DIY test: 10 questions to ask ChatGPT about your company today

You do not need an assessment to get a first signal: ten minutes and an AI assistant are enough. Open ChatGPT (preferably with web search on), Perplexity or Gemini, paste the questions replacing the brackets, and note down the answers. At the end you will find how to read them.

The ten questions

1. What do you know about [Company]? The base question. Note: year and facts cited, any errors, tone.
2. What is it like to work at [Company]? The candidates' point of view. Note whether reviews appear, from which year, and whether benefits, training and growth are mentioned.
3. Is [Company] a reliable supplier? What are its weaknesses? The buyer's question. Note the "weaknesses": they are often dated narratives.
4. Which are the best [industry] companies in [country]? The category question. Note whether your name appears and who appears in your place.
5. Which [product or service] should I choose for [typical need of your customer]? The recommendation. Same exercise as the previous one.
6. [Company] or [main competitor]: which one should I choose and why? The comparison. Note the arguments used for and against.
7. Who is [name of the CEO or founder]? The person entity. Note namesakes, wrong roles, old information.
8. Have there been controversies or problems linked to [Company]? The risk question. Note what is cited and with which date.
9. What has changed at [Company] in the last two years? Freshness. If the answer is vague or empty, recent sources are missing.
10. Which sources did you use to answer? Ask it after each of the previous ones. The sources are the part that counts: they are where you act.

How to read the answers

• Absence in questions 4 and 5 is the most important signal: the AI recommends your category without naming you. It is a problem of third-party sources (directories, rankings, communities), not of your website.
• Old facts in questions 1, 3 and 8 point to dated narratives: there is a source (often Wikipedia or an article) more authoritative than everything you have published since.
• Different answers across engines are normal: they read different sources. If one engine knows you and another does not, the problem is the coverage of the sources that engine uses (for ChatGPT and Copilot, Bing).
• Cited sources you do not look after are your priority list.

The limits of the test

The test shows one answer per question, at one moment; an assessment repeats every question several times on five engines, traces every source with URL and weight, and compares with competitors. If the test showed you something you did not expect, the free audit is the next step: we present it in 48 hours.