Chapter 11

Same question, other language, other names

A toy shop. A child at the counter asks a question, shown by two speech bubbles with different marks in them. A robot answers with two bubbles: one holds three green toy cars, the other holds other cars.

Imagine you ask the shop robot for the best toy car, first in English, then in your own language. You may hear other names the second time.

Your customers can do this too. They can ask a robot in the language they think in, and the answer can name other firms.

I tested one question in several languages. It asked for the best accounting software, which keeps track of a firm’s money.

The short version. Check each language your customers use, because the names can differ. The same words also gave other names on another internet connection, so I cannot say language did this. Read the answers yourself, because my own list missed a firm.

What I did

I asked ChatGPT for the best accounting software for freelancers. I tried some combinations of language and internet connection. Then I checked the names against country lists and an international list.

Norwegian firms came up more often in Norwegian

Asked in Norwegian, 6 of 6 answers named a firm from my Norwegian list. In English, 1 of 6 did.

Picture it. Each invoice is one answer. It is green if it names a firm from that country’s list, and white if it does not. Each white one here still named a firm from my international list.

Connection labelled Oslo, in English6 invoices, 1 of them green.1 of 6
Same connection, in Norwegian6 invoices, all of them green.6 of 6
Connection labelled Istanbul, in English6 invoices, all of them white.0 of 6
Same connection, in Turkish6 invoices, all of them green.6 of 6

At Mikko’s: like 6 of 6 shoppers speaking Finnish naming a local dealer, against 1 of 6 speaking English.

The same words gave other names on another connection

One Turkish question: my Turkish list in 6 of 6 answers on one connection, 1 of 6 on another

Picture it. Each parcel is one answer to the same Turkish question. The two Istanbul rows show the same 6 answers, checked against two lists, and so do the two Berlin rows. A parcel is green if it carries a name from the list in that row.

Istanbul label, my Turkish list6 parcels, all of them green.6 of 6
Berlin label, my Turkish list6 parcels, 1 of them green.1 of 6
Istanbul label, my German list6 parcels, all of them white.0 of 6
Berlin label, my German list6 parcels, 5 of them green.5 of 6

At Marta’s: like the same order sent 6 times from each of two warehouses. A Polish courier’s label is on all 6 parcels from the first and 1 of 6 from the second.

The same Turkish question brought different names on another connection. That does not prove the city caused the change.

A miss on my list can still be a local firm

On the connection labelled Tallinn, I asked in Estonian, Russian and English, and checked my Estonian list each time. My first reading put Russian in the middle, with Estonian high and English at zero. Then I read the answers that missed.

In Russian, 4 of 6 answers named a firm from my Estonian list. The other 2 named an accounting service I left out.

Picture it. Each invoice is one answer to the question in Russian. Pale green marks the 2 that name the provider I left out.

My Estonian list as I first wrote it6 invoices, 4 green and 2 white.4 of 6
Second look, including the provider I left out6 invoices, 4 green and 2 pale green.6 of 6

At Kadri’s: like 4 of 6 buyers naming a builder on her list, and the other 2 naming one she forgot to write down.

That service comes with software for Estonian limited companies. It does not fit every freelancer. So I keep both counts: 4 of 6 with my first software list, and 6 of 6 with the wider list.

How to check this for your business

Portrait of Samira, who runs a media agency.Samira BenaliMedia agency, FranceMade-up example

Samira runs ads for a small hotel in Lyon. Its guests ask robots in French, English and German. The report to the hotel counts how many French answers name a hotel on Samira’s list.

Samira asks in all three and reads every answer. The German answers keep naming a hotel missing from the list. Samira adds it and shows the hotel both counts.

Try this on Monday, in every language you serve

Take one question from your sheet in chapter 1. Have someone who speaks each language check that the question means the same thing. Ask it ten times in each language, in a new chat each time. Ask for help with any answers you cannot read.

Read every answer in full, and write each company missing from your list on the back. In each language, count how many of the ten answers name a rival on your list. Add the new names and count again. Keep both counts. This checks your counting and leaves the answers as they are.

References

Chapter 11. Same question, other language, other names

Revised manuscript. Dmitrij Żatuchin (2026). Query Language and Network Exit Context in Generated Supplier Mentions: An Exploratory Audit. Original preprint. arXiv:2608.30052v1. Raw records. 10.5281/zenodo.22181306.

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