In short
- Find out first
- Which of the four it is. Each has a different repair, and three of them cost you nothing if it turns out to be the fourth.
- The proof is waiting
- In the sources under the answer you are missing from. That is a list of addresses, not a hunch.
- What you do not do
- Trying to steer the answer with hidden text or bought signals. It does not work and it is the one thing that can set you further back.
Which four causes are there, and how do you recognise them?
The middle column is the diagnosis and the right-hand one the bill. Note that the four rise in cost and fall in certainty: the first is a setting, the last is a positioning conversation.
| Cause | How you recognise it | What you do about it |
|---|---|---|
| Your website is not being read | the assistant does not know your website, not even when you name it in the question | open the gate for AI crawlers and check that they come by |
| Outside your website there is too little | you get named as soon as your name is in the question — never otherwise | entries in registers and directories, plus mentions in sources the engine already reads |
| What is there is no longer right | the assistant names an old name, an old address or a service that no longer exists | write to the source it came from — the address is under the answer |
| You lose on a property | you get named, with a reason next to it why the other one fits | put the property on your website in a checkable way — or accept that you do not want to win this question |
How do I work out in half a day which one it is?
Ask three versions of the same question: one with no names, one with your competitor's name, and one with yours. What comes out points straight at one of the four. If the assistant does not know you by your own name, it is the first. If it knows you but never names you unprompted, it is the second.
If it names you with wrong details, it is the third, and the address of the error is usually in the sources underneath. If it names you correctly but picks the other with a stated reason, it is the fourth — and that reason is literally the most useful sentence you will read that day.
Can I catch up with someone who has been there for years?
For the first three causes usually yes, and faster than with classic results: there is no accumulated authority to catch up with, there is a selection per question. Whoever is in the source the engine fetches today is in the answer today.
For the fourth, catching up is not a visibility question but an offer question. If the other is chosen because they deliver something you do not, that is a product decision and not a marketing decision — and establishing that honestly is worth more than continuing to try.
Why does trying to steer the answer not work?
Because every attempt to influence the answer directly amounts to a signal that is not true, and an engine recognising it does not weigh that one page lower but your entire domain. The best-known example is hidden text with instructions in it; the most expensive is a run of reviews arriving in one week.
Frequently asked questions
Do you track my competitor for me?
We measure who appears in the answers to your questions, and names come out of that. What we do not do is build a file on another company, crawl their site or collect their data. The difference is that we measure an answer someone else happens to appear in, not someone else.
The assistant names my old company name. How do I get rid of that?
Not by asking the assistant — that changes nothing. It comes from a source, and that source has an address. The order is: find the source, get it corrected, and make sure the right name sits in the structure of your own site as a counterweight. That takes weeks, not days.
Read on
Free, no account and no card.