How do you create a brand-neutral questionnaire for an AI baseline?

A brand-neutral questionnaire consists of questions a customer might genuinely ask without naming your business, such as “Who can help me with …?” or “What is a good approach to …?”. It helps you measure whether an AI assistant mentions your organisation on its own, instead of prompting that answer by putting your name in the question.

Ask the same questions regularly across several AI assistants and record whether and how you appear. This gives you a baseline: a starting point for comparing later answers. The results are an indication, not a definitive verdict on your visibility. Responses can vary between assistants and from one occasion to another.

Why ask brand-neutral questions?

If you include your name in a question, you have already guided the assistant towards your organisation. You are then measuring what happens when someone knows your brand, rather than whether they can find you through a general question. With a brand-neutral prompt, the assistant can choose which organisations, sources or solutions to mention.

This makes the measurement relevant to people who are still exploring their options. You can see which topics and phrasings may lead to a mention, and which alternatives the assistant presents. A brand-neutral measurement does not tell you everything about how customers find you. It does show how your organisation appears in a particular question context.

Choose questions based on real customer queries

Do not start with a long list that only makes sense internally. Gather questions from customer conversations, sales and support enquiries, search queries, intake calls and frequently asked questions. Pay attention to the words people use themselves. A technical term may be useful if customers know it, but everyday phrasing is often a better starting point.

Include different question types. A problem question describes a need or obstacle. A comparison question asks about differences between approaches or types of providers. A choice question asks who or what might be suitable. A local question mentions a place or region when location matters. An explanatory question asks how something works or what to consider.

Choose a compact set that reflects the main customer needs and topics. Avoid adding variations that change only one word, unless that wording difference reflects a meaningful way customers search. Link each question to a topic or customer need. This makes it easier to see where you are and are not mentioned, without making the list unnecessarily large.

Write in the customer’s language

Phrase questions the way someone might naturally ask an assistant: clearly and without a hidden preference for your organisation. Avoid promotional wording, brand names and leading prompts such as “Why is my company the best choice?”. Ask about an approach, a provider or points to consider instead. Make the question specific enough to get a meaningful answer, without adding details that predetermine the result.

Review the questions with people who speak to customers. Do they immediately understand what each question means? Does the wording sound familiar from real conversations? Revise anything unclear or unnatural, then save the final versions. Do not change the set without recording it: if you revise a question, comparison with earlier measurements becomes less reliable.

Ask the same questions in the same way

Use the same question text, language and settings each time. Ask several AI assistants, since their sources, search capabilities and response behaviour may differ. Record which assistant you used and when you asked each question. Keep the conversation context consistent too: for instance, start a fresh conversation each time and avoid adding extra instructions along the way.

For every answer, record whether your organisation is mentioned, whether there is a source reference or quotation, and whether the information is accurate. Note relevant competitors, inaccurate claims and answers that mention no organisation. A simple table might include columns for the question, assistant, date, mention, source, accuracy and notes. Where practical, save the full answers so you can check how you arrived at your summary.

Repeat carefully and treat the results as an indication

Repeat the measurement at regular intervals using the same settings. An assistant may give a different answer to the same question. Personalisation, previous conversations, available sources, search features and timing can all affect the result. Changes to the assistant or to information available online may also play a part. One response therefore cannot prove that your visibility has consistently increased or decreased.

A baseline still gives you a starting point and ideas for improvement. Look for patterns: which customer questions lead to a mention, what information is accurate, and which questions mainly produce other names? If you are mentioned, check whether the reference can be verified and whether the description is correct. For more context, read about being mentioned or cited by AI and the difference between being quoted and being summarised.

Use what you learn to improve your content: answer customer questions clearly, make relevant pages easy to find, and connect related topics with internal links. An internal linking model for topic clusters can help make relationships between pages clearer. Then measure again using the same questions. Avoid changing everything at once; otherwise, it is harder to understand which change may have made a difference.

A short checklist

• Collect questions from real customer conversations, searches and support enquiries.
• Include problem, comparison, choice, local and explanatory question types where relevant.
• Phrase questions in customers’ words, without brand names or leading preferences.
• Ask the same questions in the same language and with the same settings across several assistants.
• Record mentions, sources, accuracy, competitors and notable answers.
• Repeat the measurement consistently and treat the findings as an indication.
• Improve your content based on recurring patterns, then compare again.