Stop Calling Everything a Flywheel

Jim Collins, the Flywheel Principle, and the importance I place on questioning every idea I'm exposed to.


I hope you are safe and sound as you navigate this terrible COVID-19 crisis.

I am. One interesting side effect of working from home and leaving the house less is that I am having more time to reflect on things.

Today, I'm writing about something that happens quite often to me. It's when I consume some content - a book, an article, a blog post, a podcast episode, an interview, and so on - and question its ideas, its presentation, its logic, etc., especially when that content was produced by someone that's believed to be a genius, a master at her craft, and/or when that content describes ideas that are cited by many other people without the sort of questioning I've gone through, and without evidence of their realizing that these said ideas are deeply flawed at their core.

On a more concrete note, I'm going to go over the thinking-trip I went through when I listened to Shane Parrish’s interview with Jim Collins for The Knowledge Project, which is Parrish’s podcast as part of the Farnam Street blog [1] (very interesting for the book recommendations).

I've been a big fan of Collins for a long time. And I've been pretty interested and confused by Collins’ concept of “Flywheels,” and the intersection of that concept, in my mind, with the concepts of moats, network effects, economies of scale, and strategy.

While I was listening to Collins I had this very disappointing feeling I have sometimes when I tentatively conclude that a mind I previously revered, such as Collins’, has produced ideas with such deep logical flaws in his or her thinking that basically they, at least for me, get off of this huge pedestal I previously put them on, only to rest on “maybe-I-found-a-hole-in-her-reasoning” ground. It’s a feeling of disappointment I'm getting more used to as I allow myself to question everyone and everything. It feels a bit like when we first realize adults are just deeply flawed human beings and not the spectacular all-knowing sentient beings we thought they were when we were kids [2].

Anyway, I decided to tackle Collins and his Flywheel Concept and try to articulate why I think it contains very flawed thinking where I previously thought there was only brilliance and insight. I hope you'll pardon me for my petulance.

But why?

You might be asking yourself “why is this guy doing this?” Why bother? Why take things apart? Why criticize Collins? Why not do better?

Just to be clear, “why” deserves a two-part answer, because I'm doing two related things: the first is thinking a lot - e.g., questioning the validity - about the ideas that are presented to me. The second is taking the time and energy to write about it.

Step 1 is something I can't help, and haven't been able to help for as long as I can remember. I've questioned everything and everyone ever since I was a little kid in school. I remember teachers telling to my parents that I was a “complainer” and that it disrupted class. This left a mark on me. My mom was probably embarrassed by being confronted by the teachers (and maybe amused by my behavior) and told me I should stop complaining about everything at school. She used to ask me “have you stopped complaining?”

Anyway, the point is I can't help it. And I can tell you it's very exhaustive.

Step 2 is something I haven't done a lot. Sometimes my mind will just get lost in so much stuff I find wrong that I feel overwhelmed at the mere thought of taking the first steps needed to articulate why. Sometimes I just feel nobody will listen to me, since the world around me - Silicon Valley, tech entrepreneurship, etc. - is a big echo chamber where people parrot each other all the time and seem to never question each other publicly. If I felt Ben Horowitz would like to listen to all I think is wrong in his most recent book, I'd gladly spend an afternoon doing so. If I felt Paul Graham, whose work I only recently started to question more intensely, would like to listen to the few things I think are wrong in one of his essays, I'd also love to chat him up. I just doubt they'd welcome the criticism.

Who am I, anyway? I totally get that.

Back to the Flywheel Concept

Back to the proper subject of our essay. First, let’s see how Collins himself describes what he calls the Flywheel Concept. For that, I’m going to borrow from Collins’ own Turning the Flywheel: a Monograph to Accompany Good to Great

But first, we have to discuss Collins’ word choice and the mechanics of an actual flywheel. He picks “Flywheel” for a reason: he believes flywheels accurately describe the sort of dynamics he identifies in some very successful companies that go from being average - “good” - to being leaders - “great” -. So the first thing I tried to do was look for what a flywheel actually was. I thought I’d never seen one. Wikipedia describes a flywheel as:

"A flywheel is a mechanical device specifically designed to efficiently store rotational energy (kinetic energy). Flywheels resist changes in rotational speed by their moment of inertia. The amount of energy stored in a flywheel is proportional to the square of its rotational speed and its mass. [3]”

In plain words, a flywheel is a heavy wheel. The first example that comes to mind is the front wheel of a spinning bike. Have you ever ridden a spinning bike? Here’s a picture of a spinning bike with my name on it (just a happy coincidence) so you know what we’re talking about:

Spinning bikes were invented to resemble the experience one has when riding a real bike - one that actually moves -. When we ride a real bike, we - the us+bike mass - build momentum. If we’re on a plane road, the faster we go, the more momentum we accumulate so that the further we go if we stop pedaling.

The way spinning bike inventors found to mimic that effect was by building a very heavy front wheel that spins as we turn the pedals. As you can see from the picture, the pedal’s shaft is connected to a hidden chain that’s itself connected to the front wheel’s axis. As we pedal, we indirectly move the heavy front wheel.

The front wheel of our average spinning bike is a good example of a flywheel. It’s a heavy wheel. Because it’s heavy (I’ve always been TERRIBLE at physics, so I’ll stick to my empirical average-Joe observations), it’s a) very hard to make it spin, and b) it’s very hard to brake if it’s spinning fast, where, in both cases, "it’s very hard" equates to “it takes a lot of energy."

Flywheels and companies

Ok. So whatever Collins wants us to understand about great businesses, he thinks a flywheel is a good shortcut to get there. Or, by understanding a flywheel’s properties, we will understand properties these special companies Collins found have.

His description of a flywheel, back to Turning the Flywheel, goes like this:

“…‘the flywheel effect’ that we’d uncovered in our research. In creating a good-to-great transformation, there’s no single defining action, no grand program, no single killer innovation, no solitary lucky break, no miracle moment. Rather, it feels like turning a giant, heavy flywheel. Pushing with great effort, you get the flywheel to inch forward. You keep pushing, and with persistent effort, you get the flywheel to complete one entire turn. You don’t stop. You keep pushing. The flywheel moves a bit faster. Two turns… then four… then eight…. The flywheel builds momentum… sixteen… thirty two… moving faster … a thousand… ten thousand… a hundred thousand. Then at some point - breakthrough! The flywheel flies forward with almost unstoppable momentum.”

So for now, we can agree that Collins is describing something that resembles a flywheel. But there’s one problem, which could be innocuous if I didn’t know it wasn’t: I’m not sure you’ve noticed it, but while describing how someone turns the flywheel example above, Collins use exponentials to describe the turns of the wheel. Let’s zoom in on this part of the quote:

"you get the flywheel to complete one entire turn. You don’t stop. You keep pushing. The flywheel moves a bit faster. Two turns… then four… then eight…. The flywheel builds momentum… sixteen… thirty two… moving faster … a thousand… ten thousand… a hundred thousand”

Ok. So at the start of the process, he cites numbers of turns on the flywheel that are multiples of two: 2, 4, 8, 16, and so on. Then, when that progression reaches a thousand, it changes to a “faster” one, of multiples of ten. If we were these “turns” on a graph, here’s what it’d look like:

As you can see, the first seemingly flat part of the graph (which is exponential, but doesn't look like it is because of the graph's scale) is where the numbers are multiplied by two. Then the inclination shifts abruptly when we multiply 1.024 by ten, and then even more abruptly when we multiply the resulting number, 10.240, by ten again.

Now, this has been taken from a book that Collins published on Amazon as a follow-up to Good to Great, because it expands on a concept Good to Great touched on, The Flywheel Concept itself, which Collins felt deserved more attention. So we must presume that Collins chose his words (and examples) carefully. And if we presume that, we have to conclude that this example is illustrative of how a flywheel works, and also of how a business that leverages “The Flywheel Concept” works.

Flywheels and exponentials

But flywheels don’t work like that.

If we keep pushing at a flywheel with the same strength, it’s not going to double its speed at regular intervals. Far from it. How can we know it? Because we’ve all - presumably - ridden bikes before. If Collins’ example was true, we’d all be able to reach very high speeds on a bike before the wind held us back. Tour de France professional cyclists would be riding at hundreds of miles an hour, since they ride in packs which reduce air drag. And it just doesn’t happen. The only way a similar effect could be achieved with a flywheel is if a) it were in a vacuum and b) we had gears to leverage our strength in keeping pedaling it.

Therefore, flywheels are great at describing something that holds a lot of momentum, but not something that behaves exponentially, or that self-reinforces itself. (Please beware that by now I'm a bit tired, so there's a big chance my reasoning and writing quality will fall off a cliff.)

For now we’re done with talking about actual flywheels.

Flywheels and Amazon

Let’s check the prime (pun not intended) example that Collins uses to describe how a “Flywheel” looks like in the business world, Amazon, and see if it makes the flywheel concept make more sense.

Collins describes how Amazon’s “flywheel” works by citing Brad Stone, author of The Everything Store, a book about the company:

“As Brad Stone later wrote in The Everything Store, ‘Bezos and his lieutenants sketched their own virtuous cycle, which they believed powered their business. It went something like this: Lower prices led to more customer visits. More customers increased the volume of sales and attracted more comission-paying third-party sellers to the site. That allowed Amazon to get more out of fixed costs like the fulfillment centers and the servers needed to run the website. This greater efficiency then enabled it to lower prices further. Feed any part of this flywheel, they reasoned, and it should accelerate the loop.’ And so, the flywheel would turn, building momentum. Push the flywheel; accelerate momentum. Then repeat. Bezos, Stone continued, considered Amazon’s application of the flywheel concept ’the secret sauce.’”

I’m wondering if you see what I see, again.

There are two very different concepts being thrown around: one has to do with geometric progressions. The other has to do with momentum. They are pretty different. But I think Collins is confusing the two and he hasn’t realized it.

A flywheel has a lot of momentum. But it doesn’t really work in a way that resembles a geometric progression.

Am I crazy or have I just found a major flaw in one of Collins’ greatest contributions to management thinking, part of one of his best-sellers, which are themselves some of the world's most influential business management books? Why myself, a meager entrepreneur from Campinas, in Brazil, have found it? Why haven’t I heard anybody else point this flaw before, as they parrot left and right about how important it is to find companies that “have flywheels” in their business models?

Anyway, I don’t know. I think people just don’t question what they read. Or I’m a dumb shit, and everybody else has already gone through the path I’m going through, but they concluded they were wrong, as I should be, and moved on.

I'd bet most people think the second option when they feel like questioning something written by someone they admire, or something written by someone that everybody around them seems to admire.

Back to Amazon

Anyway, I think Amazon is a great example of something I can’t quite point at but that seems to reinforce itself (I have no evidence that it's exponential in any way. Just that it seems to go on and on forever without the need for more energy). As it sells more, it’s able to sell cheaper, which leads it to sell more, which leads it to be able to sell cheaper, and on and on.

Businesses that find such self-reinforcing “mechanics” have something strong going for them. It's like they are predestined to build a strong moat because the strengthening of their moats takes almost zero additional energy. It's pretty logical and consequential that as Amazon sells more, it's able to buy for less and sell for less, which leads quite naturally and effortlessly to more sales, and so on.

A similar type of self-reinforcing loop (let's call it a “strange loop”) tends to happen to some social networks, at least for some time [4]: the more users that join the services, the more content that's produced, the more discoverable, useful, and attractive the service becomes to its current and prospective users, the more users it attracts, and so on and so forth.

I think that's what called Collins’ attention in the first place, but it's also what Collins messed up when he used a flywheel to describe it (and when he picked some unfortunate examples to illustrate it).

As we’ve seen, on the one hand, flywheels have a lot of momentum but aren’t exponential in any way. Amazon, on the other hand, has an interesting exponential dynamic to its business (the “strange loop”) but doesn’t necessarily have momentum (at least, Collins doesn’t talk explicitly about it). So if you are like me and read these books to try to improve your business, it’s critical to separate momentum and geometric progression for proper understanding and application of these concepts.

Shopping malls

If we look for flywheel like businesses around us, they are quite easy to find. Shopping malls seem to be a good example. Building a mall takes a lot of energy: You need a ton of capital, land, licenses, and then tenants to rent it and attract people to it. Once you spend that amazing amount of energy to put the mall into motion - in successful operation, with low vacancy and good foot traffic - it has great momentum [5], in that people keep visiting it, stores keep selling stuff, rents keep getting paid, and so on and so forth. You could even argue that a shopping mall needs very little energy to keep working for some time, just like a flywheel needs very little energy to keep spinning for some time: you basically need some security guards, a reasonably small cleaning staff, some changing of light bulbs… Essentially, the thing goes on even if management decides to take a three-month vacation to some remote country with no cell phone reception and no internet.

Shopping malls and Amazon

Now, shopping malls are not exponential in any obvious way, or at least in any way that resembles the “strange loop” that Amazon seems to benefit itself from.

You could make a very loose and weak argument that goes like this: “The more stores that rent spaces in the mall, the more visitors that are attracted, the more sales that take place, the more stores that rent spaces in the mall, and so on and so forth” but I guess that would be a Grand Canyon of a stretch. As is an example that Collins gives when asked by Parrish if he has a “personal flywheel” going for him: Collins answers something along the lines of "actually I do have. Let me illustrate it for you: when I finished Built to Last I told my wife I was going to pour all the money I was making with Built to Last into a next big project, which turned out to be Good to Great”. That’s his example of a flywheel that’s at work in his own personal business, which is doing research and selling content and consulting.

Now, can you put *that* in the same shopping cart (pun intended now) as Amazon’s “strange loop”? Or as the shopping mall's flywheel? Hardly.

I don’t know about you, but I got saturated at that point. Here’s this guy who’s arguably the new Peter Drucker, revered by entrepreneurs, world leaders and executives alike, not only basing a huge part of his knowledge production on a shaky concept that’s named and explained with a shaky mix of words and examples applying it all in a very shaky way to his own life. WTF?

Why this matters?

I hope by now you have some interesting weaponry for a (Zoom?) business cocktail party. But more than that, I hope you also start questioning the stuff you read, so as not to parrot things that don't make sense. If you are from Silicon Valley, you probably need a double dose.

If you ever feel compelled to describe Amazon as a “flywheel,” please don't. Pick another analogy. And if you ever feel compelled to describe a business such as a mall, pick “flywheel.”

Happy quarantining,


PS: If you found a hole in my argument, PLEASE tell me. I *will* love it.


[1] I went to look for what Farnam Street was and it turns out that’s the street where Berkshire Hathaway’s main office is located in Omaha, Nebraska.

[2] The same thing happened a few weeks ago when I was rereading Ben Horowitz’s What you do is who you are, and thought I had identified so many flawed conclusions and plain bad writing that I couldn’t even start to articulate it.

[3] Taken from

[4] Please don’t use the COVID-19 crisis as a counterexample. Let’s stick to a normal working world for now.

[5] It’s critical to note that these seemingly viral or exponential motions - “strange loops” - can’t go on forever. In Amazon’s case, costs go down and that leads to it being able to charge lower prices which leads it to being able to attract more customers and sell more units. A critical step is how volume leads to lower costs. It happens in two ways: first, by lowering relative fixed costs, and second, by raising bargaining power with suppliers (back to Porter…). As it sells more widgets, the portion of fixed costs (e.g., Bezos’ salary) diluted to each widget becomes lower per widget sold. Also, the more widgets Amazon sells, the more widgets Amazon buys, and therefore the more it can use its size and power to bargain for lower prices from suppliers. But these two cycles’ curves reach asymptotes: As the number of widgets sold becomes very high, the amount of fixed costs the company dilutes in any nth widget tends to zero, and therefore the prices can’t be meaningfully lowered anymore. Also, suppliers can only lower their prices so much before going out of business or just going out of that business.