Illustration of a small desk microphone by a window at night, sound ripples spreading out of the window towards many small points of light in the sky

The machines don’t check your follower count

Aug 13, 2026 | Business

AI tools cite useful, honest answers whatever the size of your audience. I've watched this pattern once before, with a blog.

"I'll start posting properly when I've built a bigger audience."

I've heard that sentence more times than I can count. I never argue with it, because I once believed my own version of it. It sounds like strategy. Sensible sequencing. Build the audience first, then speak.

Recently I read some research that takes that sentence apart. It backs up something I've been teaching from stages for years, and something I lived through more than a decade ago.

Meltwater just analysed 9.5 million citations. That's 9.5 million times ChatGPT, Gemini, Copilot or Claude used a source in an answer.

LinkedIn came out as the second most-cited source on the entire internet.

What surprised me was who gets cited:

  • 51% of the people being cited have fewer than 10,000 followers.
  • 75% of citations come from individual people's profiles. Company pages barely get a look in.
  • 72% of the content being cited is original, written by the person themselves.

Someone had something to say. So they said it.

The machines don't check your follower count before they quote you. They check whether you said something clear, specific and worth repeating.

Know, trust, like. In that order.

People buy from people they know, like and trust. Say that in a room full of business owners and every head nods.

Now ask which of the three words matters most and watch the room split. Some shout trust. Some shout like. Almost nobody says know.

Here's my opinion, and it is only my opinion:

Can someone buy from you if they don't know you exist? No. Can they trust you, or like you, if they don't know you? Same answer.

So know has to come first. Without it the other two never get a chance.

That leaves like and trust. Which matters more?

Imagine someone you care deeply about needs an operation. A doctor comes in with the most amazing bedside manner. One of the nicest people you've ever met. Then you discover their success rate in this procedure is 63%.

Another doctor walks through the ward. Rude. Obnoxious. Horrible to the nurses. Their success rate is 97.3%.

Which one do you want holding the scalpel? The one you like, or the one you trust?

Trust wins. It's not close.

Like is a bonus. A massive bonus, and the longer the relationship runs, the more it matters. I once kept paying a personal trainer I actively disliked because what he was doing worked. I eventually left him for someone who got me the same results and was decent company. So yes, like can lose you a client. But it's almost never the reason someone chooses you in the first place.

The real order is know, trust, like. And without trust, nobody keeps paying attention.

AI just took over the "know" step

For the last twenty years, the know step was the expensive one. You fought algorithms, bought ads, chased followers and treated the number next to your name as a scoreboard. The whole "I'll post when I'm bigger" belief comes from that era. Know felt like a mountain you had to climb before trust was even allowed to start.

That's what the Meltwater data quietly ends.

When someone asks an AI a question, the machine makes the introduction. It decides who gets known, at the exact moment a stranger is looking for an answer. And it makes that decision on trust signals: useful, specific, original, recent.

And recent matters more than you'd think:

  • 48% of everything the AI models cited was published in the previous three months.
  • Only 12% was more than a year old.

So what you post this month counts. A lot.

You can't make a machine like you. There's no bedside manner for an algorithm. The only thing left to compete on is trust.

This is also why I've started saying SEO is dead, long live SEO.

Search engine optimisation has become search everywhere optimisation. Google's own research found buyers need around seven hours of content, eleven touchpoints and four locations before a significant buying decision. Those four locations used to be your website, your socials, your signage, your office. Today one of them is what ChatGPT or Claude says when someone asks about your industry. You're being searched in places you can't see.

Which raises an uncomfortable question. Are there questions in your industry that nobody is answering well right now? Because whoever answers them clearly is about to become the introduction the machines make on your behalf, or on your competitor's.

I've watched this film before

Here's why I trust this pattern. I've lived it once already.

In September 2008 I was at a children's birthday party when my chest started pounding. I was breathless, drenched in sweat and quietly wondering if I was having a heart attack. I did what a lot of people do with information they don't want. I ignored it.

The moment that actually changed things was far less dramatic. A year later, travelling for work, my suit tore and I had to go and buy bigger clothes. It was catching sight of myself in the mirror that finally started the conversation I'd been avoiding.

So I changed. And I documented the whole thing on a blog: the green juice, losing five and a half stone, going from the sofa to a 10-mile race, then a marathon, then a 100-mile ultra marathon.

I had no fame and no qualifications. Just a bloke going through it, sharing what I learned as I learned it.

That blog passed a million views in its first eighteen months, using nothing but organic social media and the content itself. It led to blogging for the Huffington Post, a magazine column and sharing stages with people I'd only ever seen on television.

People read it because it was useful and it was honest. The internet worked differently back then. But it rewarded the same thing the AI models reward now: someone with something real to say, however small their platform.

And notice what it didn't take. You don't need to be the celebrity who decides to talk about your subject. You don't even need to be the expert. An expert tells you what works. Someone going through it shows you, in real time, with all the mess left in. Sometimes that builds even more trust.

The question underneath

So if the follower count was never the permission slip, why do so many smart people with real knowledge stay quiet?

My guess? "I'll start when I'm bigger" is protection. As long as you haven't properly started, nobody can judge what you had to say.

I know something about the gap between knowing and doing. It took me years from knowing my life had to change to actually changing it. "I know" and "I've decided" are different sentences and you can live in the space between them for a very long time.

So what do you know that you're still not saying?

Because somebody is out there searching for that answer.

And more and more, something is searching for it too.

P.S. You may have noticed this article has headings, bullet points and actual numbers in it. According to the same research, 100% of the most-cited articles are structured exactly like this. Now you know why.

Keep reading

Fresh off the press

Beth Ayers of Metomic speaking on stage at aiblLIVE London in a pink jacket

The trick is being lazy

A self-confessed disorganised CEO got her AI to build its own filing system. Why that’s my next step, and what it could mean for you.

Every week

Get The Mix

One short email a week. Two minutes to read.

01One track
02One idea
03One thing worth your time

Unsubscribe in one click. Your email stays with me.