Ask them
tl;dr
Let the people who use the product tell the story, and ask them for it. That idea built our first major retainer and has run through ten years of our work. Generative AI makes it more valuable: when anyone can render anything, a real person who chose you is the scarcest thing in marketing.
The bottle doesn't matter
In April 2019 we sat at the bar at The NoMad in Los Angeles with Dave Purcell, and asked him what he'd make with Plymouth Gin. He didn't hesitate. "My favorite Plymouth cocktail is a Martini."
He makes it the house way, with Manzanilla sherry in place of vermouth and a grapefruit twist, about five parts gin to two. Then he told us why he orders Plymouth everywhere he drinks. "It's so identifiably what it is that the bottle doesn't matter. People know what they're ordering."
No brand manager wrote that line. No agency could have. It came from a working bartender, unprompted, because someone pulled up a stool and asked. That trip to Los Angeles was one of several we made for the brand, and every stop sounded like that one.
Built on the people who pour it
Plymouth Gin hired us in October 2017. It was one of the best-selling gins in the world, with roots back to 1793, and the most modest social following of the six top sellers. The brief was to tell its story through the voices of the industry.
We started by archiving about a thousand old posts: stock images and blurry phone snaps from earlier management. Then we did what we'd done in print for Rémy Cointreau with 1806 Magazine, and took it to social. We traveled to bars, sat down with the bartenders who poured the gin, photographed them where they worked, and let them talk. Many of our images had no bottle in them at all.
In the first three months we featured seven US bartenders and two of the brand's own ambassadors. Instagram engagement rose 56%, from 1.08% to 1.68%, and average likes rose 64%. Against the six best-selling gins, Plymouth moved from fifth to third in engagement, passing brands with more than twice its following.
The approach held. Over the first 20 months, the accounts we ran earned 7.5 million impressions and 182,700 engagements, and drew 5,538 comments and messages from people talking back.
The further it drifted, the less it worked
Ten years and more than 40 brands later, the pattern hasn't moved. The posts that earned comments, shares and saves were the ones with a real person in them telling the truth about something they cared about. The further a feed drifted from that, toward product-on-a-plinth and lines a committee approved, the less anyone engaged.
The reason is simple. A brand saying it's great is an ad. A bartender who orders it everywhere they drink is evidence. People can tell the difference instantly, and they always could.
The Dresden Room in Los Angeles, opened in 1953, put a Plymouth Negroni on its menu. Its bartender told us the bar's guests run from 21 to 85, and that consistency is why a place lasts. Its lounge act, Marty and Elayne, had just marked 38 years on its stage. You can't write a brand story better than the one that room was already telling. You can only show up and ask for it.
What a model can't render
We build AI-native production every day, and we're optimists about it. A photographer who spent a decade perfecting a still frame can now tell stories in motion. Worlds that once needed a crew and a location budget take a week. That's a real expansion of what a small team can make.
It also changes what's scarce. When anyone can generate a flawless image in seconds, flawless stops being a signal. What can't be generated is a real person who chose you: the bartender, the customer, the teacher, the engineer, saying in their own words why they use the thing.
Where audiences draw the line on AI-generated people will keep moving. As we wrote in There is no button, every image technology has stood that trial, and audiences tend to settle on what they once refused. What won't change is where a brand's life comes from. A generated face can stand in a frame, but it has never ordered your gin, and it can't tell you why it would. The future is hybrid. The people who keep a brand alive are still out there, waiting to be asked.
What AI is for
The mistake is to point the new tools at the person. Generating a happy customer is cheaper than finding one, and it throws away the only thing the customer was worth. The right move is to point the tools at everything around the person.
Finding the stories. Brands sit on thousands of them: comments, reviews, support threads, posts from people showing what they made. AI can read all of it and surface the ten worth a phone call.
Getting more from each one. One honest interview becomes a short film, a cutdown for every platform, a still, a quote card and a line for the sales deck.
Carrying it further. Translation and captioning let a bartender in Los Angeles reach a buyer in London without losing her voice.
Building the world around it. Generative tools can extend the setting, the light and the season, while the person and what they said stay real.
Two rules hold all of it together. Nobody's story or likeness is used without their consent. And everything ships with a label that says what was captured and what was generated, which is why we publish our standard, Rendered, Not Recorded. The machine amplifies the voice. The voice still has to be someone's.
How to start asking
None of this needs a new budget line. It needs a habit.
List the people closest to the product. For a spirit, that's bartenders and brand ambassadors. For software, it's power users, support staff and the engineers who built it. Start with ten names.
Go to them. Their bar, their classroom, their desk. People talk differently in their own space, and the picture is better there too.
Ask one question and stop talking. "Why do you use it?" works better than a script.
Make it a series. One story is a post. A story every week, in a recognizable format, is a program people come back for. Ours ran for years.
Build the intake. Give people a simple way to raise their hand, watch for the ones already posting about you, and ask your sales and support teams to flag the best they hear.
Measure the conversation. Replies, saves and messages tell you whether people believed it. Plymouth's 5,538 comments and messages said more than its 7.5 million impressions.
Ask them
The technology will keep getting better at making things that look real. That's good news for anyone who makes things. It also means the brands that win the next ten years will be the ones with real people willing to vouch for them, and a way of finding those people and telling their stories well.
We learned it on bar stools in 2017. Dave Purcell said it better than we could: people know what they're ordering. The job is to find the ones who do, and ask them why.
FAQ
What is customer-led content?
Content built on the real people who use a product, telling their own stories in their own words. For Plymouth Gin, that meant bartender portraits and testimonials gathered in the bars where they worked.
Does it still work when AI can generate anything?
It works better. As generated images get cheaper and more common, a real person who chose your product stands out more. Production is going hybrid, but a brand still lives on the people who use it.
Where should AI fit in a customer story program?
Around the person, never in place of them: finding stories in comments and reviews, cutting one interview into many formats, translating and captioning, and extending the setting. Consent and clear labeling come first.
How do you measure it?
Watch the conversation it starts. In Plymouth's first three months, engagement rose 56% and the brand moved from fifth to third among the six best-selling gins, passing brands with twice its following.
by // Kyle Ford
