Lock the question and environment
Core observations use non-brand buyer questions. Brand-fit and competitor questions remain diagnostic and separate.
GEO METHODOLOGY · GOOGLE AI SEARCH
Every conclusion
must return to evidence.
We treat an AI-search answer as an observation produced under a specific question and environment. Comparison becomes meaningful only when the question set, market, language, time, source state and human QA are recorded.
DIRECT ANSWER · CURRENT SCOPE
How does CLEARCRAFT measure AI search visibility?
Before observation, CLEARCRAFT fixes the non-brand buyer questions, market, language, platform, date and available settings. It preserves raw answers, cited URLs, no-answer states, errors, unknown sources and off-topic responses. Only comparable, complete and human-reviewed observations enter the core interpretation; platform observations, search performance, website activity and commercial outcomes remain separate evidence sets.
EVIDENCE PATHRead the baseline and human-review method →
Boundary: The result is a reviewable observation under stated conditions, not an overall AI ranking, long-term trend or proof that a content change caused an outcome.
Retain raw evidence
Keep the timestamp, provider or surface, raw answer, cited URL or unknown source state and QA record.
Treat exceptions as results
No answer, errors, incomplete citations, format failures and off-topic responses remain visible and outside core KPIs.
Review before interpretation
People verify question intent, source status, entity identification and completeness before a finding is used.
WHAT THE RECORD SUPPORTS
A reviewable record,
not a marketing slogan.
Readable output
Questions, environment, raw answers, citation states, human QA, exceptions and possible actions are presented separately.
Data limitations
An answer may not exist; a source can be unknown; and platforms change. A snapshot cannot prove ranking, causation or a long-term trend.
Approval responsibility
Human QA precedes any KPI or public claim. Clients confirm their own facts and regulated language.
BUYER QUESTIONS
Know what the data
can and cannot say.
What can one observation tell us?
It is a snapshot of a question, environment and source state. A comparison needs repeated observations under the same locked configuration.
What stays outside a KPI?
No answer, system errors, unknown or incomplete sources, format anomalies, off-topic content and pending reviews. They are not zero scores.
Are Google AI, Gemini and other platforms combined?
No. Each surface, provider and source environment is recorded and interpreted separately.
USE THE METHOD
Design the observation first,
then discuss improvement.
AI tools support the workflow but do not replace evidence capture, review or accountable approval.
View service deliverables