive signal. A score that doesn't separate positive from cautionary mentions can be very misleading.
Result consistency (volatility)
A brand that scores 65% might appear in 65% of queries on average — but on any given day, that could be 30% or 90%. High volatility means the engine is not treating your brand as a stable authority — it is including you sometimes and excluding you randomly. A stable 50% is worth more than an average 65% with wild swings.
Mention vs. citation
Being named in a response text and being cited as a source URL are very different trust signals. A brand can have high mention rates and zero citation rates — the engine knows the brand exists but does not trust its content enough to link to it. A score that combines mentions and citations is blending two entirely different things.
Competitive context
A score of 68% tells you nothing about whether you are ahead or behind. If your main competitor is at 85% on the same prompt set, you are behind. If they are at 45%, you are well ahead. Absolute scores are meaningless without competitive benchmarking on identical prompts.
3. What to measure instead: four separate lenses
Replace the single visibility score with four metrics reported separately. Each answers a different question and points to a different kind of action.
| Metric | What it answers | Action it points to |
|---|---|---|
| Mention rate by prompt type | Where in the funnel are you visible? | Fix by prompt type: content for discovery, structure for comparison |
| Citation rate (URL appearances) | Does the engine trust your content enough to link to it? | Fix: content quality, schema, third-party coverage |
| Share of voice vs. competitors | Are you ahead or behind on identical prompts? | Fix: competitive intelligence, targeted content gaps |
| Consistency score | Is your visibility stable or volatile? | Fix: entity normalization, topical authority depth |
4. How to use the four metrics together
The four metrics are most useful when read in combination, because different combinations point to different root causes and different priorities.
High mention rate, low citation rate
The engine knows your brand but doesn't trust your content as a source. Entity recognition is fine; content quality and structure need work.
High mention rate on informational prompts, low on commercial prompts
You have broad topical awareness but have not built the comparison-ready, decision-stage content that gets cited when buyers are close to purchase.
High mention rate but low share of voice vs. competitors
You appear, but so does your competitor — and at higher rates. The competitor has a structural citation advantage on the prompts that matter most.
Average mention rate but high volatility
The engine does not have stable, consistent evidence for your brand. The underlying issue is usually entity coherence — your brand's signals are inconsistent across sources.
Four Lenses, Not One Score
Citany breaks down every result by prompt type, intent, and sentiment
No single score. Just the specific data you need to understand where you stand and what to fix next.