Selected work

Evidence first. Decisions with a clear trail.

These examples show how I frame a question, check the evidence and make the limits of a conclusion visible. Public case studies use Brand A and Brand B.

Proof pillar 03 / AI search visibility

Find the citation gap before choosing the fix.

DIRECTIONAL

The question

Could a product rank in classic search and still be missing from the independent roundups buyers use to compare tools in its category?

The method

I defined commercial queries, classified intent, checked candidate citations, opened source pages to verify what snippets suggested, compared category framing, and turned the observed gaps into a ranked action plan.

What the evidence showed

For Brand A, two commercial-intent queries did not surface the product in two category roundups. Full-page checks ruled out a search-snippet truncation issue. Existing directory descriptions also placed it in a broader category than the one represented by those buying queries.

Brand B was used as a comparison case. Its citations appeared across both query angles, with no comparable gap in this small sample.

Decision and limits

The recommended sequence was to correct category descriptions, pursue relevant roundup inclusion, then publish an owned comparison page with structured data. This was a plan, not a measured outcome. Direct ChatGPT or Perplexity API access was unavailable, so live web search was used as a proxy. Two queries per brand were checked, below the method’s usual three to five.

Basis: two queries per brand. Status: DIRECTIONAL. The audit did not measure changes after implementation.

Proof pillar 02 / campaign and customer evidence

Choose channels for the job they need to do.

The starting assumption

The initial brief focused on finding channels that could bring in more traffic. Reviewing the recommendation exposed a more fundamental distinction: a channel can add reach, or it can add credibility at a decision point.

What changed in the analysis

An early claim about the brand’s directory presence was wrong. A live check found an existing profile with no reviews. That changed the process: verify the premise first, then compare channel options against their intended role. The submitted analysis has no confirmed performance outcome yet.

Basis: channel research and a checked directory listing. Status: SUPPORTED for the research process; no channel lift claimed.