Technical readiness
Access, indexing, server HTML, canonical, languages, schema and crawler policy.
Methodology
Every finding has provenance.
GEO is measured through non-deterministic systems. The useful asset is not one pretty answer but a reproducible protocol.
Three metrics
Separate facts instead of one decorative score.
Observed AI visibility
Mentions, discovery and recommendations in saved answers to real prompts.
Competitive and source gap
Who owns the answer and which independent sources help the system choose a competitor.
Protocol
How evidence is collected.
Context
Define product, buyer, market, language, competitors and the decision moment.
Security
Classify data, permitted tools, retention and NDA requirements.
Prompt map
Create real commercial questions rather than only asking the model about the brand name.
Controlled run
Save prompt, model, search mode, date, answer, citations and repeats.
Normalisation
Separate mention from recommendation, fact from judgement and brand from namesake.
Competitors and sources
Measure presence and inspect which independent sources support the answer.
Technical check
Indexing, HTML, canonical, languages, schema, crawler access and accessibility.
Action plan
Priority, expected effect, difficulty, dependencies, owner and repeat measurement date.
Evidence rules
What the report will not do.
- No figures without saved answers.
- No guarantee of a position in a specific model.
- No conclusion from a single prompt.
- No mixing brand recognition with brand recommendation.
First step
Tell us about the task.
Complete the short form. Within 24 hours the team will review the context and tell you whether we can help and where it makes sense to start.
No files. No passwords. No automated promises.Scope and commercial terms are defined after the assessment.