A visibility run is a repeated check of AI answers: you ask ChatGPT, Perplexity, Claude and Google's AI the same buyer questions several times each, and record how often your product is named. It matters because AI answers change from one run to the next, so a single check says little about what buyers actually see.
Why one check is not enough
A search engine results page is fairly stable. Ask Google the same thing again and the blue links are mostly the same. AI answers are written fresh each time. The model can phrase the answer differently, run different searches behind the scenes and pick different sources.
Part of that comes from query fan-out: one question becomes many background searches, and those searches shift between runs. Part comes from the model itself, which does not produce the same text every time.
So "ChatGPT named us" after one try is an anecdote. Named in most runs is a position. Named in some runs is a foothold you can build on.
How a visibility run works
- Pick the questions. Use the ones a buyer asks before they choose, in their words: "which tool tracks AI crawlers", "how do I get my SaaS named by ChatGPT". Skip questions with your brand name in them; those tell you nothing.
- Fix the setup. Same wording every time, a logged-out or fresh session, no earlier conversation for the model to lean on.
- Ask each question on each engine several times.
- Record per answer whether your product is named in the answer text, and which pages the answer cites.
- Save it. A visibility run is only useful next to the last one.
How to read the result
Write the result as a count per engine: named in how many of the runs. That reads differently from a yes or a no.
Named in every run means the engine has settled on you for that question. Protect it: keep the cited pages current.
Named in some runs means you are close. More coverage of the related questions, a listing on a page the engine already cites, or a clearer answer on your own page can tip it.
Named in none, on every engine, usually means the engines cannot find you or have no reason to trust you yet. Check that AI crawlers can read your site, then get onto the lists and directories the answers cite.
Then compare engines. Each one searches and cites differently, so being named on one and missing on another is normal and tells you where to work next.
An example
A founder runs a visibility run on the question "how do I check if ChatGPT mentions my product". On the first run, one engine names their tool and cites their answer page. On the runs after that it does not. That is a foothold: the page is good enough to be picked sometimes. They add a clearer first sentence, get listed on a "best tools" page the engine cites, and run the same question again a few weeks later.
How to run one
By hand, a spreadsheet works: a row per question, a column per engine, a cell per run. It takes time, and the hardest part is asking the same way every time.
For founders who build with Claude Code or Codex, Shipfound does the runs and keeps the history. It asks the same buyer questions on ChatGPT, Perplexity, Claude and Google's AI in repeated runs, records whether the product is named, and compares each run with the last one.
Common mistakes
Changing the questions between runs. Then you cannot tell whether you improved or the question got easier.
Checking while logged in. Your history and past chats can steer the answer toward you.
Counting a source link as being named. If your page is cited but the answer recommends someone else, the buyer heard someone else's name. Track both, separately.
Checking one engine. Buyers use several, and they rarely agree.
Frequently asked questions
Is a visibility run about running at night?
In running, a visibility run usually means a run in low light with reflective gear. This page is about the AI search meaning: checking whether AI answers name your product.
How often should I do a visibility run?
After you ship changes, and after the engines have had time to recrawl the pages. Running it daily mostly measures noise.
How is a visibility run different from rank tracking?
Rank tracking records your position in Google's results. A visibility run records whether AI answers name you. A page can rank well on Google and still be left out of the AI answer. More in alternatives to generic SEO dashboards for AI search visibility.