What we tracked

We built a fixed prompt set around brand discovery, category recommendations, comparisons, alternatives, pricing questions, and problem-led searches. Each prompt had a buyer stage and a reason for inclusion.

For every answer we recorded whether VisibleGen was mentioned, whether visiblegen.com was cited, which competitors appeared, the wording used to describe each brand, and the sources supporting the response.

The prompt groups

The fixed list covers branded discovery, unbranded category recommendations, comparisons, alternatives, pricing questions, and problem-led searches. We do not add or remove prompts mid-study without recording the change.

Example prompt: “Which platforms help a small SEO agency track both Google rankings and brand recommendations in AI answers?” The saved record includes the complete wording, account state, model, date, answer, citations, and competing brands.

Why mentions and citations are separate

An answer can name a brand without linking to its website. It can also cite a useful article without recommending the company behind it. The tracking sheet therefore treats mention, recommendation, citation, and description accuracy as separate fields.

MENTIONWas the brand named?

Useful for awareness and recommendation tracking.

CITATIONWas a source linked?

Useful for understanding which pages and publications shaped the answer.

POSITIONHow was the brand framed?

Useful for seeing whether the answer matched the company’s actual strengths.

How competitor appearances are logged

For each answer we record which competitors appear, the reason the assistant gives, and the sources attached to that explanation. This allows the final report to distinguish our own movement from a changing competitor set.

The review rhythm

Answers are collected daily, but interpretation waits for the weekly review. The reviewer checks repeated changes, new citations, description errors, and competitor appearances before an action is proposed.

  1. Review changes that repeated across more than one check.
  2. Open the original answers before assigning work.
  3. Group findings into content, authority, technical, local, or brand clarity.
  4. Assign one owner and an expected outcome.
  5. Keep the finding open until the next measurement window.

What will be published at day 30

The update will include the dated response log, prompt-level mention and citation fields, notable description changes, competitor overlap, cited domains, and a clear limitations section.

Until then, this page remains a public research diary—not a case study dressed up with invented results. The same evidence-first workflow is available in VisibleGen AI Visibility.