Knowledge
He knows it, but doesn’t recommend it
The model can correctly describe a brand when you ask about it directly, yet fail to mention it even once when you ask about a category. These are two different things: recognition and recommendation. They are built differently and at different speeds.
Last updated: 7 September 2026 · reading time: approximately 6 minutes
The phenomenon
The same model, two different answers
The easiest way to see this is by looking at two questions asked of the same model on the same day.
Question containing a brand name
What do you know about [Your brand]?
Model response
Correct description: what it does, where it operates, and since when. Everything is correct.
Diagnosis works
Question about the category
Who does it? [service] w [city]?
Model response
Three other companies. Your brand isn’t there, even though the model knows it inside out.
No recommendations
This is neither a contradiction nor a flaw in the model. The answer to the first question stems from what the model knows about the brand. The answer to the second question stems from what it can offer as a justification for choosing that particular brand out of a dozen or so in the category. Knowledge alone is not sufficient justification.
Distinction
Diagnosis versus recommendation
The most important line is the third one. Recognition depends on you, recommendation on others, and that explains all the other differences.
| Diagnosis | Recommendation | |
|---|---|---|
| Model question | Do I know who that is? | Do I have a reason to point her out? |
| What makes them up | Content on your own domain | Signals from outside one’s own domain |
| Who oversees this? | You | Others: customers, the media, compilers of lists |
| How fast is it growing? | Weeks | Months |
| It’s clear from the question | With the brand name | A category, unnamed |
| Metrics in the report | The model recognises the brand; the description is correct | Visibility and participation in responses |
Reason
Where does this threshold come from?
Working on your own website boosts brand awareness quickly: the model learns more and more about who you are and what you do. After a few months, this curve flattens out, as the model already knows everything your website has to say.
The recommendation will remain virtually unchanged throughout this whole time, because it is based on something else. The model needs a basis on which to select you from among a dozen or so companies, and your own statement does not constitute such a basis. Nobody writes on their website that they are average.
This is precisely the moment when companies lose heart. They have done everything they could on their end, yet see no results. Meanwhile, the results are there – just on a different curve.
A typical workflow for creating your own content
Conclusion: the model knows who you are
Recommendation: the model lists you
The point at which work on one’s own content comes to a halt
Illustrative chart. It shows the general pattern of the phenomenon, not specific values.
Practical experience
What shifts the second curve?
All these things have one thing in common: you can’t write them yourself on your own website. Someone else has to say them.
Customer reviewsThe cheapestIn places that models read: maps, industry websites, directories. Ask for them from those you’ve already worked with.
Inclusion in the rankingsThe greatest leverageCategory rankings and business directories are often cited more frequently than the companies’ own websites. Being included in just one such list can boost your ranking more than ten articles.
Publications and presentationsThe most durableAn article on an industry website, a podcast, a speech. They remain relevant for years, and the model refers back to them in subsequent responses.
Working on your own website is not a waste of time. Without it, a model has nothing to showcase and won’t even build brand recognition. It’s all about the sequence of steps and not getting stuck at the first stage. We describe the first steps in the article when AI doesn’t know your business.
Diagnosis
How can you tell that this is the case?
Four signs. If they all apply, carrying on writing on your own website won’t make much difference.
✓
The model accurately describes the brand when you ask directly
He is familiar with the industry, the location and the range of services; he does not confuse it with another company.
✕
The brand is not mentioned in the questions about the category
In none of the four models, for none of the unnamed queries.
✕
Competitors’ responses contain references outside their own domain
Reviews, rankings, articles. You mainly have your own website.
✕
The result hadn’t changed after several months of work on the content
Despite the new articles and updated subpages.
Four out of four means that the problem isn’t the content, but the lack of backlinks from outside your domain.
We explain how to check this for yourself in the article how to find out what AI says about a company. In the Aelo report, this is reflected as a discrepancy between recognition and visibility in the category-related questions.
Diagnosis
How can you tell that this is the case?
Four signs. If they all apply, carrying on writing on your own website won’t make much difference.
✓
The model accurately describes the brand when you ask directly
He is familiar with the industry, the location and the range of services; he does not confuse it with another company.
✕
The brand is not mentioned in the questions about the category
In none of the four models, for none of the unnamed queries.
✕
Competitors’ responses contain references outside their own domain
Reviews, rankings, articles. You mainly have your own website.
✕
The result hadn’t changed after several months of work on the content
Despite the new articles and updated subpages.
Four out of four means that the problem isn’t the content, but the lack of backlinks from outside your domain.
We explain how to check this for yourself in the article how to find out what AI says about a company. In the Aelo report, this is reflected as a discrepancy between recognition and visibility in the category-related questions.
Find out which curve you’re on
You start with 250 AeloCoins, without a card. That’s enough to start tracking and test the modules. Once you’ve used them up, you’ll switch to a paid plan. Measurements start immediately, and a full report on your brand is included from the Starter plan onwards. Recognition and visibility in category-related queries are calculated separately.
