What does AI answer today to our own question?
This table is not an example. It is our own measurement, with our own machine, on the question our own buyer asks — and what came out is what is there, including when we do not appear in it ourselves. As long as there is no fresh measurement, nothing stands here: better a section without a table than a table with an invented name in it.
On which question are you actually losing to your competitor?
Six types, each with a weight the scan genuinely uses. This table is drawn from the same code that computes your measurement — change a weight there and this page changes with it. That is deliberate: a page showing a different weight than the report is worse than no page at all.
| question type | weight | what someone wants at that moment |
|---|---|---|
| Informational · info | 0.55 | Orientation - no buying intent yet. |
| Commercial · commercial | 1.00 | Looking for a provider, not yet comparing. |
| Comparison · compare | 0.92 | Weighing options (best, vs, alternative). |
| Ready to buy · transactional | 1.00 | Ready to buy/book - the costliest question to miss. |
| Local · local | 0.80 | Tied to a place or proximity. |
| Navigational · navigational | 0.50 | Looking for a specific brand/site. |
Which engine is he winning on, and what is that engine worth?
A competitor winning ChatGPT wins almost three quarters of the market. A competitor winning only Claude wins a little over four percent. Without this column a ranking is a list without weight, and then a lost position on a small engine looks as bad as a lost position on the largest one.
| engine | share of the market for this language |
|---|---|
| openai | 70.83% |
| gemini | 10.56% |
| copilot | 7.96% |
| perplexity | 6.78% |
| claude | 3.85% |
| deepseek | 0.01% |
Source: Statcounter · AI chatbot market share — Europe, April 2026. It measures referrals to websites, not user counts.
Why does an engine pick him and not you?
Because there was more to read about him at the moment the question was asked. An engine does not choose on sympathy or on advertising budget: it chooses what it can substantiate. Whoever sits on three independent sources while you sit on one wins — even when your service is better.
What do we not do with your competitor?
We do not track him. There is no monitoring screen for a company that happens to bother you: what we know about a competitor is what appeared in an answer to your question. That is a by-product of your measurement, not a file on him.
Frequently asked questions
Can I choose which competitors you track?
You choose the questions, not the names. Who appears in the answer is decided by the engine, and that is exactly the value: a list of competitors you supply confirms what you already believed. Two times out of three there is someone in the answer who would not have been on such a list.
Why is there no percentage next to my competitor?
Because we do not know how visible he is — we know how often he landed in your answers. That is a different thing, and a percentage would suggest we measured his entire market. What we can show is on which of your questions he appeared and which sources sat underneath.
Do you also measure whether we are spoken about positively or negatively?
Yes. Per answer we record three values, plus the caveat the engine adds itself — that last one is the sharpest signal, because 'good but expensive' is not a negative answer and still costs you the customer. The sources underneath come with it.
How often does that order change?
More often than a search result and less often than people fear. Asking the same question twice on the same day can give a different order; that is why only a re-measurement with the same prompts, the same market and the same method counts for us. One deviating measurement is noise, not a shift.
Read on
Free, no account and no card.