The Jeff Bezos Napkin and the Art of Thinking in Loops
27-11-2025 · 9 min read · By Anshul Garg
In 2013, a startup called Homejoy had what looked like the perfect growth engine. Lower prices attracted more customers. More customers attracted more cleaners. More cleaners meant better availability. Better availability attracted even more customers. The loop was spinning. Venture capitalists saw the loop and poured in $40 million. The founders saw the loop and started planning international expansion.
Eighteen months later, Homejoy was dead.
The loop was real. But hidden inside it was a friction nobody had drawn on the whiteboard: customers were booking a cleaner through Homejoy once, then hiring them directly. Every revolution of the loop was training customers to bypass the loop. The machine was building the thing that would destroy it.
Now consider another loop — drawn on a napkin in 2001 by Jeff Bezos, in a restaurant nobody remembers the name of. Lower prices lead to more customers. More customers lead to more sellers. More sellers lead to more selection and competition. More competition leads to lower prices. Two decades and $1.5 trillion later, that napkin sketch might be the most valuable doodle in business history.
Same structure. Opposite outcome. The difference wasn't the loop — it was whether the loop's designer understood where the friction lived. Thinking in loops instead of lines is one of the most underused mental tools available to anyone trying to build anything. But a loop without a friction audit is just a prettier way to plan your own destruction.
The Straight-Line Trap
Most people think in straight lines. Input leads to output. Cause leads to effect. Push a lever, get a result. This is how we're taught to plan, budget, and forecast: A causes B, B causes C, done.
Straight-line thinking works for simple problems. You're hungry, so you eat. Your tire is flat, so you replace it. The problem has a beginning, a middle, and an end. Linear.
But almost nothing important in business, health, relationships, or personal growth works this way. The interesting parts of life are circular. The output of the system becomes the input for the next cycle. The result of your action changes the conditions for your next action.
When you exercise, you get fitter. When you're fitter, exercise feels easier. When exercise feels easier, you exercise more. That's not a line — it's a loop. And it explains why the first two weeks of a fitness habit are brutally hard while month six feels almost automatic. The loop hasn't changed. You just crossed the threshold where it started working for you instead of against you.
Why Linear Plans Fail
A startup writes a business plan: we'll launch the product, acquire 10,000 users in Q1, monetise in Q2, hire a sales team in Q3, expand internationally in Q4. Each step follows the last in a neat sequence. Each step assumes the conditions created by the previous step will remain stable.
They won't. The moment you launch the product, users generate feedback. Feedback changes the product. The changed product attracts different users. Different users generate different feedback. The market you thought you were entering has shifted because you entered it.
Every plan is a straight line drawn through a curved world. The plan isn't wrong — it's incomplete. It describes one pass through the system and assumes the system will hold still for the second pass. Systems never hold still. They loop.
The Anatomy of a Flywheel
A flywheel is a specific type of loop — a reinforcing feedback loop where each step amplifies the next. The distinguishing feature of a flywheel isn't that it works. It's that it works better each time around.
The Amazon Flywheel, Decoded
Let's slow down Bezos's napkin:
- Lower prices attract more customers (because who doesn't want a better deal?).
- More customers attract more third-party sellers (because sellers go where the buyers are).
- More sellers create more selection and price competition (because supply follows demand).
- More selection and lower prices attract even more customers.
Notice what's missing from this loop: advertising budgets, sales targets, strategic partnerships, product launches. Those are tactics. The flywheel is the structure underneath the tactics — the invisible machine that makes every tactic more effective each quarter.
Amazon's advertising spend per customer has actually decreased over time. Not because they stopped marketing, but because the flywheel generates its own momentum. Happy customers tell other customers. More sellers improve selection without Amazon sourcing anything. The loop does the work.
Jim Collins and the Push
Jim Collins, who coined the business use of "flywheel" in Good to Great, describes it as a massive, heavy wheel on an axle. The first push barely moves it. The second push moves it a bit more. No single push is dramatic. But cumulative pushes, in a consistent direction, eventually create unstoppable momentum.
The key insight: no single push produces the result. It's the compounding of consistent pushes in the same direction. Most people give up because any individual push feels insignificant. They can't see the loop — they only see the push.
This is why patience looks like a virtue when it's actually a strategy. Every person who built something significant — a career, a company, a body of creative work — was pushing a flywheel that nobody could see spinning in the early months.
Virtuous Circles and Vicious Ones
Reinforcing loops don't care about your intentions. They amplify whatever direction they're spinning. Spin them forward, and you get a virtuous circle. Spin them backward, and you get a vicious circle. The mechanics are identical. The direction is everything.
The Confidence Loop
Consider a student who does well on a test. Good performance builds confidence. Confidence increases effort and engagement. More effort leads to better performance on the next test. The loop spins forward — each success makes the next success more likely.
Now consider the student who fails. Poor performance erodes confidence. Lower confidence reduces effort ("why bother, I'm bad at this"). Less effort produces worse results. The loop spins backward — each failure makes the next failure more likely.
Same loop. Same mechanism. Opposite direction. And here's the brutal part: the student isn't aware they're in a loop. The confident student thinks they're "naturally good at this." The struggling student thinks they're "just not a math person." Both are attributing to fixed identity what is actually a structural feedback dynamic.
The Trust Doom Loop
In organisations, trust works the same way. When trust is high, people share information freely. Shared information leads to better decisions. Better decisions build more trust. The loop spins forward.
When trust is low, people hoard information. Hoarded information leads to duplicated effort and bad decisions. Bad decisions erode trust further. The loop spins backward.
Most managers who inherit a low-trust team try to fix it by demanding transparency — "we need to communicate better." But demanding transparency in a low-trust environment triggers the opposite response: people see the demand as surveillance, which erodes trust further. You can't fix a vicious circle by pushing harder in the same direction. You have to intervene at a different point in the loop.
Every outcome you see — a thriving business, a failing relationship, a career on a trajectory — is not a snapshot. It's a loop. And the most important question isn't "where is it now?" but "which direction is it spinning?"
How to Draw Your Own Loops
The practical power of loop thinking isn't just understanding Amazon's strategy. It's seeing the loops in your own life and deliberately designing them.
Step 1: Find the Stocks
What are you trying to grow or maintain? Your savings. Your audience. Your knowledge. Your health. Your network. Pick one.
Step 2: Trace the Flows
What increases it? What decreases it? For your professional network: meeting people increases it, losing touch decreases it. But go further — what makes you more likely to meet people? Attending events. What makes you more likely to attend events? Having a strong network that invites you. That's a loop.
Step 3: Find the Reinforcing Mechanism
Where does the output feed back into the input? For a writer: writing consistently improves your skill. Better skill produces better posts. Better posts attract a larger audience. A larger audience creates accountability to keep writing. Writing consistently improves your skill. That's the flywheel.
Step 4: Identify the Friction
Every flywheel has friction — the things that slow the loop or stop it entirely. For the writing flywheel, friction might be perfectionism (you don't publish), inconsistency (you break the cadence), or distraction (you switch to a new project before the loop has time to compound).
Your job isn't to push harder. It's to reduce friction. Lower the barrier to the next revolution. A writer who publishes a mediocre essay every week will build a faster flywheel than one who publishes a masterpiece every six months — because the loop only compounds when it's spinning.
Step 5: Be Patient With the First Revolutions
This is where most people quit. The first turns of any flywheel are agonisingly slow. You publish and nobody reads it. You network and nothing comes of it. You invest and the returns are invisible. The loop hasn't kicked in yet because the output is too small to meaningfully affect the input.
But there's a threshold — Collins calls it the breakthrough moment — where the accumulated momentum becomes self-sustaining. The loop starts pulling you forward instead of requiring you to push. You can't predict exactly when it happens. You can only keep pushing consistently until it does.
The napkin that Bezos drew in 2001 wasn't a prediction. He couldn't have known that Amazon would become a cloud computing giant, a streaming platform, or a logistics empire. The napkin didn't describe those things. It described something more fundamental — a structure that, once set in motion, would find its own path.
That's what loops do. They don't tell you where you'll end up. They tell you that wherever you end up will be further than you could have gotten in a straight line — because each revolution builds on the last in a way that linear plans never can.
So the question isn't whether you have a plan. It's whether the last step of your plan connects back to the first. Where does the output become the input? If you can find that connection and strengthen it, you don't need a better plan. You need a better loop.
And then you need the patience to keep pushing it when nobody, including you, can see it move.