Why personalization distorts your audit
When you run AI engine queries while logged into your Google, ChatGPT, or Perplexity account, you are testing a personalized view of the engine's output. The engine knows your search history, your location, your preferences, and in many cases the businesses you have previously researched or visited.
If you have ever searched for your own business, visited your own website, or interacted with your Google Business Profile as an owner, the engine's personalization layer has flagged your business as relevant to you specifically. This means you are more likely to appear in your own logged-in queries than you are in the queries of a potential client who has never heard of you.
The cold-start test
The metric that matters for GEO is the cold-start experience: what does an AI engine recommend to someone who:
- Has never heard of your business
- Has not visited your website
- Is not in your geographic area or network
- Has no prior signal indicating your relevance
This is what a new potential client experiences. Running your audit while logged out and in incognito mode approximates this experience.
How to run a clean audit
- Open a fresh incognito window (Ctrl/Cmd + Shift + N in most browsers)
- Do not log into any AI engine account within that session
- If testing Google AI Overviews, do not log into your Google account
- Clear your browser's location permission or use a VPN to a neutral location if testing local queries from your own address
What you are actually measuring
A clean, logged-out audit measures your business's organic recommendation probability — the likelihood that an unfamiliar user, asking a relevant query, encounters your business in the AI's response. This is the only version of the metric that matters for new customer acquisition.
