Priority studies

These protocols are being built in public. A study is not labeled complete until its sample, dates, method, evidence package, and limits are published.

Protocol 01 · planned

Public bot-access benchmark

Test a defined sample of AEO and SEO sites across documented search, user-fetch, and training crawler paths. Report status, robots, challenge behavior, static content, and sitemap health.

Protocol 02 · baseline next

Fixed-prompt citation lab

Run a frozen question panel across named answer products, with repetitions, exact outputs, citations, correctness review, geography, and uncertainty.

Protocol 03 · planned

Utah answer landscape

Build the candidate universe independently of Bob, then examine how local high-intent questions are answered and which sources support the results.

Protocol 04 · ongoing

TrainingPlan.dev follow-up

Extend the fixed September 4 baseline with sanitized aggregates, link observations, index cohorts, AI citation trials, conversion evidence, and negative results.

Read the current case →

Protocol 05 · planned

Page-feature citation study

Compare consenting page cohorts with visible methods, primary citations, direct answers, original statistics, and controlled information structures.

Protocol 06 · monitoring

llms.txt adoption

Measure requests and downstream citation outcomes descriptively. Correlation will not be reported as proof that the file caused visibility.

Research release standard

  1. Prestate the question, sample, measures, scoring, and exclusions.
  2. Archive dates, exact inputs, public sources, code, and permitted outputs.
  3. Separate descriptive observation from causal inference.
  4. Report missing, null, negative, and contradictory outcomes.
  5. Disclose ownership, commercial incentives, and reviewer conflicts.
  6. Version the release and correct material errors publicly.

See the complete site methodology and disclosures.