This article documents Phase 1 of an ongoing Generative Engine Optimization (GEO) and Answer Engine Optimization (AEO) testing project.
The goal of this test is not to force visibility inside AI systems or to manipulate responses.
Instead, the purpose is to observe how and when a website becomes eligible to be mentioned at all in AI-generated recommendations.
Phase 1 focuses on a single question:
What minimum signals are required for a new or less-established website to enter the AI recommendation space?
What Was Done on the Website (Phase 1 Setup)
Phase 1 was not passive.
Before testing any AI responses, the website was intentionally prepared to provide clear, foundational signals, without artificial amplification.
The following changes were made:
- A core informational page was created to clearly explain the service and answer the most important user questions
- An About Us page was added to establish basic identity and legitimacy
- Content was written in a neutral, explanatory tone, without promotional or sales language
- The website structure was kept simple and easy to interpret
- No backlinks were built
- No authority signals were artificially added
- No prompt repetition or manipulation was used during testing
The goal was to make the website understandable and complete, not popular.
What Was Tested
The test focused on recommendation-style prompts, such as:
- “Best driving schools in my city”
- “Recommended driving schools near me”
The intent was to observe:
- how it is positioned relative to more established entities
- whether the website appears at all
- under which conditions it is introduced
Phase 1 Observations
1. No Direct Inclusion in Initial Answers
In the first round of responses, the driving school did not appear in the initial list of recommendations.
This outcome was expected.
AI systems tend to prioritize:
- long-established entities
- frequently referenced businesses
- options with stronger historical signals
Early exclusion does not indicate rejection — it reflects confidence weighting.
2. Inclusion After Extending the Recommendation List
When the system was asked to extend the list with additional recommendations, the website was introduced as an option.
This distinction is important.
The business was not treated as a primary recommendation, but it was considered eligible once the system moved beyond the most dominant entities.
This suggests that inclusion may occur in layers, rather than as an all-or-nothing outcome.
3. What This Suggests About AI Recommendation Behavior
Based on Phase 1 observations, AI recommendation systems appear to operate with at least two layers:
- a primary layer of high-confidence, well-established entities
- a secondary layer of acceptable alternatives that meet baseline clarity and legitimacy thresholds
Visibility, at least in early stages, appears to be gradual rather than binary.
What Phase 1 Does Not Show
It is important to be explicit about limitations.
Phase 1 does not demonstrate:
- strong authority
- primary recommendation status
- consistent top placement
Early inclusion does not equal dominance.
This phase only confirms eligibility, not preference.
Why Phase 1 Still Matters
Even without direct inclusion in the first response, appearing in extended recommendations confirms one critical point:
The system recognizes the website as a valid and usable entity.
For new or less-established local businesses, this appears to be the first measurable threshold in AI-driven recommendation systems.
What Comes Next (Phase 2)
In phase 2, I will focus on:
- strengthening contextual clarity
- reinforcing entity consistency
- observing whether inclusion moves closer to primary recommendations
No conclusions will be drawn until multiple phases are completed.
Closing Thought
GEO testing is not about instant visibility.
It is about understanding how systems decide who is eligible to be mentioned at all.
Phase 1 shows that basic clarity, completeness, and legitimacy can be sufficient to enter the extended answer space — even before authority is established.
Leave a Reply