Practical guide

How to measure your business’s ChatGPT visibility

One question gives you one answer. Useful ChatGPT visibility measurement needs buyer-relevant questions, a recorded environment and repetition.

Sivlo · Updated

Choose buying questions without a brand hint

Record 10–20 real buying situations: what the buyer needs, in which country and language, and under which constraints. A small marketing team’s needs differ from a multi-country agency’s. Do not add Sivlo or your own company name when testing unaided discovery.

Keep a separate branded question set for checking prices, capabilities and company facts. An answer to a branded question does not establish that the brand is recommended for a general buying need.

Record the exact measurement environment

Record the date, question, country, language, service, model and whether search was used. A consumer ChatGPT answer and an OpenAI API answer are different observations. Login state, conversation history and personal context may also affect the result.

Sivlo’s OpenAI observations come from supported API measurement. Read the settings and sources alongside the result; do not describe them as visibility across all ChatGPT users.

Check public-page access

OpenAI uses OAI-SearchBot for search and GPTBot separately for potential training use. Check robots.txt and CDN/firewall rules. An ordinary browser request or a changed user-agent string does not prove the real search crawler can access the site. Check authentic requests in server logs against published verification information.

Public access enables retrieval. It does not guarantee that the service retrieves your page or recommends your business.

Correct the information and follow the same sample

Check an incorrect claim against your pages and the answer’s sources. Add a missing use case only if it matches your actual offering. Record the change, publication and later observations. Count mentions, recommendations and citations separately and report missing responses.

Repeat questions under the same method and examine variability. If sources or results change, investigate; one favorable answer does not establish lasting growth.

Sources and further reading

Measure your own market

See who AI recommends in your market, why, and what to change.

Request a demo