Demonstration / Illustrative Example
How an AI Visibility Case Study Is Measured
An illustrative example showing how Digital Platform 271 structures baseline measurement, implementation review and controlled AI visibility retesting.
Demonstration only: this is an illustrative example of the Digital Platform 271 AI visibility measurement process. It is not a verified client engagement and contains no client performance claims.
Start with a frozen baseline before changing the website.
This illustrative workflow begins by defining a fixed prompt set, target AI engines, scoring rules and denominators before implementation.
The purpose is to create a repeatable baseline that can later be compared with a controlled retest. No example result on this page should be read as evidence from a real client engagement.
- Freeze prompts and engine coverage before implementation
- Record mentions, recommendations, citations and entity-resolution outcomes
- Keep branded and unbranded prompts separate
- Document dates, denominators and scoring rules
Diagnose the gaps using observable website and authority evidence.
The next step is to review crawlability, entity clarity, answer architecture, structured data, proof, citations and external authority signals.
The diagnosis should distinguish what is directly observed from what is inferred, unknown or not applicable.
- Check whether the business and offer are clearly identifiable
- Review structured data against visible page evidence
- Identify missing decision-useful proof and buying information
- Record external references and authority signals separately
Implement only the agreed changes and verify them on production.
Implementation can include technical, content, entity and authority work, but each completed item should be verified on the live site before it is counted as implemented.
The workflow does not assume that every recommendation was completed or that any later movement was caused by one change.
- Improve entity and offer clarity where evidence supports it
- Make important answers explicit and extractable
- Add or correct structured data only when visible content supports it
- Verify deployment, indexability and dependencies before retesting
Report observed movement without turning correlation into causation.
A controlled retest repeats the frozen prompts and reports the results by engine, prompt type and denominator.
Mixed results should remain mixed. Improvements, declines, unchanged results, entity-resolution errors and implementation limitations all belong in the final report.
- Use the same prompts for the baseline and retest
- Report engine-level results instead of hiding variation in one headline number
- Separate observed change from causal claims
- Keep implementation status and limitations visible
Results
- Prompt set: Define → Freeze
- Baseline: Unknown → Measured
- Implementation: Prepared → Verified live
- Indexability: Unchecked → Confirmed
- Retest: Not run → Same prompts
- Reporting: Assumed → Evidence-led
- Limitations: Hidden → Disclosed
Illustrative GEO Measurement Path
This illustrative workflow moves from baseline and diagnosis through implementation verification, controlled retest and evidence-led reporting.
Real-world application
See this measurement methodology applied to Scrubsy using a frozen-prompt test across ChatGPT, Gemini and Perplexity.
Frequently Asked Questions
What is GEO?
GEO, or Generative Engine Optimization, improves how clearly AI engines understand, cite, and recommend a brand inside generated answers.
Is this a real client case study?
No. This page is an illustrative methodology example and does not present verified client performance results.
Does this replace SEO?
No. GEO builds on SEO. Crawlability, content quality, internal links, schema, and authority still matter, but the optimization target expands to AI-generated answers.
Can small brands win AI visibility?
Yes. Smaller brands can compete when their entity signals, structured content, proof, and citations are clearer than larger competitors.