Flywheels: building a business where every turn pays for the next one

The land under the hamburger stand

Ray Kroc's first McDonald's franchise deals paid him a royalty of 1.9% of each restaurant's sales, and he had to pass part of it on to the McDonald brothers. It was too thin to pay for growth. Every new restaurant made the chain bigger without making Kroc's business any stronger.

Harry Sonneborn, a former finance executive at Tastee-Freez, proposed a different deal. McDonald's would control the land under each restaurant and lease it to the franchisee at a markup. The real estate company that did it was incorporated in 1956.

Now each new restaurant added rent, and the rent, along with the land itself, could pay for the next site. The markup started at 20 percent of lease costs and was later raised to 40 percent. By 1960 the restaurants were grossing roughly $56 million a year across some 200-plus locations.

The model never left. At the end of 2025 about 95% of McDonald's restaurants were franchised, and its real estate holdings were worth $37.7 billion on the balance sheet.

That is a flywheel: a loop where each turn produces what powers the next one. The word is now used for almost any growth plan with arrows in a circle. This guide is about the difference, and about the decisions companies made to get a real loop turning, or to keep one from stopping.

What flywheel is

The test is simple to state. In a flywheel, the second turn is cheaper than the first, because the first one paid for part of it. In an ordinary growth plan, every turn costs the same and has to be funded from outside. Both can grow. Only one gets easier.

Why it is so hard

Drawn on a slide, every loop works. In a real company, three things get in the way.

The ways companies did it

The decisions on record take four shapes. Each is a different answer to the same question: what do you give up to keep the loop turning?

What 13 companies did

Here is every decision on record, placed by the year it was made and marked by how it turned out. Hover a mark, or open the table, for the company and the outcome.

Of the 12 decisions that can be judged, 11 were never undone and 1 was walked back. Shopify's 2025 change to its developer fees appears here as too early to judge, and under Ecosystem Lock-in as held.

"Never undone" is not the same as "worked". Roblox's 2025 rise in what it pays developers was never undone either, but it still left them well below what other platforms pay. A loop like that keeps turning because of the platform's reach, not because the people feeding it are well paid, and it is only as safe as that reach.

How to run this decision

If you are weighing a version of this choice, here is what the record suggests doing, in order.

1. Write the loop as a sentence that ends where it started

   More sites, more rent, more sites. Better pricing, better drivers, better data, better pricing. If the sentence cannot close without adding money or effort from outside, it is a growth plan, and it should be funded as one.

2. Check that the second turn is cheaper than the first

   This is the whole test. In a real loop the cost of the next customer, seller or site falls because the last one paid for part of it. If each turn costs the same or more, the diagram is decoration.

3. Find anything that shrinks as you succeed

   TikTok's was the pay per creator, because the pool was fixed and the number of creators was not. Anything divided by a growing number belongs on this list, and a loop that contains one will slow itself down.

   Seen in: TikTok, 2020
4. Own what the loop leaves behind

   The land under the restaurants, the data behind the loyalty card. The loops on record that lasted built up an asset the company controlled, not just momentum. Momentum stops when you stop pushing; an asset keeps paying.

   Seen in: McDonald's, 1956 · Tesco, 2001
5. Decide in advance which side you protect

   The two sides of a loop will eventually want opposite things. Make the rule for that day while things are calm, and set aside what it will cost to keep the other side on board, so the call is not made for the first time in a crisis.

   Seen in: Airbnb, 2020
6. Get one side in place before you open the other

   Sellers need buyers before they will build, and buyers need sellers before they will come. Shopify waited three years after launch to open its App Store to developers, so that they arrived to a market that already existed.

   Seen in: Shopify, 2009 · PayPal, 2015
7. Track one number per turn, on a schedule

   The cost of the next unit, or the rate at which the asset grows. One number, one owner, one date to check it. Loops fail quietly, because the total keeps rising after the mechanism has stopped.

Warning signs

What it has looked like when this decision was going wrong.

• A fixed pool of money is split among a growing number of participants.TikTok, 2020
• Each participant earns less as the platform grows.TikTok, 2020
• There is no rule for which side you protect, so the first time you decide is in a crisis.Airbnb, 2020
• Suppliers keep a far smaller share than they would elsewhere, and only your reach keeps them.Roblox, 2025

All 13 cases

Each case is one company at one decision: what was on the table, what it chose, why, and what followed, with every fact sourced.

Questions

What is a flywheel in business?

A loop in which each turn produces what powers the next one. More customers draw more suppliers, who draw more customers, and each new participant costs less to win than the last. McDonald's land-and-lease model, where each restaurant's rent helped pay for the next site, is a classic example.

What is the difference between a flywheel and a growth strategy?

In a flywheel the second turn is cheaper than the first, because the first one paid for part of it. In an ordinary growth plan every turn costs the same and has to be funded from outside. Both can grow; only the flywheel gets easier as it grows.

What are examples of company flywheels?

McDonald's owning the land under its franchises, Tesco buying the analytics firm behind Clubcard, DoorDash's DashPass subscription, Shopify's merchants and app developers, and Progressive's pricing based on driving data.

Why do flywheels fail?

Usually because something in the loop shrinks as it grows. TikTok's first creator fund was a fixed pool, so each creator earned less as more joined, and TikTok replaced it. Loops also stall when the company takes too much out of the step that feeds the next turn.