What we asked and what we found
Every entry answers the same eight questions, in this order.
Question
Can we observe how a business appears in AI-era search without inventing an AI score?
Why it matters
The market is full of dashboards that score a business's presence in AI answers. Most are ordinary search data with a new label. A business deserves to know what is measured, what is not, and what we refuse to infer.
State
Running experiment Built and running, being proven.
What we built or tested
- Google Search Console data on how a site appears in AI-generated results, labeled by the surface it came from, with the raw source details kept.
- AI-generated results kept in their own record, apart from ordinary search.
- A firm refusal to guess undocumented values or to turn impressions into clicks, leads, return on spend or a ranking score.
Evidence
- What is measured
- What Google exposes, what it does not, what we record, what we refuse to infer
- How we check our own work
- Earlier checkpoints found code that existed with no production use, and said so
What we learned
- The source is the product. A number without its source is a guess with a font.
Limitations
- Search Console does not give you a ChatGPT ranking. Nothing here claims it does.
- Before and after observations are not cause and effect.
What changes next
Add direct answers from ChatGPT, Claude, Gemini and Perplexity beside the Google data, each dated. This has started. The AI Observatory page describes how answers are checked.
Failed tests are published as results. A state on this page describes today, not plans.
Next in Labs
- Next: Start with the business, not the homepage
A redesign that starts with the business, not the homepage, changes the structure of the site before any page is designed.
Field notes
Get Upstream
One email a week: what we asked the AI, what it got wrong, and how we checked.
Our research reaches 17,000+ followers across our channels. Upstream is the weekly version.
Weekly. Unsubscribe anytime. Privacy