This page shows the structure of a decision-grade AI Visibility report. It is an anonymized methodology example. It does not publish fabricated metrics, imply a client relationship, or claim that a sample observation represents permanent AI visibility.
Executive report structure
- Executive summary: material representation risks, decisions required, and the most important next action.
- Scope and boundaries: entities, markets, languages, providers, surfaces, period, and exclusions.
- Query register: exact questions, buyer roles, decision stages, proof requirements, and priority.
- Observation environment: provider, visible model or mode, date, time, timezone, language, location context, session, and retrieval status.
- Entity findings: recognition, naming, category, services, locations, experts, relationships, and ambiguity.
- Answer findings: completeness, directness, accuracy, omissions, outdated statements, and misinformation.
- Source and citation findings: cited sources, owned-source use, third-party dependence, and source conflicts.
- Competitor and recommendation context: which alternatives appear, why they may be considered, and what cannot be inferred.
- Knowledge and evidence gaps: missing canonical answers, claims, proof, relationships, or governance.
- Roadmap: prioritized corrections, implementation dependencies, acceptance criteria, and monitoring plan.
- Confidence and limitations: what is observed, inferred, unknown, or pending human review.
Illustrative finding record
| Record field | Sample treatment |
|---|---|
| Observation ID | SAMPLE-OBS-001 |
| Entity | Illustrative organization, not a named client |
| Exact query | Recorded in the project query register |
| Provider and mode | Recorded at the time of observation |
| Date and timezone | Required, using Asia/Jakarta where applicable |
| Observation status | Observed, not observed, not measurable, provider failure, empty output, retrieval failure, or needs human review |
| Interpretation | A bounded analyst interpretation, not an absolute truth |
| Confidence | Stated with the evidence available |
| Limitation | Provider, session, wording, retrieval, location, source availability, and time |
Evidence layers must remain separate
| Evidence layer | Meaning |
|---|---|
| Observed evidence | What an AI system returned in a defined environment and time. |
| Implementation evidence | What page, schema, source, relationship, or control was actually changed. |
| Outcome evidence | A comparable change measured against a valid baseline. |
| Independent validation | Relevant evidence produced by an independent third party. |
| Methodology evidence | How the query, observation, classification, and validation were performed. |
Failure handling
A provider failure, blocked response, empty output, or retrieval failure must not be counted as brand absence. The report records the failure state and excludes it from any denominator where inclusion would distort the result.
What the report cannot claim by itself
- Permanent ranking, citation, inclusion, or recommendation.
- Universal results across users, locations, accounts, models, or query wording.
- Causal improvement in leads, sales, conversion, revenue, or market share without valid outcome evidence.
- A client relationship when the material is only a public observation or methodology example.
- Independent validation when the work was reviewed only by Undercover.co.id.
Related routes
Review the measurement methodology, evidence architecture, and monitoring program. A company-specific report requires an approved scope and observation environment.