The Flywheel Effect: Jim Collins on Why There Is No One Big Push

Collins needed an image for something the data kept showing: great results took a long time to arrive, and nobody inside the company could point to the moment they started. The image he reached for was physical.

After a lot of work you get one giant slow creaky turn. Then two. Then four, and they start to build on each other, then eight and sixteen and thirty-two, then a thousand, then a hundred thousand. At some point the cumulative momentum produces something that looks from the outside like a breakthrough.

The word carrying the most weight in that sentence is direction, and Collins is emphatic that the pushing is not random.

So the flywheel is downstream of the hedgehog concept. Momentum in the wrong direction is worse than no momentum, because it takes longer to stop than it did to start.

Most explanations of the flywheel stop at the metaphor, which is the least useful part of it. Collins says so himself: people talk about flywheels, we have a flywheel, whatever. The principle is generic. The value only appears when you can name the specific loop your own business runs on.

He learned that from watching someone do it better than he had. Around the publication of Good to Great in 2001, Collins was invited up to Amazon, coming out of the dot-com collapse, and taught them the principles.

What Amazon did next is the part Collins credits them for. They took the principle and made it their own, by getting specific about what their flywheel actually was. He narrates the loop from memory: lower prices on more offerings, so you almost can't help but increase customer visits, which almost can't help but bring more third-party sellers, which expands the store and extends distribution, which grows revenue against fixed costs, which funds lower prices on more offerings again.

Every step is a because, not a wish. That is the test of a real flywheel. If you can move from one node to the next and honestly say you almost cannot help it, the loop is doing work. If a step needs a heroic effort every time, it is a plan, not a flywheel.

The comparison companies in the study had a distinct pattern, and Collins named it alongside the flywheel. They bought into dramatic moments: radical transformations, cultural revolutions, savior chief executives, anything that promised to jump straight to breakthrough in one big step instead of building the wheel. That is the doom loop, and the two ideas share a chapter.

The third failure is the real one, and it is the mirror image of the doom loop. Persistence is the mechanism here, which means the model has no built-in way to tell you when to stop. Collins studied companies that were right about their direction; the ones that pushed a wrong wheel for a decade did not make the sample. Pair it with a decision rule that can fire, like Annie Duke on kill criteria, so that keep pushing stays a choice rather than a default.