The short version

We will ask ChatGPT 100 buying questions and record which businesses it recommends and which sources it uses to support those answers. We will not publish a ranking or any percentages until every answer has been saved and reviewed.

The question we will try to answer: when a buyer adds a location, audience, budget or specialist need, what will change in the shortlist, and what public material will appear beside each recommendation?

How we will build the 100 prompts

We will write ten buying situations and ten natural-language versions of each. They will range from broad questions, such as “What are the best options?”, to decision-stage requests, such as “Which provider is best for a small team that needs monthly client reporting?”

  1. We will separate branded from non-branded prompts. A brand question will measure recognition; a category question will measure discovery.
  2. We will keep the buyer’s need visible. Location, company size, budget and desired outcome will be part of the prompt where they matter.
  3. We will record the full answer. We will capture the businesses named, their order, the wording, linked sources, unlinked mentions and the reasons given.
  4. We will code how certain the answer sounds. “Consider”, “a strong option” and “the best choice” will be recorded as different levels.
  5. We will repeat the questions over time. We will treat a single answer as one observation, not a trend.

What we will record for each answer

For every prompt we will record the businesses returned, their order, the reason given, the cited domains and the date. We will record a mention, a direct recommendation and a linked citation as separate fields.

We will report patterns only if the reviewed answers support them, and we will publish the rows so readers can check them.

FIELD 01Category clarity

Will the answer explain what the business does and who it serves?

FIELD 02Independent sources

Will other sites, customers or industry sources describe the business the same way?

FIELD 03Answer-ready pages

Will the cited pages answer comparison, suitability, location, pricing and outcome questions directly?

The questions the test will check

1. Will a clear category help?

We will check whether businesses that state a specific category, audience and job appear more often than businesses with vague descriptions.

2. Will other sources matter?

We will check whether reviews, news coverage, association pages, comparison articles, community discussions and directories appear beside recommendations more often than company-owned pages alone.

3. Will matching proof matter?

We will check whether the evidence cited matches the buyer’s question, for example agency features for an agency question.

4. Will consistent details matter?

We will check whether consistent names, locations, services and pricing language across sites go together with more recommendations.

What we will measure

We will measure five things for every answer: whether the business is recommended, whether its site is cited, its position, the reason given and the type of source. We will add competitor overlap and change over time, and we will keep the original answer beside every score.

When we will publish

We will publish the method, the reviewed answers and their limits together, and only after the review is complete. Until then, this page will describe the plan only. The same workflow will be available in VisibleGen’s AI Visibility Tracking and AI Source Opportunities.