Reputation Was Always the Moat
The pipe changed. The moat didn't.
More and more, people are picking who to hire without ever landing on a website.
They ask a chatbot. It gives them three names. They call one.
If your business depends on being found, that should get your attention. For twenty years the path was at least legible. Show up in search, get found, earn the click. It was slow and grindy, but you could see it. You knew what page one looked like. You knew roughly what it took to get there, and you knew whether you were on it.
Now somebody asks a machine who to hire and there's no page one. No list of ten. No comparing tabs. Three names, take it or leave it. And the machine only ever names a sliver of the businesses that exist.
So the question I'm getting from owners right now is the obvious one.
How do I get in front of it? How do I work the new system?
I have been answering a version of that question for twenty years. Back when it was just SEO. The answer has not changed, and I want to walk through why, because the reasoning matters more than the conclusion.
The short version
You can game a system. It works. For a while.
Then it stops working, and everything you built on top of the tactic goes with it. Meanwhile the person who spent those same years getting better at the actual work is still standing there, and now they are the one getting named.
I'm not making a moral argument. I'm not telling you to be virtuous and trust the universe. I'm telling you the math is against you, it's against you structurally, and it always has been.
Here's the part most people never look at.
What the system is actually paid to do
Every discovery layer ever built has had the same job: hand the user the best version of what they are looking for.
Not because the people building it are noble. Because that is the only thing that works.
Google was never trying to be fair to you. It was trying to give people good answers, because good answers hold attention, and attention was the product. Every improvement Google ever shipped to its ranking, every penalty it dropped on link farms and keyword stuffing and content mills, was self interested. A search engine that returns junk loses its users, and a search engine without users has nothing to sell.
So the entire history of SEO is one long story of people finding a shortcut and the system closing it. Keyword stuffing worked, then it didn't. Link buying worked, then it got you delisted. Content farms worked, then Panda ate them. Every one of those tactics had a shelf life, and every one of them was killed by the same force: a company protecting the quality of its results because quality was the business.
The people who did well over the whole arc weren't the ones who kept finding new tricks. They were the ones who kept being worth finding.
Where the models are different
Now here's the part I want to be careful about, because the easy version of this argument is wrong.
You will hear people say the models are running the same play as Google. They're not. Google's alignment was actually a little loose, and it's worth being honest about that.
Attention monetized through advertising is a volume business. A perfect answer on the first try costs Google an ad impression. Good enough to keep you coming back was enough, because for two decades you had nowhere else to go. That gap between "good enough" and "best" is the exact space where gaming lived. It's why the tactics worked as long as they did.
The models are playing a tighter game. Two reasons.
Switching costs are close to zero. You can move from ChatGPT to Claude to Gemini in about four seconds and lose nothing. No index to rebuild. No history you cannot live without. No habit deep enough to hold you. The only thing keeping a user is that the last answer was good. That is a harsher master than search ever had.
And the bigger one: the model's name is on the recommendation.
Google handed you ten blue links and let you sort it out. If the fourth one was garbage, that was your problem for clicking it. The search engine never claimed the fourth link was good. It just put it on the list.
The model doesn't do that. The model gives you three names and vouches for them. When it says "call this one," it just spent its own credibility on you. And if you hire that person and they are terrible, you do not blame yourself for picking wrong. You blame the machine that named them.
That is not an algorithm. That is a referral.
And anybody who has ever made a referral runs the same filter, whether they say it out loud or not: is this going to make me look bad?
That is the whole thing. The model has skin in the game in a way the search engine never did. It's not a new system to crack. It's a referral partner protecting its own reputation, which means it applies roughly the same test your best client applies before they give your name to a friend.
Which is why the way you get picked has not actually changed.
The two objections, and they are both fair
I want to take the counterarguments seriously, because I've gotten both and they're not stupid.
"Ads are coming to the models." Probably right. There is already movement toward shopping and sponsored placement, and I would bet money that some form of paid surface shows up inside these things. So do not let me sell you on the idea that the models are pure. They are companies, and companies monetize.
But it doesn't break the argument, because the argument was never about purity. It is about accountability. Whoever ends up paying the bill, the recommendation still carries the model's name. That's the part that doesn't change. An ad slot is labeled and discounted by the reader. The recommendation is the thing being trusted, and the moment it stops being trustworthy the product is dead. They know that better than we do.
"AI SEO is already an industry." Also right. There is a whole cottage business selling GEO and AI visibility optimization right now, and some of it works today.
So I won't tell you it's ungameable. That would be a lie, and you would catch me in it within a week.
What I will tell you is narrower and, I think, sturdier: it's gameable right up until it isn't. Correcting for manipulation is the one thing these systems are built to do forever, because their existence depends on it. Every dollar of that correction budget is aimed at the shortcut you just bought. You are renting an advantage from a landlord who is actively working to evict you.
That's a fine trade if you know that's the trade. Most owners buying it think they're building something.
So what do you actually do
This is where I lose people, because the answer is not clever.
Be worth talking about. Not a slogan. The actual work. Do the job so well that a normal person, unprompted, mentions you to someone else. Everything downstream depends on this, and there's no way around it. If the work is mediocre, the rest of this is decoration.
Make the proof exist somewhere other than your own website. Your site is you talking about you. Nobody weighs that heavily, and neither does a model. Real customers saying real things in public, on platforms you do not control, is the thing that carries. Most owners I meet have done real, good work for a decade and have almost no public record of it. That's not a marketing problem. It's a filing problem.
Be consistent everywhere someone might look. Same story, same offer, same name, same claims. Inconsistency reads as risk, to a person and to a machine.
Then run the test yourself. If someone recommended you tomorrow, would it make them look good? Would it make them look smart for knowing you?
If the answer is no, no tactic fixes that. If the answer is yes, you are already doing the durable version and the pipe does not much matter.
The part I keep coming back to
Do good work and people talk about it. They leave feedback nobody had to beg them for. That feedback gets seen. The people you want find you.
It's slower than the shortcut. It's always been slower than the shortcut. That's the trade, and it's the same trade it was in 2006, and the people who took it then are the ones who are still here.
Reputation was always the moat. The pipe changed. The moat didn't.
Who's the last person you recommended without being asked? And what made you do it?