I’ve had this hobby horse for a while. It came up again on a podcast prep call this week, so I figured I’d finally write it down.
You know Moore’s Law, right? Compute power doubles on a curve, so things keep getting faster. I think search works kind of the same way, except what’s speeding up is how fast each era gets replaced.
Here’s how I think about it
The first search engine was the phone book. In my head it had about a 70-year run. Then Google came around and owned search for roughly 25 years. Then AI disrupted that. I’d guess this version of AI search lasts maybe another 10 to 12 years before something else shows up.
When I actually went and looked up the dates, the phone book number was off, but in the direction that makes the point stronger. The first telephone directory was a single piece of cardboard in 1878. The first Yellow Pages was Chicago in 1886. So paper directories ran for over a century before Google showed up in 1998.
And AI didn’t ease in. ChatGPT launched in November 2022 and was at roughly a billion weekly users by July 2026, according to an OpenAI spokesperson quoted by The Verge. The phone book never moved like that.
Could I be wrong on the 10 to 12 years? Totally. I’m pretty confident about the direction, though.
Nobody saw this coming
If you’d asked me six years ago what search would look like today, I wouldn’t have gotten close. Nobody would have.
And honestly, I was getting a little burned out on SEO before all this. I’ve been doing it for 23 years, since I was 18. For a long stretch the fundamentals barely moved: build links, write content a certain way, fix the technical stuff. AI search is what got me excited about the work again, because testing things and finding out what actually works is fun again.
So what do you do with this?
If eras are getting shorter, I’d be careful about going all-in on tactics that only work for one interface. A lot of prompt-specific tricks are going to age badly. I already see citations decaying faster than people expect (I wrote about that here).
What held up from the phone book to Google to AI is pretty boring: be a business people actually talk about, make sure the information about you is accurate everywhere, and publish stuff that adds something new. That’s what I mean by being recommendable, which is tomorrow’s post.
And keep testing. If you wait for a documented best practice, you’ll probably get it right about when the era’s over.
I also think the next shift is already starting. Agents are skipping search and going straight to the source. More on that later this week.
Want help building something that outlasts the current interface? See our services or get in touch.
Leave a Reply