Your AI visibility score is a summary of the answers measured for that report. To use it well, review the prompts, completed provider coverage and available response excerpts and sources behind the number. Some responses have no saved excerpt. It describes a sample, not what every potential customer sees.
Read the evidence behind the number
- Check the provider and query set.
- Check the measurement date and whether the run completed.
- Distinguish no mention in a successful response from missing or failed coverage.
- Inspect any saved excerpt and returned sources, rather than relying only on a percentage. A missing excerpt cannot explain the full answer.
AnswerMetrix supports Perplexity, ChatGPT, Claude, Gemini, Grok and DeepSeek. It records direct URL citations from Perplexity and mention-based evidence for the other five. Inspect available excerpts and sources with their dates, prompts and coverage. Some responses have no saved excerpt; a missing response is not a measured absence.
Keep report metrics distinct
Website audit scores, AI visibility measures and Share-of-Voice have different definitions. Compare like-for-like measurements and the dates shown. A change in provider coverage or query scope can make two scores difficult to compare.
Use your own baseline
Build a baseline around questions your customers would ask. Compare the accuracy and usefulness of the returned answers over time. A universal score cannot tell you whether a business will be recommended, trusted or profitable.
What to do when a score changes
Inspect the underlying evidence first. Check for new website issues, different query coverage and provider changes. Improve a concrete customer-facing problem, then follow later measurements. Do not assume schema or a meta description caused the movement.
Start from the findings you already have. Run a fresh scan when the site has changed or the evidence is too old for your decision, and check your plan allowances first. Saved reports remain available without starting another analysis.
