The LinkedIn number I care about now is 83%

I think I had LinkedIn slightly wrong.

For a while I worried that posting almost every day was too much — that I was cluttering people’s feeds, and that people I knew through Sage and previous jobs would eventually think here he is again. But I was experiencing my own publishing completely differently from everyone else. I see every post because I wrote every post. My network doesn’t. LinkedIn decides which tiny fraction of everything published actually reaches each person, and the evidence was sitting in my own analytics the whole time: most of what I post gets fairly modest reach. It isn’t being pushed relentlessly at everyone I know.

Then one post escaped. I’d written that editorial jobs aren’t simply disappearing because of AI — that the valuable part of the work is moving upstream, into research, judgement, narrative, positioning and deciding what an organisation should say before somebody produces the words.

When I first wrote this article, 83% of the post’s reach was coming from outside my existing network. That was the number that changed how I thought about what was happening.

An update: 83% became 93%

The latest LinkedIn analytics make the signal stronger.

The dashboard now shows 8,308 impressions, 5,830 members reached, 93% out-of-network reach, 48 profile viewers from the post, 16 followers gained directly from the post, 32 saves and 17 visits to the linked article.

LinkedIn’s reported impression total has also shifted slightly from an earlier snapshot, so I would not treat every live dashboard number as a fixed historical counter. But the more important direction is clear: the share of discovery outside my existing network increased, and LinkedIn is now directly attributing follower growth to the post rather than leaving me to infer it from timing.

That matters because it changes one of the limits I originally put on the argument. I still can’t say that every new person who followed me around this period came because of this one post. But I can now say LinkedIn itself attributes 16 followers to it.

Why that number, and not the impressions

Most of my LinkedIn network was built through Sage and earlier jobs. There’s nothing wrong with that network, but most of those people didn’t connect with me because they wanted to follow Editorial Intelligence. Judging whether EI had legs by whether my existing network engaged with it was always a fairly weak test. What I actually needed was for an idea to escape that network and find another one, and for the first time, that seems to have happened.

The people arriving around the post include people working in editorial, communications, content strategy, narrative, journalism, PR and knowledge work. It is the first meaningful external evidence I’ve had that some of the things I’ve been thinking about inside my own work bubble — which I’ve written about needing an outside edge for before — resonate with people outside it. That matters more to me than the impressions do.

I’m also building an audience that belongs to the ideas

There is another reason the out-of-network number matters to me: I’m beginning to create an audience that exists because of the ideas themselves, rather than because of where I happen to work.

Most of my professional network is inherited. It accumulated through employers, clients, projects and years of being connected to people for reasons that had nothing to do with Editorial Intelligence. This audience is different. It is starting to be earned argument by argument. I don’t need everyone I already know to become interested in Editorial Intelligence. I need to keep publishing useful enough ideas that the people who are interested can find me.

The new follower attribution makes that less abstract. Sixteen people did not just see the post. According to LinkedIn, they chose to follow after seeing it. That is a much more useful signal for what I am trying to build than another few thousand impressions would have been.

That feels valuable in both directions. I don’t see independent work and a day job as competing intellectual lives. Working inside a real organisation gives me difficult, consequential problems to think about. Publishing independently gives me a place to test general ideas, encounter people outside that environment and get external behavioural feedback on what travels, what gets saved and what brings people back.

The useful loop is becoming something like:

practice → idea → public experiment → audience signal → learning → better practice

There are obvious boundaries. Confidential employer knowledge stays confidential, and a personal LinkedIn result isn’t a benchmark for a corporate account. What can travel between the two is the professional learning: how audiences discover ideas, how evidence changes an argument, how AI changes the economics of experimentation and how editorial judgement decides which signals are worth acting on.

It’s also making me rethink the AI slop complaints

AI undeniably makes it cheap to publish mediocre content. But LinkedIn doesn’t show every user everything that gets published — most of it disappears. So the problem isn’t simply that too much content exists. I’ve argued before that the slop debate stays at the wrong altitude, and this is a specific, mechanical version of that same point: the more interesting question isn’t how much gets produced, it’s what earns distribution. Which ideas get saved. Which ones travel outside the existing network. Which ones make somebody follow because they want the next argument, not just this one.

That’s where volume gets interesting rather than just noisy. I’ve been posting almost every day. Most posts have done very little. Then one travelled — and I probably couldn’t have sustained that much publishing alongside a full-time job without AI helping me research, develop, challenge and edit ideas. The obvious interpretation is that AI lets me produce more content. I think that undersells what actually happened, and I’ve made a version of this argument before about AI generally: AI didn’t just give me more content, it gave me more experiments. More attempts meant more feedback, and more feedback meant more chances to discover which problems other people actually recognise. Eventually one of those attempts crossed into a different network. The useful equation isn’t AI → more content. It’s AI → cheaper experimentation → more serious bets → faster audience learning.

This has implications for brands too

Corporate LinkedIn can generate impressive-looking engagement that’s heavily driven by employees, existing customers and people already close to the organisation. That can be useful, but it’s a different thing from discovery. For thought leadership specifically, I increasingly think the better questions are whether a piece reached people who didn’t already know you, whether they stopped, whether they saved it, whether they read more, whether they followed, and whether some of them came back. That’s closer to building an audience than simply accumulating engagement — corporate reach and discovered reach can look identical on a dashboard and mean almost opposite things.

It also resolves some of my own anxiety about publishing too much. The question probably shouldn’t be am I posting too often. It should be is this idea distinct enough to deserve another attempt. AI can give me the capacity to publish more. Editorial judgement still has to decide whether I should — which is a healthier use of abundance than building a content factory.

The title of this article is already out of date. The number is now 93%.

I think that’s useful rather than inconvenient. 83% was the moment I realised the idea had escaped the bubble. 93% — plus 16 followers LinkedIn attributes directly to the post — is stronger evidence that it found an audience outside it.