Insight · 9 min read

Reviews and mentions: why AI trusts others

An AI system only half believes your own website. It looks for confirmation from others. This is which reviews and mentions count, and how to get them in a decent way.

By NXEO · 17 Sep 2026
In short
  • Your own website says what you think of yourself. A system looks for sources that confirm it.
  • A review with text is worth far more than five stars without. It gives AI words to write about you.
  • A few real mentions in places that fit your trade do more than a hundred bought ones.

Suppose you are looking for a builder and find two companies. The first has a splendid website stating that it is the best. The second has a simple website, forty reviews in which customers describe what was done, and a listing with the trade association.

Whom do you call?

An AI system reasons in a comparable way. What you say about yourself is a start. What others say about you is decisive. Reviews and mentions in AI therefore play a larger part than many business owners think.

In this article we explain why, which sources count and how to get more of them in an honest way.

Why a system looks for confirmation

An AI assistant that recommends a company takes a risk. If the recommendation turns out badly, the user loses trust in the assistant. The systems are therefore built to be careful.

That care shows in the search for several sources. If only your own website says you renovate bathrooms in Apeldoorn, that is one voice, and a partial one. If your business profile says the same, there are twenty reviews about bathrooms and a supplier lists you as a dealer, there are four voices confirming each other.

There is a second reason. A system has to explain its recommendation. It can only write "customers value this company for its clear planning" if customers have written that down somewhere. Reviews are the material with which a system can say something about you.

Your website tells a system what you do. Others tell it whether that is true.

Which sources count

Not every mention is worth the same. This is how we rank them, from most to least influential for most local companies.

1. Reviews in your Google Business Profile

For Gemini and the AI answers in Google this is the nearest source. Google uses the details from business profiles for local questions. But other systems also come across these reviews when they search for you.

2. Review platforms from your sector

Every sector has its own places. For construction, for care, for hospitality, for business services. A system looking for a "good physiotherapist in Zwolle" often ends up on such platforms, because they are precisely about that question.

You find out which platforms those are for you by asking the question yourself and seeing which sources a system quotes.

3. Listings with organisations that know your trade

A trade association, a quality mark, a professional register. They confirm that you exist and meet certain requirements. To a system they are reliable sources, because they were not written by you.

4. Partners and suppliers

A manufacturer that lists its approved installers per region. A software supplier with a list of partners. A colleague who refers to you for work they do not do themselves. These are mentions with context: they say what you do and often where.

5. Local media and associations

A piece in the local paper, a listing with the business association, the sports club you sponsor. No big names, but they connect you to a place. For local findability that is valuable, as you can read in getting found locally in AI.

6. General business directories

They count least, but they can do damage. A directory with an old address or a service you no longer offer confuses a system. So check them above all for mistakes.

What makes a review valuable

Five stars are pleasant. To an AI system they are almost nothing. A number does not say what you did, for whom and where.

Compare these two.

Review one: five stars. "Great!"

Review two: five stars. "Our flat roof leaked at the roof vent. Within two days someone came to look, and a week later the whole covering had been replaced. Clear quote beforehand and no extra work afterwards. House in Deventer, Colmschate."

The second review tells a system five things. You do flat roofs. You solve leaks. You respond quickly. Your quotes are right. You work in Deventer. Each of those can come back in an answer.

So a good review contains:

  • what was done, in ordinary words;
  • where, the town and possibly the neighbourhood;
  • what went well, concretely;
  • how the process went.

How to get reviews like that

You cannot dictate to a customer what to write. You can make it easy for them.

Ask at the right moment. Straight after completion, when the customer is satisfied. Not three weeks later by email.

Ask in person. "Would you be willing to write a review? It really helps us." That works better than an automated message.

Give a direction. "It helps if you briefly describe what we did and where." Most people do not know what to write and are glad of a pointer.

Make it one click. Send the direct link to the form. Every extra step costs you half the responses.

Make it a habit. One review after every job is better than a campaign in which you collect thirty in a week. A sudden wave stands out, and a steady stream shows that your company is alive.

Respond to every review

A reply from you is also text a system reads. Use it.

"Thank you, Marieke. Glad the new roof covering in Colmschate is to your liking." That one sentence contains the type of work and the town once more.

Respond to criticism too, calmly and on substance. Explain what happened and how you resolved it. A company without a single critical review is not believable. A company that responds decently to criticism inspires trust. With people, and in what a system summarises about it.

How to get mentions

You cannot force mentions, but most business owners leave a number of them lying there for the taking.

  1. Go through your memberships. Are you a member of a trade association or business association? Check whether you are on their website, with the correct details and a link.
  2. Ask your suppliers. Many manufacturers have an overview of dealers or approved companies. If you belong on it and are not there, ask.
  3. Ask your partners. Do you often work with an architect, an estate agent or a colleague in another trade? List each other.
  4. Share your knowledge. A local paper or a trade magazine is always looking for people who can say something sensible. A short piece on a subject you know about produces a mention you cannot buy.
  5. Show projects with others. An architect who shows their project and names you as the contractor. A customer with a company who thanks you on their website.

It is not about the quantity. Five mentions in places that fit your trade and your region weigh more than fifty on websites nobody visits.

What not to do

Buying or inventing reviews. It is prohibited, it can be recognised and the risk is large. Platforms remove them and can restrict your profile. And one exposed fake review does more damage than ten real ones make good.

Swapping reviews. You write one for me, I write one for you. The platforms recognise the pattern.

Buying links. Packages of a hundred mentions on unknown websites deliver nothing and can work against you.

Selecting only satisfied customers in an unfair way. Asking everyone for a review is fine. First sounding out customers and only sending the satisfied ones on to the public form is not allowed on various platforms.

Paying customers for a positive review. A reward in exchange for a good rating is misleading. A thank you afterwards, for whoever wrote a review of any kind, is something else.

Make sure it is the same everywhere

Mentions only work if a system understands they are about the same company. That goes wrong more often than you would expect.

On your website you are called "De Vries Construction". In your business profile "De Vries Bouw BV". On the review platform "De Vries Builders Deventer". The trade association still has your old address.

To a person it is clear. A system sees four companies that resemble each other, and is not sure whether the reviews of one belong with the website of another.

Choose one spelling for your name, one address and one phone number. Adjust them everywhere. It is dull work for an afternoon, and it makes sure all your mentions add up.

How to see what a system thinks of you

You can simply ask.

  1. "What do customers say about [your company name] in [your town]?" The answer shows which reviews a system knows and how it summarises them.
  2. "What is [your company name] known for?" Does something come out that is right? Or does it stay vague?
  3. Look at the sources. Many systems show where they get their information. Are there platforms among them that you are not on? Or directories with old details?

If the answer is "I could find little information about this company", you know what to do. If the answer is wrong, read AI says something wrong about my company.

Reviews on your own website

Reviews on your own website are useful for visitors. To an AI system they count for less, because you decide what is there.

That does not mean you should leave them out. Put them on, with the customer's name and town and what was done. Link to the place where the review originally stands. Then a visitor sees the proof, and a system can check it.

What works even better: turn a good review into a project description. The customer says the roof was replaced quickly. You add what was wrong, what you did and how long it took. Then you have the customer's confirmation and the tradesperson's explanation on one page.

How many is enough?

There is no magic number. What counts is how you compare with the companies you are compared with.

If the competitor who keeps being mentioned has a hundred and twenty reviews and you have eight, that is a difference a system sees. If you both have around thirty, it comes down to the content.

More important than the number is the regularity. Reviews from three years ago say little about now. A few per month, year after year, shows that your company is running.

The order in which to tackle it

  1. Make your details the same everywhere. Otherwise your mentions do not add up.
  2. Respond to all the reviews that are already there. The old ones too.
  3. Start asking after every job. With a direction for what the customer can write.
  4. Collect the mentions that are there for the taking. Memberships, suppliers, partners.
  5. Look for one chance per quarter to share your knowledge. A piece, an interview, a talk.

This is work that grows slowly. You cannot finish it in a month, and you do not have to. It adds up.

What we do and do not do here

For completeness: NXEO writes and publishes content on your own website and measures how AI mentions you. We do not collect reviews for you. That should stay between you and your customer.

What we do show is where it rubs. Which sources a system quotes when it writes about your trade in your region, and whether you appear there. That is in the free baseline check. The other causes of a company being missing are in why am I not in AI recommendations.

Curious how AI mentions your company today?

Enter your website. Within 48 hours the report is in your inbox, free.

Read next

Insight7 Nov 20268 min

Is ChatGPT the Google of 2026?

More and more people put their question to ChatGPT. Is it replacing Google? No, but it is changing where customers make their choice. This is what that means for a business owner.

Read article →

Where can we send the report?

In your inbox within 48 hours. No newsletter, no spam.

Website