Part 1: What is GEO?

A lot of people are talking about GEO right now. Generative Engine Optimization. AI visibility. Brand presence in AI answers.

Most of what I see in the market misses the mark.

Over the next posts, I will share concrete learnings from work I have been doing at Harvard Business School and through GEO Advisor. How to measure GEO properly, which metrics actually matter, and how to move from interesting data to clear actions.

Let’s start with why most GEO efforts currently fail.

Why most GEO audits are not actually measurement

I have seen marketers run 10 queries in ChatGPT, take screenshots, call it a GEO audit, and move on.

That is not measurement. That is a vibe check.

AI responses vary wildly.

• Same query, slightly different wording, different answer

• Same query, different implied persona, different recommendations

• Same query, different day, different result

One query tells you nothing.
Ten queries tell you almost nothing.

If you want to understand how your brand actually appears in AI systems, you need statistical volume. Patterns only emerge when you test at scale.

Visibility is not binary

Most early GEO efforts stop at a single question.

Is my brand mentioned or not?

That is table stakes. The real questions are:

• What exactly does the model say about your brand?

• How are you framed when compared head to head with competitors?

• What drives positive versus negative sentiment?

That last question is where most GEO work breaks down.

Sentiment by itself is not useful unless you understand why the model is expressing it.

From mentions to drivers

When we analyze AI responses at scale at GEO Advisor, we do not just log mentions. We extract the drivers behind those mentions.

For example:

• Pricing complexity

• Time to value

• Integration depth

• Product scope clarity

• Target customer fit

These drivers are what actually shape AI recommendations.

This is where GEO becomes actionable. Instead of saying sentiment is negative, you can say sentiment is negative because pricing limits surface repeatedly in automation related queries. That is something a product, marketing, or pricing team can fix.

What GEO Advisor focuses on

At GEO Advisor, we are building systems that:

• Run hundreds of AI queries per category

• Normalize responses across models and prompts

• Extract sentiment and the drivers behind it

• Turn that into clear recommendations, not dashboards

No theory. No screenshots. Just what actually works when you want to influence how AI talks about your product.

This is Part 1.

Next, I will break down the core GEO metrics that actually matter and why most teams track the wrong ones.

If you are working on AI visibility and want to go deeper, you are in the right place.