Your AI strategy is the new "Spotify model"

Spotify model vs. AI strategy: the industry that produces identical organizations is still running

Around 2012, two Spotify engineers published a document describing how the company organized its teams. It was a new vocabulary, with squads, tribes, chapters and guilds, and the promise of coordinating a lot of people without the weight of traditional management.

The text went viral. It became slides, consulting, the “Spotify model,” and for a few years it was copied by companies all over the world that wanted a bit of that agility.

There was just one problem. Spotify itself didn’t use the Spotify model that way.

In 2015, one of the authors, Henrik Kniberg, published a piece with a blunt title: “No, I didn’t invent the Spotify model”. He explained that he was only the messenger, that it portrayed the company at a specific moment, still evolving, and that the point of looking at what another company does is to adapt it to your own context, not to copy the formula. Years later, a former employee described the so-called model as aspirational, never fully implemented as the structure the world was copying.

I’m bringing this story back because it’s about to repeat itself, though maybe with other players.

The cycle organizations insist on repeating

The script is always the same. An admired company does something that works for it, under conditions that are its own. Someone writes a beautiful case study. The case loses its context along the way and becomes a universal method. And then comes the expensive part: other companies roll out the formula by decree, in realities that have nothing to do with the original.

When it doesn’t work, the hasty conclusion is that the method is no good. But the method was never the point. What can’t be copied is the context that made the method work there.

I’ve seen this with squads. I’ve seen it with OKRs. I’ve seen it with agile cells. And I’m watching it start again, now with the acronym of the moment.

This reflex to copy has a name in organization theory. Paul DiMaggio and Walter Powell, in a 1983 paper, called it mimetic isomorphism: the tendency of one company to imitate another that seems successful, even without a clear reason to believe it will work in its own case. And they pointed to the condition that most triggers this reflex: uncertainty. When no one knows for sure which way to go, imitating whoever seems to know becomes the safest shortcut, or at least the one that’s least work to justify in a meeting.

Notice that few things produce as much uncertainty today as the question “what do we do with AI?” It’s the perfect ground for this reflex. And there’s an irony in the middle of it, because imitating under uncertainty doesn’t dissolve the uncertainty. It only pushes it forward, to the day the copied formula doesn’t fit and no one can explain why.

The new formula is called AI

The new irresistible case study is the company that is “already AI-driven.” It shows up on the conference stage, showing the fully automated workflow, the customer service that answers on its own, the team that shrank and the productivity that went up. The audience takes notes. The next day, the question in the boardroom is “why aren’t we doing it this way yet?”

It’s the Spotify model all over again, in new clothes. The same showcase, the same forgetting of context, the same rush to copy.

And context, in the case of AI, matters even more. The company on stage has a kind of data you may not have. A clean history that took years to organize. A volume that justifies the investment. A problem that, as it happens, is exactly the kind the tool solves well. None of that shows up on the slide. What shows up is the result, never the soil it grew in.

Copying another company’s AI strategy is copying the answer without knowing the question it was answering.

What the Spotify story teaches about adopting AI

Kniberg’s lesson wasn’t “don’t look at Spotify.” It was the opposite: look, but to learn how to reason, not to copy.

It’s worth doing with AI what almost no one did with squads. Before asking “how does company X use AI?”, ask the following about your own organization:

• What real problem of yours would this tool solve? Does that problem resemble the one at the company you’re copying, or is only the tool the same?
• Does the context that made it work there, such as data, volume and maturity, exist here?
• If you adapt instead of copying, what from the original formula makes sense to keep, and what needs to change to fit your reality?

None of these questions shows up in an on-stage case study. All of them separate those who will learn from AI from those who will only repeat, three years from now, that “that AI thing didn’t work.”

Food for thought

• When you look at a company that seems to be ahead in AI, are you trying to understand its reasoning or to copy its configuration?
• What in your context, from data to team maturity, is different from the context of that case that impressed you, and how much does that change the result you can expect?
• Three years from now, will your AI adoption be remembered as a decision of your own, designed for your problem, or as the time you tried to wear someone else’s model?