The scan captured enough evidence to support the finding.
Use the finding to prioritize work.Trust & methodology
Know what the score saw. Know where certainty stops.
Gemmetric connects AI Visibility findings to observable evidence, keeps access and uncertainty visible, and gives teams a traceable path from scan to next move.
Results are directional and bound to the evidence captured in a specific scan. They are not guarantees of model behavior or business outcomes.
Inspectable evidenceVisible uncertaintyComparable re-scans
Evidence chainInspect the reasoning
- 01ScopeWhat was evaluated
- 02EvidenceWhat the scan captured
- 03BoundaryWhat remains unknown
- 04MeaningWhat the signals support
- 05Next moveWhat to improve first
Evidence boundaryNot verified is not the same as absent.Access limits and unavailable sources stay visible.
Re-scan comparable evidence after changes go live ↻
The standard
A useful result should survive the question: Why?
If a team cannot see what was scanned, what evidence supports the conclusion, or where the method reached its limit, the output is difficult to defend. Gemmetric is designed to make that reasoning inspectable.
- 01
Inspectable
A result should point back to the page, source, response, or scan evidence that supports it.
- 02
Bounded
The method should show what was evaluated, what was not verified, and what could not be accessed.
- 03
Comparable
History is useful when teams can compare like with like and interpret movement across repeated scans.
- 04
Traceable
Priorities and recommendations should follow from the evidence instead of appearing as black-box advice.
Evidence before advice
Every decision needs a visible chain from scope to action.
The chain begins with what the scan could evaluate. Evidence and its limits shape the finding; the finding shapes the priority; a later comparable scan tests whether the picture changed.
- 01ScopeWhat was evaluated
- 02EvidenceWhat the scan captured
- 03BoundaryWhat remains unknown
- 04MeaningWhat the signals support
- 05Next moveWhat to improve first
Comparable re-scanDid the evidence move after the change?
Repeated movement matters more than one isolated swing.
Evidence boundaries
Unknown, unavailable, and absent are not interchangeable.
A credible method does not turn a collection problem into a business failure. The evidence state changes how a finding should be interpreted. It also shows whether a team should act, investigate, or wait for a better scan.
Evidence stateHow to interpret itWhat to do
The selected scan did not establish the fact either way.
Treat it as an open question, not a failure.Access, source, or collection limits prevented evaluation.
Disclose the limit and avoid inferring absence.Progress with context
A score move matters when the evidence explains it.
History should do more than draw a line. Teams need to know whether the scans are comparable, which underlying signals changed, and whether the movement persists after changes go live.
Comparable historyNorthstar Supply
Illustrative example, not customer resultsWhat changedThe business definition became clearer and more independent sources supported it.
What did not changeThe score still does not guarantee how every AI system will answer.
Method boundaries
What the evidence can support and what it cannot promise.
These limits are part of the methodology, not fine print. They protect teams from over-reading a scan and keep AI Visibility work grounded in what can actually be observed.
A scan has a defined scope
It reflects the domain, sources, and access available when it ran. It cannot capture every possible signal on the internet.
Model behavior can change
AI Perception records observed behavior for a scan. It is not a promise that every model, prompt, or future answer will behave the same way.
Unavailable is not absent
If access or collection fails, the honest result is an evidence limit. Gemmetric should not convert uncertainty into a negative fact.
Scores are directional
Scores help compare signal strength and find priorities. They do not guarantee rankings, citations, recommendations, traffic, or revenue.
Authorization matters
Deeper managed-domain workflows are intended for sites a customer owns or is authorized to manage.
Repeated movement matters
One score change can be noisy. Comparable movement across scans, supported by changed evidence, creates a stronger progress story.
Related transparency pages
Methodology questions
How to read a Gemmetric result responsibly.
Straight answers about inspectability, uncertainty, score movement, outcome claims, and the role of re-scanning.
Can I see why a Gemmetric score or finding exists?
Gemmetric is designed to connect scores and findings to inspectable scan evidence. The available detail depends on the workflow and plan, but the method is built around showing what was observed rather than asking users to trust an unexplained number.
What happens when Gemmetric cannot access or verify a signal?
The limit should remain visible. A blocked or unavailable source is not treated as proof that the signal is absent, and a partial scan should not be presented as complete evidence.
Why can an AI Visibility Score change between scans?
The website, outside sources, scan access, and observed model behavior can all change. The useful question is not only whether the number moved, but which evidence changed and whether the scans are comparable.
Does Gemmetric guarantee that AI systems will cite or recommend a business?
No. Gemmetric measures directional signals and observed behavior. External models, queries, sources, and answer contexts remain outside Gemmetric's control.
How should teams use the methodology?
Use it as a repeatable decision process: define the scan scope, inspect the evidence, respect what the scan did and did not verify, choose a supported next move, and re-scan after changes are live.
Start with the evidence your public site makes available.
Run a lightweight GEO-first Snapshot, or explore the full workflow when your team needs deeper evidence, prioritized work, and comparable history.