About

Why I built this

Visits from chatgpt.com kept showing up in my own traffic logs. Somebody had asked an AI about my site and been given an answer. I could not find that answer anywhere.

The tools that could have told me started around $29 a month, and they were built for agencies juggling dozens of clients. I tried a few anyway. The rate they showed me swung week to week while I changed nothing, and few of them wrote down how many times they had asked. Without that count you cannot separate a real change from noise. Some services do repeat their queries. Those start in the hundreds of dollars a month and want a demo booked before you see anything.

So here each question goes to 4 engines, 7 times each. You get the count itself, how many of those answers named you, with a margin of error attached. Next week, when the rate climbs, you can see whether the climb is bigger than that margin.

I run this on my own right now. A one-person operation is a fair thing to be wary of, so let me be direct about it: on the quality of the product itself I have no intention of finishing second. And as the methodology page sets out, this measurement has limits. Saying those out loud before you run into them is the part I care about most, and it is the same reason the free scan never asks you to sign up first.

What this deliberately does not do

A sentiment score

The tone of each answer is already recorded, and subscribers see it run by run in the detailed report. What I do not build is a single rolled-up number like “sentiment 72”. Averaged over 7 samples, tone reads far firmer than the evidence behind it. A favourable tone also leaves you nothing to act on, which is the job the cited sources do.

Racing on engine count

Putting another logo in a comparison table takes an afternoon. Keeping a scraper honest on 4 engines takes the rest of the year. I would rather run 4 engines whose numbers hold up than a longer list whose numbers are soft. Engines get added once they take real search share.

A single visibility score

Rolling it into one score buries the thing you most need to see here: how far the answer moves between runs. So you get a range instead, published even in the weeks it looks bad. That is what makes next week's rise worth believing.

Reaching a person

Email ysc9550@gmail.com. It reaches me directly with no queue in between, which is the one clear upside of a one-person operation. Refunds, deletion requests, and “this number looks off” all land in the same inbox, and that last one is the one I am always glad to get.