Choose one important service and several different customer needs: selecting a provider, understanding terms, and comparing options. Phrase the questions in everyday language. If you want to find out whether an AI tool will name the company on its own, leave the company name out of the question.
In reChecker, you can save from 1 to 5 unique questions, up to 250 characters each. They are used for the next scheduled check. Start with a few precise formulations whose answers you are prepared to review.
Start with a real customer need
Choose one priority service and describe what the person wants to achieve. For example, the owner of a small office needs ventilation maintenance: they are choosing a provider, finding out what the work includes, and comparing maintenance options.
This framing includes the need, type of property, and relevant constraints. It does not require promotional language about your own company. First, phrase the question without the brand name and check that it can be answered meaningfully.
Use real customer conversations if you have the right to analyze them. Remove personal data and sensitive circumstances. A general situation is enough for a sample set; do not include a private address, customer name, or contract details in the questions.
Distribute questions across different needs
Do not devote the whole set to one question about the “best company.” Choose different stages of the decision: finding a provider, selection criteria, a specific service, comparing approaches, and asking about brand facts.
| Sample intent | Possible question | What to look for in the answer |
|---|---|---|
| Choose a provider | Who maintains office ventilation in the relevant city? | Whether suitable companies are named |
| Understand terms | What should I ask a contractor before signing a maintenance agreement? | Relevance to the actual service |
| Find a specialization | Who works with small office properties? | Fit with the target audience |
| Compare options | What is the difference between a one-time repair and maintenance? | Context for choosing a solution |
| Check the brand | What services does the selected company provide? | Accuracy of the information |
Avoid prompting the result
The question “Why is our company the best?” already dictates an evaluation. It tests whether the model will continue the premise, not whether the company name appears independently in the answer. Use neutral wording for monitoring.
Do not include your domain in every question if your goal is to see spontaneous mentions. Otherwise, a positive result may reflect the question itself. Record separately which questions include the name intentionally and why.
Also, do not ask for an unverifiable ranking that guarantees a choice. An AI response may be incomplete, incorrect, or depend on the wording. Read any mention of the company together with the text, the company’s role, and whether it matches the customer’s requirements.
Specify the relevant location
If a service is limited to a city, include it in the question. Without a location, the answer may refer to another area. But include only conditions that genuinely matter to the customer and are supported by your offering.
Questions about urgent callouts, round-the-clock service, or specialized equipment should reflect the company’s actual capabilities. Do not add these attributes just to make the brand appear if the business does not offer them.
For international services, you can treat language and location as separate conditions. Five questions will not cover every combination. Choose the priorities and explain to your team which demand segment is not being monitored.
Check the length of each question
Each question must be no longer than 250 characters. Do not try to fit a complete technical specification into it. Keep the key situation and one clear question. Additional requirements may make the answers too varied to compare.
Read all five questions in a row. If two differ only by an adjective, consider what distinct conclusion each one provides. Textual uniqueness is technically necessary, but the usefulness of the set depends on the differences in intent.
To move from search keywords to customer situations, see the guide to choosing keywords. A list of search phrases helps identify topics, but an AI question usually needs clear context and natural, conversational phrasing.
Save the questions before checking
In the “Search positions” section, click “Set up questions,” enter the questions, and click “Save questions.” Saving the settings does not start a new check.
Before changing the questions, record the current wording and the date. You can do this in your team’s working document. This gives the next monthly results clear context, even without a built-in archive of question versions.
Saving a new list affects subsequent checks. Refreshing the report page does not mean that all services will be checked again immediately. Do not tell a colleague to expect new answers as soon as the form is edited.
The monitoring includes ChatGPT, Alice, DeepSeek, Gemini, Claude, Grok, and Perplexity. Scheduled mention checks are monthly; the availability of individual responses may vary. Read the actual report data rather than treating the expected total as a result already obtained.
More information: Website semantic core: how to build and use it for SEO.
Check results are available in the website monitoring dashboard.