I’ve been doing this for years and I’ve ridden the ups and downs. You learn the rules, the rules shift, you learn them again. Over time I got pretty good at the whole dance.
Then the traffic started to change.
Not all at once, and not in a way that set off alarms at first. Just a shift in the shape of it.
What does a shift in shape look like?
Nothing dramatic enough to act on. That’s the problem.
Total numbers that haven’t collapsed. A different mix underneath them. Queries that used to arrive and no longer do. A slow change in which pages are doing the work. None of it triggers anything, because there’s no line on any dashboard that says the mechanism has changed.
That’s how nearly everybody experiences this. Not a crash. A gradual alteration in the composition of something you were only watching in aggregate.
So I did what I always do when something changes. I started reading.
The book on this: The Day Your Website Died is forty-two chapters on how answer engines decide who gets named, assembled because I went looking for a clear account and couldn’t find one.
Where did the reading go?
I started reading about AI, and from there about answer engine optimization, and I found a whole different world.
That was the point at which this stopped being a traffic problem and became a subject. The mechanism by which an AI system decides whose work to name when somebody asks it a question isn’t the mechanism by which a search engine decides what to rank, and I had been assuming, without examining it, that the second would carry over into the first.
It doesn’t. Different signals, different structures, different results, arriving differently.
What surprised you most?
The thing most likely to throw you, so I set the expectation before anything else in the book.
This isn’t like SEO. With SEO there is, eventually, a kind of big bang. A day arrives when Google decides to rank you, and the traffic shows up. You can point at it. There’s a before and an after.
AEO doesn’t work that way. There’s no single moment, no flip of a switch.
That absence is psychologically brutal and it’s the reason most people abandon the work. Everything you know about optimization was learned in a system that eventually confirms you. This one doesn’t confirm you. It accumulates, quietly, and the evidence arrives as somebody mentioning that a machine recommended you.
Why write down an SEO strategy that works slowly?
I wrote it because I went looking for a clear account of how this works and couldn’t find one, so I assembled the one I wish I had been handed at the start.
That’s nearly always why I write anything. The material existed in fragments, scattered across posts by people with something to sell, most of them describing tactics without the model underneath. What was missing was somebody laying out the mechanism plainly enough that you could reason about it yourself.
It’s been a long and twisty road, and an interesting one, and so far a successful one.
Does SEO stop mattering?
It hasn’t stopped mattering and it’s stopped being sufficient, a more awkward position than either extreme.
Both systems are running. They reward overlapping but different things, and a great deal of effort that helps one does very little for the other. That means the work roughly doubles, and it means some of what you’ve been doing for years is now maintenance instead of growth.
Nobody wants that answer. The clean versions, that SEO is dead or that nothing has changed, are both more comfortable and both wrong.
How often should you post and comment on LinkedIn?
Watch the shape, not the total. The total moves last and by then you’ve lost a year.
Don’t wait for the big bang. It’s not coming, and waiting for it is how people conclude this doesn’t work.
Learn the mechanism instead of the tactics. Tactics go stale in months. The model behind them doesn’t.
Expect the evidence to be anecdotal at first. Somebody telling you a machine recommended you is a real signal even though it’s not a chart.
Assume both systems are live. The work doubles.
I got pretty good at the old dance over a lot of years. What I didn’t expect was that the useful skill would turn out to be the reading instead of the dancing, and that it would all start with a change too small to alarm anybody.