Who this is for, and what it should answer
Useful when buyers need to compare specialist suppliers, a team is entering a new market, or existing brand claims need checking against observable answers. Agree the buyer role, product category, language and market before selecting questions.
The questions include whether the brand is mentioned, how its capabilities are described, which alternatives appear and what sources an answer lists. A listed citation is recorded separately from an independently checked source. An API answer is not evidence of what a consumer AI interface displayed.
What we need before starting
- The official website, confirmed brand names and aliases, target products, markets and languages.
- Public source pages or approved text excerpts that support important capability claims; identify disputed or confidential material.
- A shortlist of buying questions and competitors, the intended observation surfaces and any restrictions on access or reuse.
- An agreed sample scope, execution budget and reviewer. Provider credentials use the protected configuration process, never a public form.
Deliverables to specify in the scope
| Deliverable | What it contains |
|---|---|
| Frozen observation plan | Versioned questions and a matrix of language, market, surface, model or product conditions, repetitions and planned observations. |
| Original evidence | Answer text, observation time, model or product identifiers where available, listed citations, usage evidence and explicit error or uncertainty states. |
| Baseline findings | Brand and capability observations with planned, valid and missing counts. API, search-enabled API and actual product-interface observations remain separate. |
| Prioritised work plan | Issues linked to supporting observations, proposed knowledge or content changes, owners and the evidence needed for a later review. |
How the schedule is established
The sequence below is a planning example, not an elapsed-time promise or SLA. Schedule the agreed scope after the contract and required funding conditions are met. Authorise actual provider calls and their budget separately.
| Stage | Completion condition | What affects timing |
|---|---|---|
| Scope and protocol | Questions, surfaces and measurement definitions approved. | Unresolved markets, access or source permissions. |
| Preparation | Priority company facts checked and the observation matrix frozen. | Missing materials, contradictions and review turnaround. |
| Observation | Agreed samples attempted and outcomes recorded. | Provider availability, rate limits and separately arranged interface access. |
| Review and handover | Evidence and findings reviewed against the agreed acceptance checklist. | Corrections, missing observations and customer review. |
Acceptance and the current capability boundary
- Accept against the agreed questions, evidence fields and review record. Failed samples remain missing observations; they do not become negative answers or an inflated completion rate.
- The hosted workspace contains owner-triggered API-observation code and evidence storage. Real production audit acceptance is still pending; a saved connection or successful connection test is not an executed audit.
- Consumer AI interfaces, search-enabled APIs and AI search summaries each require separate access, scope and evidence confirmation. Do not describe one channel as coverage of another.
- No automatic GEO score, inclusion guarantee or causal sales claim is included. Follow-up content, publication and comparable remeasurement need their own agreed work and acceptance.