Case studies.
Real buyer prompts, run live against AI answer engines, every citation written down.
GeoCited
ChatGPT named us the top GEO agency in Chennai. We ran it 18 times to see how often that is actually true
Our own audit: named in 4 of 18 runs, all four from one phrasing. Reorder the same six words and we vanish. The 11 runs we lost are published at the same size as the 4 we won.
GeoCited
How a brand-new domain with 7 backlinks reached Google's AI Overview in under 48 hours
Our own site: 7 referring domains, zero indexed blog articles. One research study crawled 90 minutes after we asked, ranked first inside a day, and pulled into the AI Overview source pool — with everything we cannot prove written down next to it.
Zakro
Known by name. Invisible by category.
A factory-operations SaaS cited on 2 of 10 buyer prompts — both brand-name queries. Invisible on every question a first-time buyer actually asks.
How these baselines are run.
Every study on this page is a live measurement, not an estimate. The method is the same one we run for clients, and it is published here so the numbers can be checked rather than taken on trust.
Real buyer questions, not keywords
We start from the questions a buyer actually types when they are choosing software — the problem-shaped ones, not the brand-name ones. Both get measured, because the gap between them is usually the finding.
Run live against the major answer engines
Prompts go to the engines buyers actually use, in fresh threads with no memory and no personalisation carried over. A logged-in session with history would measure our own account, not the market.
Every answer archived
The full text of each response is saved, along with the sources it cited and the date it was run. AI answers change, so an unarchived measurement cannot be verified later.
Reported as Share of Answer
The headline number is the percentage of those questions where the product gets named. Because answers vary between runs, it is tracked on a schedule rather than measured once.
AI answers are non-deterministic: the same prompt can return a different answer an hour later. That is why every figure on this site carries the date it was captured, and why a single run is a snapshot rather than a score.