When the work starts testing the theory

Something has changed for me over the past few weeks.

For much of this year, I’ve been developing Editorial Intelligence partly as a way of making sense of how content work is changing. A lot of that thinking began outside the immediate demands of my day job, and I kept returning to the same handful of questions:

• What happens when producing another piece of content becomes the easy part?
• What becomes valuable when AI makes drafting, summarising and repackaging dramatically cheaper?
• How should an organisation decide what it genuinely needs to know, what it already knows, and what evidence is worth preserving?
• What is the role of an editor when the problem is increasingly not making something, but deciding what information should move where?

Those questions became the basis for a lot of the work I’ve been publishing through Editorial Intelligence. I wrote about the editor becoming more important as systems become more automated in The system still needs editors. I’ve been developing the idea that GEO can become an editorial input rather than simply a post-publication optimisation discipline in Becoming the GEO journalist and GEO-informed thought leadership. And increasingly I’ve been interested in routing: the idea that good AI-enabled knowledge work is often less about generating another answer and more about getting the right evidence, context and judgement to the right place.

What has been interesting recently is that the work itself has started producing these problems. Not as Editorial Intelligence projects. Just as work.

The question stopped being “how do we optimise this page?”

I’m now involved in activity around GEO where the question isn’t simply how to optimise a page. It’s closer to a small set of questions asked in order:

• What are people actually asking?
• What does the organisation currently answer?
• What evidence supports that answer?
• Is stronger evidence sitting somewhere else?
• And what do we genuinely not know yet?

That last distinction has become particularly interesting.

Historically, a content gap often became a production brief: something appears missing, so make something. But there are at least three different situations hiding inside that decision. Sometimes the existing answer is basically right and needs to be renewed. Sometimes the evidence already exists — inside research, a customer story, an interview, a video or another part of the organisation — and needs to be routed into the answer. And sometimes there really is a knowledge gap, which means somebody needs to go and report, interview, research or investigate.

I’ve started thinking of that simply as:

Renew → route → report.

That sounds like a content model, but I think it is really a knowledge model.

The same pattern keeps showing up elsewhere

I’m working on customer stories where the interesting question increasingly isn’t only whether the finished story is good. It’s whether the useful evidence inside it remains findable afterwards: what did the customer actually experience, what changed, what proof did we get, what question could that evidence answer somewhere else, and what did we fail to ask. That turns a case study from a disposable publishing asset into a small addition to organisational knowledge.

Research-led video creates another version of the same problem. If you already know what evidence is weak before the conversation happens, you can potentially shape a small number of prompts upstream and collect better material in the first place.

Events become interesting for the same reason. Instead of asking we’re attending an event, so what content should we produce?, you can ask what important questions do we still lack credible practitioner evidence for, and can being in the room help us answer them? That’s a subtle change, but quite a profound one.

It means demand intelligence can influence reporting. Reporting can strengthen existing content. Customer evidence can travel beyond the customer story. Video can become a source of reusable evidence rather than simply another channel. And old published pages stop being quite so “finished”.

Reality is constraining the theory, and that’s the point

None of this proves Editorial Intelligence as a grand theory. In fact, I think the opposite is happening: it is becoming more useful precisely because reality is starting to constrain it.

Some ideas I developed abstractly become much simpler once I try to use them inside an actual organisation. Some distinctions matter more than I expected. Others matter less. Publishing systems, ownership, approvals, existing plans and the simple fact that other people are already doing good work all place useful limits on what a framework can claim.

That is probably the stage I’m most interested in now. Editorial Intelligence is becoming less of something I explain and more of something I can test through work. The loop now looks something like:

Theory helps me see the work → the work tests the theory → the theory gets smaller and more useful → I take it back into the work.

I suspect that is how a framework becomes a practice. Not when an organisation adopts the terminology. When the underlying problems keep appearing in reality — and the framework helps you make slightly better decisions about them.