Chapter 1

Robots make a list, and you can check it

A toy shop. Three robots stand behind a counter, in front of a wall full of toy cars. Each robot has a speech bubble with a few toy cars in it, and the lists have different lengths. A child looks up at them, and a man with a clipboard takes notes.

Imagine you walk into a toy shop and ask a robot which toy car to buy. The wall behind it holds hundreds of cars. The robot names some of them.

Your customers can now do the same with AI. They can ask which accountant, pizzeria or car dealer to choose, and the answer can be a list of names.

So I tested how such lists behave, on companies in five kinds of business.

The short version. The same names keep coming back, so you can count them. Check more than one robot, because they do not always agree. If no name on your own list comes up, the robots may be naming someone you left out.

What I did

Picture asking the three shop robots the same questions, with your own list of names on a clipboard. You tick a name each time a robot says it. My list held 50 companies, and a computer program did the ticking.

The same names keep coming back

For this picture, I left out 50 questions that named the companies in the question.

Four pizzas, each cut into 10 slices. Three pizzas have 8 or more green slices. The fourth has 5 green slices.
On about 3 of every 4 questions, at least one company kept coming back

Picture it. Each pizza is all the answers to one question. Green slices are the share that named the same company. On 3 pizzas out of 4, that company is on at least 8 of the 10 slices.

At Javier’s: like 3 nights out of 4 when at least 8 of every 10 order slips ask for the same pizza.

Often that company was not alone. One answer can name many companies, and on many questions two or more companies reached 8 slices.

Robots often match, and often do not

Ten boxes, each with three toy cars. In 6 boxes all three cars are the same green. In 3 boxes the cars do not all match: in two of them two green cars match and one differs, and in the third all three differ. In the last box one car is missing.
All three robots gave the same top name on about 6 of every 10 questions

Picture it. Each box is one question. Each car is the company from my list that one robot said most often. In 6 boxes the three cars match. In 3 they do not all match, though two often do. In the last one, a robot named nobody from my list or gave no answer.

At Samira’s: like three staff members picking the same ad in 6 of every 10 campaigns.

Months later, most top names were the same

This picture uses the 237 questions all three robots answered in February.

A car park with 100 spaces. 88 cars are green and 12 are white.
GeminiA bar filled to 92 out of 100.92
GPT, behind ChatGPTA bar filled to 91 out of 100.91
PerplexityA bar filled to 73 out of 100.73
From February to September, the top name from my list stayed the same on 88 of every 100 questions

Picture it. A car park with 100 spaces, one for each question. The car in a space is the top name for it. Come back in September, and 88 cars are still in the same space. White cars are new. The bars show this count for each robot.

At Marta’s: like 88 of every 100 product groups keeping the same best seller.

A single answer can still change when you ask again. Perplexity changed most in this test.

Bigger companies came up more often, with clear exceptions

On my list, companies with bigger sales tended to be named more. The pattern was clear for software and for payment and banking companies. In the other three kinds, the numbers were too uncertain to say. Some companies break the pattern, even in software.

Five shops in a row. Two are green and three are white.
In 2 of my 5 kinds of business, the biggest seller was also the most-named

Picture it. Each shop stands for one kind of business. A green shop means the company with the most sales is also the one the robots name most. A white shop means the robots name another company most.

At Daniel’s: like the biggest law firm also being the one people name most, in 2 of every 5 towns.

One pizza. 85% of it is yellow, and the rest is white.
Salesforce made 85% of the sales on my software list. The robots still named HubSpot most.

Picture it. All the sales as one pizza, shared by 10 companies. The big yellow part is Salesforce: 85 of every 100 coins earned. The white slice is HubSpot, second on the list by sales, and the other eight.

In Giulia’s town: like one accountant earning 85 of every 100 euros spent on accountants, while people keep naming another.

HubSpot’s lead was small, and Salesforce came first on some kinds of question.

If your list comes up empty, check your list

On some questions, no company on my list came up in even half the answers. It looked as if no company stood out. Then I had a robot read those answers, with my list put away, and I checked its work.

Follow Up Boss was in all 15 answers, and it was not on my list

One question asked for the best CRM, which is software that keeps track of customers, for people who sell homes. Homes have keys, so each key here is one answer.

HubSpot, the top name on my list15 keys, 6 of them green.6 of 15
Follow Up Boss, missing from my list15 keys, all of them green.15 of 15

At Joost’s: like 15 of 15 readers naming a book missing from the catalogue, and 6 of 15 naming the catalogue’s most-named title.

250 small white houses in rows. 19 of them are grey.
19 of my 250 questions looked empty. On a second look, 3 still did.

Picture it. A small town with 250 houses, one for each question. The 19 grey ones look empty, so you go and knock.

The same 19 houses in a row after the second look. 15 are green, 1 is pale green and 3 are still grey.

In the 15 green ones, someone is home: a company I had not listed came up in more than half the answers. In the pale green one, it came up in exactly half. In the 3 grey ones, the robots named many companies, and none in more than half the answers.

At Kadri’s: like 19 of 250 new homes looking unsold, until the sales book shows that only 3 really are.

How to check this for your business

Portrait of Mikko, the car dealer.Mikko VirtanenCar dealer, FinlandMade-up example

Mikko writes down the dealers he competes with. He asks the robots who is best for a family car, and his list rarely comes up.

Is nobody being named? Or are the robots naming a dealer he forgot? He will only know if he reads the answers.

Every business has its own questions. Here is a first one for each of our owners.

Javier, pizzeria. Where can we get good pizza for a birthday party in Seville?

Giulia, accountant. Who can do the books for a small building firm in Milan?

Daniel, lawyer. Who can check my work contract in Berlin?

Marta, online shop. Where can I buy a good birthday present online in Poland?

Samira, media agency. Which agency in Lyon can run ads for a small hotel?

Joost, publisher. Which Dutch publisher takes a first novel?

Mikko, car dealer. Which dealer in Tampere is good for a family car?

Kadri, homes. Who builds energy-saving family homes near Tallinn?

Try this on Monday, by hand, with a sheet of paper

  1. Write one question a customer would ask when choosing a business like yours. Use their language, name your town and leave your own name out.
  2. Pick two robots, such as ChatGPT and Gemini. Ask each robot three times, in a new chat each time. One answer can change when you ask again.
  3. On your sheet, give each company named a row and each robot a column. Add a tick each time an answer names that company. Tick a company only once per answer.
  4. Compare the columns. Did the robots name you? Who kept coming back? Keep this sheet and add more questions as you go.

This is a starting look. Chapter 4 explains how to plan a bigger check.

References

Chapter 1. Robots make a list, and you can check it

Peer-reviewed, journal version (paid access). Dmitrij Żatuchin (2026). The synthesised consideration set: brand inclusion and concentration in large language model recommendations across five industries and three systems. Journal of Marketing Analytics, published online 31 October 2026. 10.1057/s41270-026-00562-3.

Free preprint. arXiv:2606.23057. Data, version 3.0. 10.5281/zenodo.22927143.

What does AI say about your business?

Put the book into practice. Tell us your company and website to request an AI visibility audit. We’ll reply by email to agree what to check.

What the audit shows

  • Where your company appears in the answers, and next to whom
  • The pages AI cites in your market
  • What to publish first on your own site

Fields marked * are required.

Who is it for?

Your choice about cookies

We use cookies this site needs to work, and others only if you say yes. Choose what you allow, and change your answer whenever you like. Read our privacy policy