Where the Money Moves

Interdependence vs. Modularity

Imagine you're building a house out of marble. Every wall is carved from a single slab. The arch supports the ceiling which supports the roof. If you want to change the kitchen layout, you might have to rebuild half the house — because everything depends on everything else.

Now imagine building a house out of LEGO. Every brick connects to every other brick through the same standard interface. Want to change the kitchen? Pull out those bricks, snap in new ones. The rest of the house doesn't care.

That's the difference between interdependent and modular architecture — and it's the most important distinction in business strategy that most people have never heard of.

These aren't two categories — they're the ends of a spectrum. Every product, service, and industry sits somewhere between purely interdependent and purely modular. And where it sits has enormous implications for strategy.

The place where any two components fit together is called an interface. When interfaces are unpredictable (you don't know how changing one side will affect the other), the architecture is interdependent. When interfaces are well-defined and standardized, the architecture is modular.

Think about USB ports. Before USB was standardized, every computer company had its own proprietary connectors. If you bought a peripheral, it only worked with that one brand. The interfaces were undefined — interdependent.

Then the USB standard emerged. Suddenly, the interfaces became perfectly defined. Any device could plug into any computer. That single act of standardization allowed the entire computer industry to dis-integrate. Companies could specialize in making just one piece — monitors, keyboards, mice, external drives — and it would all work together.

LEGOs are the purest example of modularity. Every brick connects to every other brick through the same standard interface. You can build anything without asking permission from any other piece. As Christensen put it, the modularity of LEGOs allows you to do almost anything you want — and it engages grandparents like him.

The Diagram That Explains Everything

If you've read Post 1 in this series, you've already seen a powerful diagram: the one that shows how technology overshoots what customers can absorb, creating the conditions for disruption.

Now we're about to add a new layer to that same diagram — and it changes everything.

In the early months and years of any industry, the product is not good enough. It doesn't do what customers need. When that's the case, an interdependent architecture develops naturally. Whatever company integrates fastest and best takes over — because integration is the only way to wring maximum performance out of a not-good-enough product.

But as technology improves and the product becomes more than good enough, something profound happens. The basis of competition shifts. Customers no longer care about incremental performance gains — they care about speed, customization, convenience, and price. And the only way to deliver those things is through modularity.

Look at smartphones. In the early stages, producers were obsessed with the overall functionality — getting the phone to access the internet at all. The architecture was deeply interdependent. Apple succeeded because they controlled everything: hardware, software, chips, and services, all optimized to work together.

Over time, as smartphones became "good enough" for the average consumer, the basis of competition shifted. Now it's about being quick to market, allowing customization, and responding to what consumers actually want. The Android ecosystem — modular by design — enabled hundreds of manufacturers to compete on speed and customization.

This pattern repeats across every industry. And the critical realization is this: you don't get to choose when it happens. The market shifts with or without you.

The Overhead That Kills You

Christensen once spent an entire week standing in the middle of a giant engineering room at General Motors' steering gear division. Because he was tall, he could see what everybody was doing.

What he observed was remarkable. The engineers were organized into groups by component — one group made hydraulic pumps, another made hoses, another did tie rods. And the way they interacted with each other was entirely determined by the architecture of the product.

When there was interdependence between what one group did and what another had to do, you could see people going back and forth all day, every day. They constantly needed to understand: "If I change this, what do you have to change to accommodate it?"

But when the components didn't interface with each other — when the interfaces were well-defined — the engineers responsible for those parts didn't ever need to talk to each other.

Here's why this matters for your income statement:

When the product isn't good enough and the architecture is interdependent, you need overhead to coordinate how all the teams interact. That overhead is essential — it's how you optimize functionality and reliability.

But when the market evolves toward modularity, those development teams don't need to interface anymore. The overhead that enabled coordination now just slows you down. You don't have to do everything to be successful — you just have to be successful at one piece of the system.

Banking: The Strategy Table in Action

To see the strategy table play out in real life, consider the banking industry.

At the beginning, banks had to do everything in order to play in the game. They were integrated because if you were going to deliver a loan, you needed to get the deposits, process transactions, close the loan (which required lawyers), service the loan every month (send out a bill, track who pays and who doesn't), and handle collections when borrowers fell behind. If they didn't do some of these components, they couldn't do any of the components.

But as banks got better at their job, something familiar happened. At the bottom of the market, a set of modular players emerged — companies that weren't integrated. Step by step, the banking industry became more and more modular. There are now companies that just close loans. Other companies that just service loans. Every piece of the system has been taken by individual, focused companies.

These specialist companies are very fast — they get efficient at their one piece. They can customize what they offer to individual customers. And because they don't carry the overhead of a full integrated bank, their costs are lower.

What has been the reaction of the big banks? Exactly what the theory predicts: they went upmarket. They moved to bigger and bigger transactions — deals that are very complicated and very nonstandard. The margins on those deals are large enough to cover the cost of being an integrated bank. But the modular players at the bottom just keep getting better and better at defining how the pieces fit together.

The result? The traditional integrated banks are slowly being eaten from below — the same pattern you'll see next in the music industry, and the same pattern disruption theory predicts.

From Vinyl to Streaming: A Century of Architectural Shift

The music industry is one of the best case studies for understanding interdependence and modularity because it shows the entire journey — from deeply interdependent to radically modular — playing out over more than a century.

The Interdependent Era: 1880s through the CD Age

Up through the 1880s, the music industry was completely interdependent. You had composers, performers, and venues — and what you could do in one depended on what you were doing in the other. A handful of people dominated the industry.

The industry really took off when RCA purchased the Victor Talking Machine Company — putting together all the pieces of the system required to make recorded music a viable industry. That acquisition was integration in action: no single piece could succeed alone, so RCA had to own the whole thing.

Then the technology kept evolving — but the fundamental structure barely changed:

• 1931: RCA's researchers developed the first vinyl record. A piece of music couldn't last more than 4 minutes before you hit the end of the vinyl and had to flip it. Deeply interdependent.
• 1950s: The disk speed dropped from 78 RPM to 45 RPM, fitting more music per side.
• 1960s: Down to 33 RPM — even more music per disk.
• 1964: The cassette tape arrived — 45 minutes of music on a single tape.
• ~1980: Sony and Philips developed the CD — 80+ minutes of music, double the cassette.

Here's what's remarkable: while the industry was growing at a rapid rate, the fundamental structure didn't change. The way performers had to interact with the technology, the labels, and the distribution channels — that didn't change much at all. It was still very interdependent.

Why the Labels Were Unbeatable

If you wanted to build a career as a musician, you couldn't do very much to affect that on your own. There were just a few labels, and you had to be on one of them. Those labels had interdependencies with the performance venues, the recording technology, the radio stations, and the devices customers owned. Everything was coupled.

CDs typically sold for $12 to $20 apiece. The cost of making them was in the low single digits. That enormous gross margin covered all the overhead required to manage the interdependencies — the A&R teams, the distribution networks, the radio promotion, the marketing machines.

The Digital Revolution: Modularity Spreads Like Wildfire

Then the transformation from analog to digital happened — and modularity spread like wildfire across the music industry.

Sales of physical media began to decline dramatically. You could get files online. And where the ability to make money shifted was to live performance. The performer taking themselves on the road and filling up venues — that became the piece of the business that made the money.

Today, the music industry is very modular. And the integrated companies that dominated for a century? They've been struggling ever since.

Taylor Davis: The One-Person Record Label

To make modularity concrete, consider Taylor Davis — a professional violinist and independent recording artist who embodies everything the theory predicts.

Growing up, Davis was told the only way to make a living playing an instrument was to either teach or get into an orchestra. And with major orchestras facing cutbacks, even that path was closing. The music world was interdependent, and the gates were guarded by a few powerful labels.

Then she discovered something: people were making music independently from their homes. Consumer-level technology was available. It wasn't expensive. With very little investment, she decided to give it a try.

The contrast with the 1980s is staggering:

Davis uses a MIDI keyboard that plugs right into her computer. She can assign those notes to any instrument — a French horn, a cello, drums. The results are so realistic that people sometimes can't tell the difference between her computer-generated orchestration and a live recording.

When she wants to distribute, she logs on to TuneCore, clicks which stores she wants (iTunes, Amazon, Spotify), and within a week her music is available worldwide — for a flat annual fee, no percentage taken. She can literally climb out of bed, walk downstairs, make an album, and have it on iTunes that same day.

Social networks are the main reason Davis exists as a musician. She's never paid for advertising. Everything has spread through sharing — people clicking a share button and showing their friends. The integrated overhead that labels once needed to handle promotion, distribution, and marketing? Each function is now handled by a specialized, modular player.

The Cost of Not Integrating Enough

We've been talking about the natural drift toward modularity. But there's an equally dangerous mistake: not being integrated enough when the architecture demands it.

Three case studies illustrate what happens when companies try to be specialized in a world that still requires integration.

Mayo Clinic: Integration Done Right

Our healthcare system is organized around components — specialists in the heart, lungs, digestive system, bones. Each specialist is excellent at their piece. But they don't have ways to integrate their insights to understand what's going on in the life of a complicated patient.

The exception is the Mayo Clinic in Rochester, Minnesota. They're organized to integrate the disparate perspectives — combining insights from different specialists to get a whole-system view of what's happening with a patient.

Nobody can compete with the Mayo Clinic. And the reason isn't that they have the best specialists. It's that they have a way to integrate the insights from the specialists to get a complete picture. When the "product" (patient care for complex cases) isn't good enough, integration is how you optimize functionality and reliability.

DuPont's Kevlar: The Failure That Taught a Lesson

When Kevlar emerged, it was better than nylon and polyester by almost every dimension. DuPont thought they could just swap it into tires — replace the existing cord material and everything would be fine.

It turned out there were so many subtle interdependencies between the properties of Kevlar and how well it performed in a tire on the road that you had to change all kinds of things in the design of the car and the tire to accommodate Kevlar's unique properties.

For DuPont to succeed with Kevlar in tires, they needed to integrate forward into tire design — taking responsibility for all those interdependencies themselves. They weren't willing to do that. They spent a lot of money, and in the end, they failed.

But here's the twist: where Kevlar did succeed was in applications where the interfaces were already modular. Bulletproof vests were the first — you could take out the current material (polyester and cotton) and put in Kevlar without redesigning anything else. Step by step, DuPont found more and more of those "plug and play" applications. Ultimately, Kevlar became a wildly successful product — just not in the interdependent world of tires.

RCA and Color TV: The Chicken-and-Egg Problem

RCA made the first color televisions, and everybody knew color TV would blow black and white out of the water. But despite having the best product, nobody would buy RCA's televisions.

The reason? Nobody was broadcasting in color. And nobody was broadcasting in color because nobody had color TVs in their homes. A classic chicken-and-egg problem — caused by interdependence between organizations.

The solution? RCA and NBC merged together. Because they were solving the whole problem — both the TV and the broadcasting — color televisions exploded. If you didn't put your arms around the whole problem, a market wouldn't emerge.

The IBM Story: Forward, Then Backward

If Part 5 was about the danger of not integrating enough, Part 6 is about the equally dangerous failure to dis-integrate when the time comes. And no company illustrates this better than IBM.

The Chain Reaction Forward

In 1952, a group of engineers in IBM's San Jose lab invented the world's first disk heads. Great invention for data storage. But nobody would buy them — because nobody was making disk drives that needed heads.

So IBM had to integrate forward into making disk drives, to create a market for their heads. But nobody would buy the disk drives either — because nobody was making computers that needed disk drives.

So IBM integrated forward again into making computers. But still nobody would buy, because nobody was writing software to run on computers. So IBM integrated forward into software.

And even then nobody would buy — because these early systems crashed nearly every day and customers needed someone to fix them. So IBM had to build a service organization too.

Each step forced the next, stretching IBM further and further from their original invention until they finally reached the customer — what Christensen calls the decoupling point.

The Painful Retreat

As the industry matured, interfaces became better defined. Service companies emerged that could handle computer maintenance. IBM no longer needed to provide service. Then it became clear how software should interface with hardware — so IBM didn't need to make software. Then disk drive interfaces standardized. Then head-and-disk interfaces.

The decoupling point retreated, step by step, back toward IBM's origin.

But here's the brutal truth: dis-integrating is much harder than integrating. When the decoupling point comes back toward you, you have all these costs you need to jettison — but those costs are embedded in your organization. The people, the processes, the overhead. IBM's profitability went off a cliff. They ultimately had to exit the disk drive business entirely.

Skating to Where the Money Will Be

As modularity advances and specialized players emerge at the bottom of the market, something frightening happens to the integrated companies at the top: commoditization.

It develops for two reasons. First, the product becomes more than good enough — customers underneath the integrated companies are no longer willing to pay a premium for better products. Second, modularity defines how the pieces perform — so the ability to differentiate becomes minimized. Prices drop. Nobody makes money.

But here's the critical insight that Christensen emphasizes: profit almost never completely disappears from an industry. Rather, where you can make money shifts to different places in the value chain over time.

Being able to recognize that your current business is moving toward commoditization allows you to shift your focus to the performance-defining component — the piece of the system that provides the functionality customers care most about.

Higher Education: A Personal Example

Christensen used his own industry — education — as an example. Historically, where the money was made in education was in teaching. Some students were MBA candidates, some were executives, but the money was at the level of the performance-defining subsystem: the teacher in the classroom.

Then online learning started getting better and better. Many more people began offering the materials that elite universities had monopolized. The ability to make money in education began shifting away from teaching — and toward the products that enable a larger population of teachers to do better and better at teaching.

The money didn't disappear. It moved. It moved from the teaching itself to the tools, platforms, and content systems that make teaching scalable. The successful players aren't necessarily the best teachers — they're the ones who recognized the shift and skated to where the money was going.

Your Defense: Organize Around the Job to Be Done

So how do you defend yourself against commoditization? The answer connects directly back to Post 3 — Jobs to Be Done.

Literally everybody doesn't want to be commoditized. Literally everybody doesn't want to be disrupted. The defense is to integrate your company around a job to be done.

If all you're doing is making a product, disruptive competitors can copy your product. They will disrupt you and cause you to live a life of commoditization. But if you organize your system around a job to be done, differentiation is natural.

Why? Because understanding the job tells you what experiences you need to provide. Understanding the experiences tells you what to integrate and how to integrate it. And that integration — the way you've organized your processes around the job — is what's hardest for attacking disruptive companies to copy.

This is the complete picture. The money moves from commoditized areas to the performance-defining component. Your defense is to organize around the job to be done — which naturally creates the kind of integration that competitors find hardest to replicate.

Remember Christensen's framing: "We're successful because we're here — at this point in the industry's structure. Not because we're smarter than everyone else." If the ground shifts, you have to shift with it. The world changes. You need to go where the money will be made.