Trust & Content

Dave Foulkes at Business Film Booth sent a newsletter this week that stuck with me. A new study read 10,000+ stories and spotted the human 93% of the time, just from how the story was built. The detection question is interesting. The commercial question underneath it is the one I want to talk about.

The internet is splitting in two, and your brand has to choose

Dave’s framing in his newsletter is sharp. The web is becoming two places. One built for humans. One built for chatbots. Shopify has 60-odd listicles ranking the best e-commerce platform and, surprise, Shopify wins every time. No human reads those. They are written to be scraped, so that when someone asks ChatGPT where to set up a store, the bot quotes Shopify back to itself.

That is the internet now. Content engineered to bend an answer. And as the feeds fill up with it, you stop being able to tell whether a human or a bot is reading and replying to your work. The reason to post starts to dissolve.

Here is the opening in that. The giant audience is not the prize anymore. The smaller, dedicated, trusting one is. Anywhere people still believe what they are reading is a commercial asset. That is where the buying decisions live.

What the study actually found, in plain English

The StoryScope paper from researchers at Maryland and Google DeepMind is the bit Dave was pointing at. They ran 10,272 writing prompts past humans and five big AI models, and analysed the structure of the stories, not the wording. Not vocabulary tics. The actual shape of how a story moves.

They could spot the human 93% of the time. And the tell was not style, it was decision-making. AI stories over-explain their themes. They tie everything off neatly. They favour protagonist-driven, single-track plots. They render emotion through bodies and weather rather than naming it. They avoid specific references to real people, real places, real works.

Humans, by contrast, were messier. More ambiguous endings. More flashbacks and time jumps. More direct address to the reader. Nearly double the rate of naming specific texts, brands and people. The researchers put it well: AI writes as though no one is watching. Humans write to someone.

That last line is the one I would underline for anyone running marketing, sales or CX. The machine writes into a void. You do not have to.

“AI writes as though no one is watching. Humans write to someone. That is the entire commercial argument for showing up as yourself.”

Why this matters for customer growth, not just content

Win, Keep, Grow runs on trust. Every stage of it. People buy from brands they believe, stay with brands that show up, and recommend brands that made them feel seen. None of that is generated by a model that has been trained to tidy up the ending and avoid naming anyone.

The risk is not that AI content is bad. Some of it is fine. The risk is that it is average in a very specific way. Tidy. Tasteful. Thematically explicit. Slightly over-perfumed. The exact opposite of the slightly awkward, specific, opinionated content that makes a human go, oh, these people actually do this for a living.

That is the ugly mirror moment for a lot of brands. If you stripped your name off your last six posts, your last sales email, your last case study, would anyone know it was you? Or could any of your competitors have published it? What can be seen is being managed. Right now, a lot of brands are managing themselves into the AI cluster the researchers identified, the one where every model sounds like every other model.

Real people, real talk: the content shape machines cannot fake

Dave’s prescription is simple and I agree with it. Get good at putting you and your team out there. As yourselves. With names, opinions, customers, specifics, awkward pauses, the lot.

Conversations, not monologues. Two people who actually know the work, talking about a real problem a real customer had this quarter. That is the format. Podcasts, customer interviews, founder-and-operator chats. The energy of a conversation is the bit a model cannot synthesise yet, and even if it could, getting caught faking it would burn the trust so badly it would not be worth the upside.

Named, not generic. Name the customer. Name the industry. Name the number. Name the mistake you made last year. The study found humans reference specific texts, people and places at almost double the rate of AI. That specificity is the signal. It is also, not by coincidence, what makes a sales conversation land.

Messy, not tidy. Leave the ambiguity in. The deal you nearly lost. The customer who churned and told you why. The bit of the process you are still figuring out. Tidy stories read as fiction. Messy ones read as truth.

What I would do this quarter if I ran your marketing

Pick three people in the business who actually talk to customers, and get them having a recorded conversation once a fortnight. Nothing polished. A repeatable, sit-down chat about something that happened with a real account. Fifteen minutes. Cut it into a couple of clips. Send the long version to your list.

Build the email list and the small community properly. Treat it as the asset it is. Dave’s line was that any community that feels real now is a valuable asset, and he is right. That is where trust compounds. That is where the introductions, renewals and referrals come from.

Stop writing for the algorithm and start writing for the eight people whose names you could list right now as your next best customers. If those eight feel seen, the commercial maths works. If you are chasing reach with content the chatbots wrote for the chatbots to read, it does not.

The opportunity here is not to out-volume the machines. You will lose that race this week. The opportunity is to be undeniably, specifically, on-the-record human. That is the Moments That Matter™ work. That is how you make trust visible at the scale that actually moves revenue.

The two internets, and which one your customers live on

The future Dave describes, two internets, one for humans and one for chatbots, is already here. You can feel it in your feed. The question is not whether it splits. It has. The question is which side of the split your brand can be found on, and whether the people who would buy from you can tell you are a real outfit run by real people who care about the work.

The good news is the bar to be on the human side is not creative genius. It is showing up as yourself, with named customers, named opinions and named numbers, often enough that people recognise you when it counts. Behaviour drives growth. The behaviour here is simple. Be specific. Be in the room. Be on the record.

That is the antidote to the slop. Not better prompts. You.