A dated measurement, zeros included: what our own product showed
Why measuring AI answers takes a date and the same questions and assistants before and after, and why we keep the zeros. Example: ShotBook.
Published
4 min read
In short
- A number about AI answers can only be checked against three things: the date, the questions and the list of AI assistants.
- The questions and the assistants are written down before the start and don't change between measurements; each question goes into a fresh chat.
- The zeros stay in the report: without them, there is nothing to compare with. For ShotBook, five checks in a row gave 0 of 46; the sixth, on September 7, 2026, gave 10 of 46.
“AI has started mentioning the company more often” could mean anything. You can only check a claim like that against three things: the date, the questions and the list of AI assistants.
Here is why each one matters, and what a series of measurements looks like on our own product, zeros included.
What one measurement is
One measurement means asking the chosen AI assistants your customer's questions, in their words and without the company's name, each in a fresh chat.
For every answer, we record who it names and where it gets the answer. The result is a number for each assistant: how many answers out of how many named the company.
Why the date
A number without a date can't be checked or compared: you don't know which moment it describes. And AI answers change - the ShotBook example below shows a zero turning into a different number. That's why we don't quote numbers without a measurement date.
The date also keeps an old number from passing for today's.
Why the same questions and the same assistants
You can only compare before and after if nothing else changed. Swap a question for an easier one, and the growth may come from the wording. Add an assistant that already knew you, and the same thing happens.
Before
After
So the questions and the assistants are written down before the start, and we don't change them between measurements. Each question goes into a fresh chat, word for word. What counts as a result is written down before the start too.
Why show the zeros
Zero is the starting point: it's the only way to see whether anything changed. And zero is an honest answer to “is it working?”.
We don't hide the zeros: if AI still doesn't name you, you'll be the first to know.
The zeros stay in the report, and you decide whether to continue.
Example: ShotBook
ShotBook is our own product and a public case study, a catalog of video references for directors and editors. We measured how many times AI named ShotBook.
How many times AI named ShotBook
0of 46
10of 46
The five zeros aren't an awkward part of the story to cut. Without them, the sixth check has nothing to be compared with.
What this series doesn't show
Why zero turned into ten. A measurement records what changed in AI answers, not the cause. And a number from September 7, 2026 describes September 7, 2026, not what AI answers today.
Our own zero
We apply the same approach to ourselves. On October 2, 2026, when we were still called shyash.ai, we asked AI assistants buyer questions about our service, without our name. None of them named us: 0 of 96 answers.
That zero is the starting point. We will run the next measurement with the same questions under our new name, aigo, and count mentions of our former name, shyash.ai, too.
- answers named us, measured on October 2, 2026
- 0of96
What to ask about any measurement
When someone shows you growth in AI answers, ask:
- On what date was each measurement taken?
- Which questions were asked, and did they include the company's name?
- Which AI assistants, and were they the same before and after?
- Did each question go into a fresh chat?
- What did the first measurement show, even if it was zero?
If there are no answers, there's no way to check the number.
Same questions, same AI assistants, fresh chats. We show you the difference. Zeros included.