Buyer-question worksheet for your AI visibility scan
Use this worksheet to turn the problems you solve into saved discovery questions that a potential client might actually ask.
On this page
Good scan questions sound like a person looking for help. They do not need to sound like keywords, and they should not contain a disguised version of your biography.
This worksheet helps you turn the work you provide into a small set of discovery questions. A discovery question leaves your name out. It asks who can help with a problem, decision, or project that a potential client might face.
The output is a saved question set you can inspect and reuse. It is not evidence of search demand or a prediction of how clients will phrase every request. You can download the worksheet and fill it in away from the browser.
Start with the job a client needs done
Write down one area of work you want to be hired for. Use ordinary language, not a service-menu label.
| Prompt | Your notes |
|---|---|
| What work do I provide? | [for example: review a paid-search account] |
| What situation makes someone seek this help? | [for example: leads are expensive and poor quality] |
| Who is usually making the decision? | [for example: a B2B software marketing lead] |
| What context genuinely affects fit? | [industry, location, language, regulation, business model] |
| What work would I decline? | [boundary to keep out of the question set] |
Do not copy the example into your own worksheet unless it describes your real work. The detail that matters is the one a client who has never heard of you would use to decide whom to approach.
If you have recent enquiries, use the non-confidential part of the language. Remove names, financial details, and anything covered by a confidentiality obligation. If you do not have enquiries yet, label the question as an assumption based on the work you can honestly provide.
Turn the problem into a first question
Try this sentence shape:
Which [kind of independent expert] can help [kind of client] with [specific problem or outcome]?
Here are fictional examples of the shape, not researched high-volume prompts:
- Which independent consultants can help a US B2B software company diagnose poor lead quality from Google Ads?
- Who can help a small architecture studio document its client onboarding process?
- Which independent researchers can help a nonprofit plan user interviews before redesigning a service?
The question should identify the work and, where it matters, the client context. It should not pack in your former employer, exact town, obscure credential, and the name of a past client. Those details can make you an obvious match while telling you little about a normal discovery question.
Build five questions with different jobs
One narrow service can produce several reasonable questions. They should test different situations, not rearrange the same phrase five times.
Use these five prompts. Leave any blank when it does not fit your work.
- Direct service: Which [expert] can help [client] with [core service]?
- Pain or diagnosis: Who can help [client] understand or fix [specific problem]?
- Decision point: Which [expert] can advise [client] before [important decision or change]?
- Second opinion: Who can review [project, account, plan, or process] for [client]?
- Implementation support: Which [expert] can help [client] put [specific change] into practice?
For each candidate, write a short reason it belongs in your set. If the reason is “it would probably find me,” cross it out. A useful reason is “a prospective client could plausibly ask this before they know my name.”
Check the question against a fresh reader
Read each question without looking at your website. Then ask:
- Does it name a problem or decision I can genuinely help with?
- Does it describe an individual expert if I sell my own expertise rather than an agency service?
- Would a client who does not know me use these details?
- Does it avoid my name, website, previous employer, and private client details?
- Is the location or language included only because it changes who is relevant?
- Would I be happy to receive an enquiry from someone asking this?
Questions with your name still have a place. “What does [name] do?” can reveal a confused biography or old association. Keep that in a separate identity check. It does not belong in the discovery baseline because the answer has been given the identity you hope to observe.
Read the full question selection guide for more examples and tradeoffs.
Save the scope with the questions
The wording alone is not enough for a later comparison. Record the conditions alongside each question:
| Field | Save this |
|---|---|
| Question | Exact text, including location or language details |
| Purpose | Direct service, pain, decision, second opinion, or implementation |
| Locale and language | The setting used for the check |
| Engines | The engines included in the scan |
| Date | When you ran the question |
| Notes | Any legitimate scope detail that affects interpretation |
Gistful saves question and scan scope for explicit rescans. Whether you keep your own worksheet or use the product, do not silently rewrite the question set when you update your profile. Save the original set first. A changed question deserves a new baseline.
Run, read, then revise carefully
Once you have five candidates, run the questions and read the returned answers. Check whether an apparent name is really you. Record a completed answer with no appearance as an observation, and record a failed or missing answer as unmeasured.
Do not rewrite questions just because you do not appear on the first run. First confirm that the question reflects a real service you provide. Then inspect the names and sources that did appear. Why AI recommends other experts instead of you gives you a way to review the evidence without pretending to know the engine’s hidden reasons.
If you later change the question, label it as a new baseline. If you keep it, you can compare a later scan under the same scope. Use the measurement guide to make that comparison honest.
Your finished worksheet
Before you run a scan, make sure you can answer yes to each point:
- I have up to five questions that represent different client situations.
- Every question describes work I actually offer.
- No discovery question supplies my name or a disguised biography.
- I have recorded locale, language, engine scope, and date.
- I know which questions are assumptions rather than language taken from real enquiries.
- I will save the returned answers and sources, not only a score.
That is enough for a useful first set. Add questions later only when you can name a distinct service or decision that the original set does not cover.