FICTIONAL SAMPLEEvery company, domain, observation and result below is synthetic. No AI provider was queried and no real organisation was assessed. This page demonstrates the structure and evidence boundaries of a SemanticRisk Unified Report.
Unified Report · Compare & Monitor example
Northstar Analytics
A worked example showing how SemanticRisk separates technical access, buyer-question visibility, interpretation accuracy, competitive displacement and measured change.
01 · Executive meaning
Reachable and generally understood—rarely surfaced.
Action required
The fictional company is technically reachable and is interpreted accurately when named, but it appears in only 9 of 40 neutral buyer-question observations. Beacon Metrics appears more often and owns most third-party citations.
Decision: Strengthen public evidence for the two buyer tasks where Northstar is absent, then repeat the same compatible matrix before claiming improvement.
How to read this: technical readiness, visibility and interpretation are separate measurements. A reachable website can still be absent from buyer answers; a visible company can still be described inaccurately.
02 · Measurement snapshot
What this cycle observed
This compact demonstration uses 40 questions with one synthetic observation per question. Live report totals equal questions × provider lanes × repetitions; for example, 40 questions across two lanes produce 80 observations. Counts here are not market-share estimates or guarantees about future AI responses.
Illustrative observations completed40 / 40Controlled illustrative observations
Target surfaced9 / 4022.5% observation coverage
Owned-domain citations3Captured supporting citations
Claims supported5 / 8Against synthetic current evidence
03 · Access and readiness
AI-oriented technical evidence
Observed checks
- PassHomepage reachable
Synthetic HTTP 200 observation.
- Passrobots.txt
Synthetic policy permits evaluated crawlers.
- Warningsitemap.xml
Three product-support pages are omitted.
- Observedllms.txt
Present; downstream use is not claimed.
- WarningOrganisation identity
Schema lacks sameAs and product relationships.
What this means
Northstar's synthetic public site is accessible enough to evaluate. The warnings may weaken discovery or entity resolution, but they do not by themselves explain every visibility result.
Causal boundary: helper files and schema can improve machine-readable evidence. Their presence does not prove that a specific AI provider used them.
04 · Buyer-question visibility
Who appeared in the controlled answer set?
Observed mentions
A single answer may mention more than one company. Counts therefore do not need to total 40.
Interpretation
Northstar appears for technical-evaluation questions but is usually absent from procurement and governance questions. Beacon's synthetic third-party evidence is cited more consistently.
Not measured: consumer-chat market share, revenue impact, hidden training data, or universal recommendation likelihood.
05 · Interpretation assessment
What the evaluated answers got right—and wrong
Supported · 5
Category, primary capability, deployment model, security documentation and headquarters align with the synthetic public evidence.
Partially supported · 2
Two capability statements are broader than the wording on the fictional product pages.
Contradicted · 1
The answer describes Northstar as enterprise-only, while its fictional current pricing includes self-service.
Assessment rule: unsupported is not automatically false, and an uncited statement is not automatically a hallucination. Stronger language requires stronger stored evidence.
06 · Change since compatible cycle
Movement was observed; causation was not.
| Measure | Previous cycle | Current cycle | Observed change |
|---|
| Target mentions | 4 of 40 | 9 of 40 | +5 observations |
| Owned-domain citations | 1 | 3 | +2 citations |
| Supported claims | 4 of 8 | 5 of 8 | +1 supported claim |
Change boundary: these are associations between two compatible synthetic cycles. This sample does not claim a website edit caused the movement. Exact answers, timestamps, model identity and methodology compatibility would remain available in a real evidence appendix.
07 · Prioritised actions
Do this, then verify it
1
ActionPublish one evidence-led page for the missing procurement buyer task.
Why nowNorthstar was absent in all eight synthetic procurement-oriented observations.
VerificationRepeat the same eight questions in the next compatible cycle.
2
ActionCorrect the enterprise-only ambiguity across pricing, product and organisation schema.
Why nowOne material interpretation conflicts with the synthetic current offer.
VerificationRe-run the branded interpretation probe and preserve its answer and sources.
3
ActionAdd omitted support pages to the sitemap and monitor discovery.
Why nowThree relevant fictional pages are public but absent from sitemap evidence.
VerificationConfirm sitemap inclusion, retrieval and later citation movement separately.
08 · Monitoring watchlist
What the next cycle should answer
- Does Beacon Metrics continue to dominate procurement-oriented questions?
- Does the enterprise-only description persist after the evidence is clarified?
- Do owned-domain citations increase without reducing third-party corroboration?
09 · Method and limits
Evidence before certainty
Included in this demonstration
- Access and crawl-readiness observations
- Neutral buyer-question visibility
- Competitor displacement
- Claim-level interpretation assessment
- Compatible-cycle change
- Prioritised actions and verification steps
Never promised
- Control over downstream AI systems
- Knowledge of hidden training or retrieval
- Guaranteed recommendation or citation
- Causation from before-and-after association
- Search ranking, market share or revenue guarantees