Back to Blog

Measuring AI visibility: findings for decisions

Measure AI visibility for decisions: a report with questions, systems, findings and explicit priorities.

Editorial review:

Reviewed factual claims and sources; added distinct examples and acceptance criteria. Observations, methodological recommendations and unconfirmed causes are distinguished.

TL;DR: Key Takeaways

A useful measurement grid explains the observation and the decision it supports. The cause of an inaccurate answer remains a hypothesis until investigated.

  • ChatGPT Search: Check the price source and assign an owner.
  • Google Search: Record as the technical foundation.
  • Perplexity: Review repeated runs; do not assign a cause yet.

A baseline needs an agreed purpose. Does management want to find incorrect service claims, understand available sources or track a region? That purpose determines the questions, systems and scope. Ten to fifteen questions per language can be a starting point; they are not a universally sufficient sample.

Context and scope

Four columns make the report readable: business question, tested system, finding and priority. Add the date, question-set version and raw evidence. Distinguish Google search results, AI Overviews, ChatGPT Search and Perplexity. A single percentage otherwise hides different observations.

A wrong city or outdated price is an important finding. It might come from conflicting pages, old directories or other sources. The error alone does not establish its cause. First check the attributed source and the current offer.

SOURCE/01 Full connects technical evidence, AI observations and a prioritised 90-day work list. The free short check covers one domain, three questions and three observations. These distinct scopes support an informed decision before a deeper analysis.

From a measurement to a management decision

Hypothetical report for a Swiss service business: three surfaces are reviewed separately. An inaccurate service claim gets high priority if it could lead customers to make an unsuitable enquiry. A missing mention is first repeated and investigated. The matrix below does not contain actual client results.

Review pointExample / subjectInterpretation
ChatGPT SearchOutdated price stated; an old PDF visibly linkedCheck the price source and assign an owner.
Google SearchCurrent service page indexedRecord as the technical foundation.
PerplexityOwned page not cited in this runReview repeated runs; do not assign a cause yet.
ManagementWhich change supports the buying decision?Agree the correction, effort and review date.

How to apply this

  • State the business purpose in one sentence.
  • Attach a source or saved test run to each finding.
  • Assess impact, risk and effort separately.
  • Plan repeat measurement with the same set and disclose changes to it.

Sources and further reading

The next step

Next step

The free short check provides an initial assessment of one domain, three buying questions and three observations. An introductory call establishes whether deeper analysis is useful.

Request an introductory call

Frequently Asked Questions

Which metric belongs first in the report?

The decision determines the metric. When service claims are wrong, factual accuracy matters more than an overall mention rate.

Can systems be aggregated?

Only with a disclosed calculation, weights and visible individual results. An aggregate does not replace separate reporting.

Does an inaccurate answer prove the owned page is wrong?

No. It triggers a review of the sources and offer; it does not establish the cause.