V6, Wave 2 · Methods preview
Methods before headlines
The design aims to separate observed AI visibility from the way answers are collected. This page describes intended reporting; it does not certify a completed protocol or collection run.
Status of these methods
Design notes can change before the final protocol is locked. The final methods, deviations and completion report must accompany the published findings.
Planned measurement and reporting
- Panel and sampling. Revisit the cleaned earlier business panel. Disclose its discovery procedure, exclusions, replacements and the matched panel available for comparison. Report any creator arm separately.
- Collection routes. Identify consumer surfaces and API conditions separately, with search settings, collection dates and session conditions. Do not pool them into a single visibility headline.
- Prompts and repeats. Retain intent families, exact wording, paraphrases, repeats and time blocks. Distinguish repeat variability from sensitivity to the prompt menu.
- Outcome states. Separate valid no-mention answers, no-trigger AI Overviews, refusals, collection failures and unresolved entity matches. Publish completion and unknown rates.
- Dependence and uncertainty. Preserve shared-answer and prompt-family dependence in paired comparisons. Entity labels from the same answer are not independent generations.
- Comparison across waves. Account for changing prompts, opportunities and surfaces before interpreting a change as a business becoming more or less visible.
- Scope of inference. Describe a benchmark presence rate for the tested menu. Zero observations do not establish that AI cannot recognize a business. Observational associations do not establish causality.
Fields awaiting the final research record
| Research record | Status |
|---|---|
| Protocol URL, timestamp and version | [Pending final research review] |
| Completed business panel, markets and exclusions | [Pending final research review] |
| Exact surfaces, settings and collection dates | [Pending final research review] |
| Prompt menu, repeats and paired time blocks | [Pending final research review] |
| Scheduled, completed, failed and unresolved measurements | [Pending final research review] |
| Entity matching audit and error rates | [Pending final research review] |
| Estimands, uncertainty and sensitivity analyses | [Pending final research review] |
| Protocol deviations, funding and conflicts | [Pending final research review] |
| Paper, citation and aggregate release links | [Pending final research review] |
V6, Wave 2 has no published results on this site yet. The registered Phase 3 trial from the earlier program never launched; it must not be represented as causal evidence for this wave.
Publication and disclosure
Joel House founded Outrigger, whose commercial interests overlap with these research questions. Final disclosures and any release license are pending. No business-level dataset access is offered by this preview.
Read the published program’s methodology · Press kit preview
See where your brand stands
Start with what AI says about your business.
See what AI says about youSee pricing