All posts

What a film about D-Day weather forecasting taught me about SEO in 2026

This month I was at brightonSEO — my first time at the conference, and something I’d been meaning to do for a while. I’ll write more about the talks separately, but one of the unexpected highlights wasn’t a keynote. It was a film.

On the Thursday evening, I needed something to do to pass an evening in an unfamiliar city. I chose to visit the cinema and watch Pressure — a drama about the meteorologists tasked with predicting the weather for the D-Day landings. Andrew Scott plays Captain James Stagg, the chief meteorologist, and Brendan Fraser plays Eisenhower. The premise is simple but genuinely tense: with tens of thousands of lives and the entire Allied invasion plan hanging on the decision, Stagg has to stand in front of the most powerful military leadership in the world and tell them whether the weather will allow a landing at Normandy — knowing that the models he’s working with are incomplete, the data is patchy, and the science can only take him so far.

He can give his best professional judgement. He cannot give certainty.

I’ve been thinking about that film a lot since I got home.

The comparison I keep coming back to

SEO has always involved a degree of uncertainty. Google doesn’t publish its algorithm. The factors that determine where a page ranks are numerous, complex, and constantly changing. You can follow best practice, produce excellent content, build a well-structured site, earn good links — and still not end up where you hoped, or find that what worked brilliantly six months ago has stopped working today.

That was always the deal. Most experienced practitioners accepted it.

But something has shifted. The rise of AI-powered search — Google’s AI Overviews, ChatGPT, Claude, Gemini, Perplexity, and whatever comes next — has introduced a new layer of unpredictability that’s qualitatively different from what came before. These systems don’t just rank pages. They synthesise information, make judgements, and produce answers. The outputs are probabilistic in a way that traditional search rankings never quite were. The same query, asked twice, can produce a different answer. The same business can appear in an AI recommendation one week and disappear the next, for reasons that are almost impossible to reverse-engineer.

And the companies building these systems — for entirely understandable commercial reasons — don’t publish enough about how they work to make the outcomes fully predictable. Even if they wanted to, the nature of large language models means that full explainability is genuinely hard.

Stagg’s problem, in other words. More data than ever before. More variables than any single model can hold. And a room full of people expecting a definitive answer.

What this means in practice

I want to be careful here, because I think there are two wrong responses to this situation.

The first is to pretend the uncertainty doesn’t exist. To promise clients that the right content strategy, the right schema markup, the right keyword targeting will produce predictable outcomes in AI search — because that’s what they want to hear, and because saying “it’s complicated” doesn’t make for a satisfying sales conversation. This is what bad SEO agencies have always done, and the AI transition has given them a fresh set of vague promises to sell.

The second wrong response is to throw up your hands and say that nothing can be known, nothing can be done, and the whole field has become too unpredictable to navigate. I’ve seen a version of this in some of the more anxious corners of the SEO industry too — a kind of learned helplessness in the face of change.

Stagg didn’t do either of these things. He didn’t tell Eisenhower what he wanted to hear. And he didn’t say the weather was unknowable. He gathered every piece of data he could, analysed the patterns, made his best professional judgement, communicated his reasoning clearly — and was honest about what he couldn’t know.

That feels like the right model to me.

What I’m actually doing differently

Concretely, what this looks like in my work is:

Publishing content that is specific, factual, and genuinely useful — rather than content optimised primarily for keyword patterns. AI tools are good at recognising when something is written to game a system rather than to help a reader.

Building structured data into every site I work on, so that whatever tools are trying to read and understand a business’s content have the clearest possible signal to work with. Schema markup isn’t a guarantee of anything, but it’s the closest thing to speaking the language these systems are trying to learn.

Being honest with clients about timescales and outcomes. “Here’s what we’re trying to achieve, here’s why we think this approach gives us the best chance, and here’s how we’ll know if it’s working” — rather than promises I can’t keep.

Watching the data carefully and being willing to change approach when the evidence suggests something isn’t working. Not panicking at short-term fluctuations, but not being complacent either.

And, genuinely, staying curious. The people at brightonSEO who seemed most useful to listen to weren’t the ones promising certainty. They were the ones sharing what they were observing, what they were testing, what they didn’t yet understand. That intellectual honesty is, I think, the right posture for this moment.

One last thing about the film

Without giving too much away: Stagg makes his call. He gives his recommendation, knowing it might be wrong. The decision is made. The weather does what it does.

What stays with you isn’t whether he got it right. It’s the quality of the reasoning that got him there.

I think about that when I’m advising clients on their online presence. I can’t tell you exactly where you’ll appear in Google’s AI Overviews in six months. Nobody can. But I can tell you what the evidence suggests, what the best current thinking is, and what I’d do if it were my business. And I can be honest when that reasoning is based on incomplete information — which, at the moment, it usually is.

That’s the job.

I try to write about what I’m actually thinking about in my work, including the parts that don’t have easy answers. If you found this useful, I’d love to hear from you.

Got a project, or just a question?

Drop me a line and you'll get a friendly, jargon-free reply from me — usually the same day.