Method
How the measurement works.
Written out in full so you can argue with it. A method you cannot check is a number you cannot trust.
Step One
The site is read before a single question is written
Your style, the venues you keep coming back to, the words you use for what you sell, the price signals you show. Twenty questions built on a template of city plus craft would measure nothing, because no couple asks that way twice.
Step Two
Twenty questions, none of them containing your name
Broad questions, buying questions, price questions, style questions, and questions about the specific venues and niches you actually work with. The point is to see whether you surface on an open question, the way a couple who has never heard of you would ask it.
Step Three
Four services, three runs each
Perplexity, ChatGPT with web search, Gemini and Claude. Each question is asked three separate times to each service. Twenty questions across four services across three runs is two hundred and forty answers, and every one of them is stored, hit or no hit.
Step Four
A majority decides, not a lucky run
Named or cited in at least two of the three runs and you hold the question. Exactly one of three and it lands in a separate section, the near wins, which is usually where the least work is left to do. Zero of three and the report names the businesses that took the place instead.
The alternative rule, counting a hit if any single run finds you, inflates the number every time you run it again. It measures generosity, not visibility.
What counts as showing up
Two things count, and the report keeps them apart. Your name written into the answer, and your website cited as one of the sources the answer was built from. Being cited without being named is a real position: it means the service read your site and used it, which is the step before it starts recommending you.
The two are not evenly distributed. Across the measurements run in September 2026, Perplexity cited a measured business site in one hundred and ninety eight answers and Claude in ninety three, while Gemini returned no readable source list at all. If you only counted names, you would throw away most of what Perplexity tells you.
What the method cannot tell you
It cannot promise you a place in an answer. Nobody controls what these services say, and anyone selling you a guarantee is guessing.
This measures visibility. It does not influence it, and no tool can. What follows a measurement is ordinary published standards: letting the crawlers in, marking up your facts so they are read correctly, stating your prices and service area plainly, and being present where the assistants already look. Anyone selling influence over an AI answer is selling nothing.
It does not measure Google rankings. It does not see what any individual couple was shown, since answers vary by person, phrasing and moment. And a single month is a snapshot. The number worth watching is the direction across months, which is why the questions stay identical from one measurement to the next. Change the questions and you have thrown the trend away.
Every answer is stored, including the ones where nothing was found, so any question about a past measurement can be answered without running it again.
The short version
Run it on your own site.
Four questions, two services, about a minute. No card.
Run my free check