Keep the best, speed up the rest

The bike I rode down the Great Divide has no chain. There is a carbon belt where the chain should be and a gearbox sealed inside the frame; a GPS told me where to turn, and a dynamo in the front hub kept the batteries topped up.

None of this technology pedalled for me. A purist might not approve, yet I still had to get forty kilos of bike and 100 kilos of me over every pass between Banff and the Mexican border, and I doubt anyone thinks the belt and navigation made it less of a ride.

This edition is about change and the long tail; our profession is having a daily argument about new technology that challenges tradition, and can unlock the stories we never get to.

01  The pushback

The mood around AI has turned. Half of US consumers told Gartner they would rather give their business to brands that do not use generative AI. Pew finds that 52% of American adults are now more concerned than excited about AI in daily life, up from 37% five years ago; and for the first time that includes a majority of the under-thirties.

I get it. We're all confronted with the slop, daily, and we've already covered what this is doing to trust.

And slop is the least of it. The data centres are hungry for power and cooling. Whether true or not, AI is the most-cited reason for announced job cuts in the US so far this year. Last week OpenAI disclosed that its own agents had been probing US government websites.

People are using these tools to fake things and to steal.

In our own work the fear has a particular shape: will my interview end up training somebody's model? The answer is no, with paid systems, frontier models and firm agreements.

Then there are our friends and colleagues. More than a third of translators told the Society of Authors they had already lost work to generative AI, and that was back in 2024. This summer ProCopywriters asked 573 copywriters whether they would press a button to erase it from the world; 64% said yes, up from 45% a year before. Yet I'm sure that over 80% of the same group use it.

So I am not surprised that there are teams with a no-AI rule for customer stories. At the other extreme there are marketers who 'prompt and publish' whatever comes back, seemingly without a second glance.

02  Two ways to be wrong

The first is obvious. Hand a model a thin source, publish the result, and add to the pile that's causing the pushback. I have lots to say on this, but nothing new.

The second is harder to spot, disguised as a principle. Keep the tools out, do everything the way we did it in 2019, and call that quality. It feels like integrity. In practice it does not serve the need; it protects a nine-week queue and a budget that stretches to half a dozen stories a year.

I say this as someone who has made a career of living and championing the traditional process. A prepared interviewer, a real conversation, an advocate who enjoys the experience; nothing a model does comes close to that. For value stories it's a no-brainer and I would actually spend more money and time on this than most programs do.

We've been here before. In 2020 we all stopped flying to customer sites and did our interviews over a video call, because we had to. Many of us never went back to flying for a forty-five-minute conversation; the experience is different but the interviews are no worse for it.

The tools changed; the craft was always in the questions.

03  Keep the best

So which parts of the old way are the best of it?

The conversation. Forty-five minutes with someone who did the homework, listens hard, and follows the thing the advocate had not planned to say.

The advocate's experience. How they were asked, how their time was used, whether they would do it again. There is no version of this work in which a poor advocate experience is acceptable.

The sign-off. The advocate approves the exact words, every time.

The editor's eye. Somebody has to notice that the best line in the interview was the throwaway one at the end.

Film, when the story earns it. A crew, a location and a day of somebody senior's time are still worth it for the right customer.

Looking at that list, what I suppose I'm saying is that the things to keep are the moments when one person is really paying attention to another.

The demand for skill remains. Good journalistic interviewers, editors, QA folks and story managers should be busier than ever; the drafting got cheaper, not the judgement.

04  Speed up the rest

What about everything else? Transcribing, drafting, scoring against the brand rulebook, the whole publish kit. Then the cutdowns: the executive summary, the press release, the six translations and the social cards. Working out who has the document and who is sitting on it. Chasing them.

This is where the weeks go and the largest gains can be made. I am certain that nobody became a customer marketer for the love of chasing approvals.

05  People still buy boxes

(Sorry, brand team!)

The world loves video, yet video eats budget like I eat princess cake. Brand teams want the cinematic, emotive story, with the drone, and the founder walking in slow-mo, and for the right story they should have it!

Meanwhile, people still buy boxes. Whatever the marketing solution name, somewhere a buyer is choosing a server, a licence or a service contract, and what they want to know is whether it worked for somebody in their position. A film about another company's transformation will not tell them that.

06  Which brings me to the long tail

If we ignore flexing the bill of materials for a story, the traditional method costs much the same for your fortieth-largest customer as for your largest; it is the full treatment or nothing at all.

People buy because a company like theirs did, and the company like theirs is usually in the long tail: your only customer in Nordic logistics, or your only one running the product in a hospital.

Forrester now counts thirteen people inside a typical buying decision, each hoping to see somebody with their own job title. And for all the AI pushback, half of B2B software buyers told G2 they now start their research with an AI chatbot more often than with Google.

If the only stories you have are about four famous customers, you are unlikely to be in the answer.

So the tail is where innovation can hit hardest; these stories can often be built from a conversation that has already happened (the QBR, the webinar, the event).

Judge the material hard; a model will return eleven hundred fluent words from almost nothing and will not tell you the source was thin. And when a tail story turns out to have some jeopardy in it, bring in the old process and the A-team.

07  Where the money goes

Pay flagship prices for flagship stories: your best interviewer, the film crew, the executive's afternoon. I would pay for the top out of the middle.

Use the tech for the long tail. The process changes but the people stay the same. Your in-house team, your agency and your freelancers are the ones with the judgement all of this depends on; and since taking on the long tail is new work for them, what's not to like?

08  What comes next

This is the eighth issue of Proof Positive.

Next issue: found, or invisible. If buyers are putting their questions to an assistant, a customer story the assistant cannot find, read and cite is not doing its job; and very few were written with that reader in mind.

Subscribe if you have not. Forward to one person with a no-AI rule and a nine-week queue.

§

Robin Hamilton is the founder of Proofpoints, the customer storytelling platform. Previously, cofounder and CEO of inevidence. Proof Positive is published fortnightly.

Two weeks from now, in your inbox. Fortnightly, no spam, easy unsubscribe.