Before he became famous in innovation circles, Clayton Christensen was hired to study milkshakes.

This was not some quirky sabbatical project. Christensen, the Harvard Business School professor whom The Economist once dubbed “the most influential management thinker of his time,” was brought in by a fast-food chain to solve a stubborn problem: milkshake sales weren’t growing. The company had tried everything: new flavors, chewier textures, lower pricing, all based on what their customer personas said milkshake drinkers wanted. Nothing moved the needle.

So Christensen and his team tried something different. They parked themselves inside one of the chain’s restaurants and recorded 18 hours’ worth of milkshake purchases. Who bought them? What were they wearing? Did they sit down or drive off? Were they alone? With kids? With coworkers? They gathered all the surface-level data they could.

They found something unexpected: nearly half of the milkshakes were sold in the early morning, to people who didn’t stick around. They were commuters. But why a milkshake, of all things?

The next day, the researchers took a different tack. They stood outside the restaurant and intercepted milkshake buyers as they walked to their cars. And that’s when the real insight landed. The milkshake, it turned out, was meeting a specific need, whether it was keeping hands occupied during boring drives, avoiding the mess of a donut, lasting longer than a banana, and, ultimately, tiding them over until lunch.

In other words, the milkshake wasn’t just another sweet treat. Christensen argued that it was being hired to do a job.

Method First, Then Meaning

This milkshake story has become a legend in business schools. But to me, it’s less a lesson about marketing than it is a story about the persistence that’s required in research.

Christensen’s first approach — standing in the store and noting observable behavior — got him what was happening, but not why. He only reached the deeper insight by leaving the building, standing outside, and simply asking people. It wasn’t fancy, but it worked, and it only happened because he was willing to shift his method.

Start Where You Are, Then Iterate

In research, there’s an unspoken pressure to nail it on the first try. To choose the “right” method from the start and avoid what feels like backtracking. But in reality, iteration isn’t a detour. It’s the path forward.

You begin with the best lens you have, knowing it may only bring part of the picture into focus. Once you see where it’s blurry, or where your data contradicts your expectations, you adjust. You reframe the question, tweak the method, or layer in a new one. That’s still the scientific method: form a hypothesis, test it, evaluate, refine, repeat.

Christensen’s team didn’t find the milkshake’s “job to be done” in their first round of observations. They found it by stepping outside, literally and metaphorically, and asking different questions. That wasn’t failure, it was progress.

At Bregma, we approach every project as a unique problem, not just another dataset to process. We start with one method, but we give ourselves permission to pivot if early findings suggest a better path. Some of our best insights have emerged not from a single “perfect” study, but from stitching together results from multiple methods, with each adding a piece that the others couldn’t.

Science, Humility, and Milkshakes

This is a reminder to researchers — especially those of us with traditional training — that flexibility is not weakness. It’s wisdom. You can be rigorous without being rigid, and you can follow the scientific method while letting your methods evolve, as long as you’re critically evaluating your data along the way. Most importantly, you don’t have to know why people are doing what they’re doing right away. Sometimes you just have to go outside and ask.

The trick isn’t having the right tool from the start. It’s knowing when to reach for a different one. In the end, the data are only as good as the questions being asked, and our willingness to ask them in more than one way. Sometimes that means getting back to basics with primary research: setting aside dashboards and models, and simply asking people how they feel about something. Nothing fancy, just humans talking to humans about their preferences, needs, and choices.