Ask twice, get two answers

It is not a fault in the check

Put the same question to the same assistant twice and you can get two different answers, naming different businesses. No engine publishes why, and there is no factor list to look up — not from OpenAI, not from Google, not from anyone. Anybody printing one is guessing.

What is observable is that it happens, and that it happens often enough to change what a single result can honestly claim.

One pass is a sample, not a rate

Ten questions asked once tells you what two assistants said on one day. It does not tell you how often they say it, and no arithmetic applied afterwards can turn one pass into a frequency.

This is why you will not find a percentage on a result here, or a decimal point anywhere. A percentage implies a rate. A rate needs repeats. Printing one off a single pass would be dressing a sample up as a measurement, which is the specific thing this product exists not to do.

What the free check does about it

The free check asks each question once, and the result says so on its face rather than in a footnote at the bottom. Read it as one day’s sample and it is useful. Read it as a score and it is not.

The free scan is not averaged over repeated runs. There is no checkout on this site, so buying more runs is not something you can do today either — what is here is one pass, described as one pass. What costs what is on its own page, including the parts that have a number and no way to pay it.

What repeats would actually change

Repeats do one thing, and it is the only thing: they turn a yes into a count. Cited on 3 of 5 runs is a fact you can hold next to next month’s. Cited, from a single run, is an anecdote about a Tuesday.

That is the whole reason a single-check number is never published as if it were a rate, here or anywhere else in this product.

How to read your own result without fooling yourself

• Re-run on a different day before you believe a change. The per-connection count resets at midnight UTC, so tomorrow you have another five.
• Compare the shape, not the number. The same competitor named on the same questions across two different days is a signal. One question moving from missing to mentioned is weather.
• Treat a single missing as a question worth looking at, not as a verdict about your business.
• Take the citations more seriously than the mentions. Being read is a harder thing to do by accident than being named. The difference is here.

What this rules out printing

• A single blended visibility figure.
• A percentage derived from one pass at each question.
• A trend line drawn through one point.
• A ranking position, on any surface. An AI answer has no positions in it.

If this page did not answer it

Write to the one inbox. There is no chat widget, no ticket queue and no phone number — a message arrives with the person who wrote the thing you are asking about, and gets answered by hand. Send the domain you checked, what came back, and what you expected instead. That difference is the whole message.

Any question that arrives twice becomes a page here, so writing in is also how this section gets longer.