SEO and AI search, explained for the person paying for it

A free course for business owners who are buying search work and want to judge whether it is real. Eight lessons. No email required, nothing to install, and every number on these pages comes from data we can show you.

By the end you will be able to hold an intelligent conversation with an agency, recognise a claim that cannot be true, and know which questions to ask. You will not be able to do SEO yourself, and that is the point.

1What search is nowWhat actually happens when someone looks for a business like mine?2Why your traffic changedMy traffic dropped. Is that AI, or is it something we did?3How AI engines decide what to citeWhy does an AI answer name some businesses and not others?Our own data4What actually moves the needle, and what does notOf everything an agency could sell me, what is worth paying for?Our own data5How to read an SEO reportI get a monthly report. What am I actually looking at?6What to ask an agencyHow do I tell a good agency from a plausible one?7What you cannot buyWhy can't I just pay to fix this faster?8What to do nextI have limited time and budget. Where do I start?

Get the one-page field guide No email required. It is a web page: print it or keep the tab open.

Where the numbers come from

Two of the eight lessons carry figures. They come from our own corpus: 12,778 AI answers collected across 5 engines over 164 days, containing 109,674 citation events across 8,577 distinct domains. Recomputed 2026-09-17 from the data on disk.

The other six lessons carry no figures at all. That is deliberate. Our corpus measures how AI engines cite sources, so it can support lessons 3 and 4 and nothing else. Rather than borrow someone else's statistic to make the other lessons look denser, they carry worked examples instead.

What this corpus cannot tell you:

Where we use a third party's number, it is labelled and linked at source, and we report what it measured rather than the version that flatters us.