Don't just gather content. Gather the evidence you're missing.

The interesting question about a conference or a webinar may not be what content you can make from it, but what you could find out that you currently can’t prove.

The standard model for a conference, a webinar or a customer call is familiar enough. You attend, you capture a few quotes, and afterwards somebody produces a recap, some social clips and perhaps an article. It works, in the sense that assets come out of the other end.

What it doesn’t do is change what the organisation knows.

I’ve started thinking about these differently, and the shift is small enough to sound trivial. Instead of asking what content we can get from this, ask what the organisation needs to know that it currently can’t demonstrate — and then design the activity around answering it.

That sounds like a planning refinement. It’s actually a different job.

Optimisation can only work with what you already have

Most GEO and AI-visibility work concentrates, understandably, on the packaging. Structure the pages properly, make information accessible, strengthen entities, answer relevant questions clearly. All of it is worth doing.

But optimisation can only ever rearrange the knowledge an organisation already possesses. If the visibility analysis shows that competitors are consistently associated with a question you barely answer, or that nothing credible connects your company to a problem you claim to solve, rewriting a page is unlikely to fix it. The page isn’t the weak part. The evidence behind it is.

Which produces a different loop. Signals from search and AI visibility identify a gap. Somebody designs questions to close it. First-hand evidence gets gathered from people who actually know. That evidence is synthesised, published in a form both people and machines can use, and the visibility is measured again afterwards.

At that point GEO has stopped influencing how content is packaged and started influencing what the organisation goes out and learns. That’s a much more interesting thing for it to be doing, and much less common.

The format matters less than the access

A conference is the obvious example, because it concentrates practitioners, customers, experts and competitors in one place for two days. But nothing about this is specific to events.

A webinar exposes the questions an audience actually asks, and the points at which they are confused. A roundtable can surface disagreement between experts that neither would express in a written contribution. A customer interview produces operational detail that no market research will reach. A video shoot contains an hour of first-hand testimony even when three minutes reach the final film. An advisory board tests whether an emerging argument survives contact with people who do the work.

Sales calls expose recurring objections. Support queries show where the official explanation is failing. Community conversations reveal the language people actually use, which is rarely the language in the brief.

What these have in common isn’t the format. It’s temporary access to knowledgeable people, and a set of recurring questions somebody has already noticed.

Deciding what you want to find out

This is where the editorial part becomes load-bearing, and where most of the difficulty sits.

The visibility analysis can identify gaps. Search data shows demand. Sales and support teams can supply the questions they keep meeting. But none of that adds up to an enquiry. Somebody has to turn the signals into a small number of things worth asking: what are we actually trying to establish, who would genuinely know, what would count as evidence rather than another opinion, and what follow-up gets past the answer somebody has already prepared.

That preparation is the whole difference between opportunistic content capture and designed evidence capture. It’s also the step that gets skipped, because it has to happen weeks earlier than anybody wants to think about it, and because the person capable of doing it is usually not the person going to the event.

Making the evidence escape the recording

Capturing the material isn’t sufficient either. Most of it dies in a transcript or a video archive that nobody opens again.

Getting it out might mean a practitioner Q&A, verified findings added to an evergreen guide, an article built around a pattern nobody had spotted, a transcript-derived knowledge page, or an existing narrative updated because new evidence changed it. The form matters less than whether the knowledge ends up accessible, attributable, connected to related evidence, and reusable beyond the activity that produced it.

That is also where a claim about AI visibility becomes defensible rather than decorative. Not we ran an event, therefore GEO, which is the version that will get built. But: we identified something the market needed a better answer to, acquired original evidence from people who actually know, and turned it into public knowledge that didn’t previously exist.

The honest reason this rarely happens

I should be clear that this is more work, and that the reason it doesn’t happen isn’t ignorance.

Designing evidence capture requires somebody senior enough to know what the organisation can’t currently prove, involved early enough to shape the questions, and present at an activity they are usually attending for other reasons entirely. Events are booked by marketing teams with targets that have nothing to do with knowledge gaps. Webinars exist to generate leads. The customer call is happening because somebody in sales arranged it.

Adding an evidence brief to any of those means asking people to do something that doesn’t help them hit the number they’re measured on. That isn’t a failure of understanding, it’s an organisational design problem, and I don’t have a clean answer to it beyond noticing that the cost of asking better questions during an activity you’re already running is close to nothing.

Which gives editorial teams a more useful test than the one they normally apply. Not what content can we get from this, but what could we learn that would strengthen what the organisation knows, says and can prove.

If the answer is nothing, the activity probably doesn’t need an editorial plan at all. If the answer is substantial, it has a job before anybody turns up — and that is a considerably better reason to be in the room than collecting quotes.