What is an AI visibility score, and what does it miss?
Learn what an AI visibility score measures, why it varies, and why it is not a SearchSeal metric.
An AI visibility score summarizes how often a brand appears across a chosen set of prompts and model responses. It can be useful for a controlled research program, but it is not a universal rank, a direct measurement of crawler activity, or a SearchSeal metric.
How AI visibility scores are produced
A tool selects prompts, runs them against one or more AI systems, records mentions or citations, and combines those observations into a score. The result depends on the prompts, model versions, locations, timing, sampling method, and weighting formula.
- Mention frequency across the selected prompt set
- Position or prominence inside generated answers
- Citations or linked domains
- Comparison with selected competitors
- A proprietary weighting formula
Why the score is not a universal rank
There is no single result page shared by every user. Responses can change with wording, context, model updates, browsing behavior, and randomness. Two valid studies can produce different scores without either calculation being fraudulent.
The score is evidence about a defined experiment. It should name the prompt set, platforms, period, sample size, and formula. Without that context, a precise-looking number can imply certainty it does not have.
Keep four measurements separate
Answer samples show what a defined prompt study returned. Human AI referrals show visits from AI products. Analytics can connect those visits to sessions, goals, and revenue. Crawler evidence shows requests from identified agents. These measurements answer different questions.
A human AI referral is evidence of a visit. It does not prove that a crawler caused the visit, that a citation caused the visit, or that the visit converted. A crawler request is evidence from infrastructure. It does not prove indexing, citation, recommendation, or conversion.
What SearchSeal measures instead
SearchSeal does not publish an AI visibility score. It provides privacy-friendly website analytics, including human AI referrals, sessions, goals, and revenue, alongside separate crawler evidence for requested paths and HTTP outcomes when available.
Use prompt studies to examine model answers. Use analytics to measure human visits and business outcomes. Use crawler evidence to diagnose access and delivery. Keep the labels separate, and never turn an unknown period or uncovered source into a zero.
Questions to ask before accepting a score
- Which prompts, models, locations, and dates were included?
- How were mentions, positions, citations, and sentiment weighted?
- How large was the sample, and how often was it repeated?
- Can the underlying observations be exported and reproduced?
If those answers are missing, treat the score as a directional vendor metric. Do not use it as a factual statement about total market visibility.