Network Effects. Do Things That Compound

I was talking to a founder recently who was frustrated. His startup was growing, adding thousands of users each month, but it felt like pushing a boulder uphill. 

The moment he stopped buying ads, growth stalled. "I don't get it," he said. "We're doing everything right."

The problem wasn't what he was doing. It was what he wasn't building: network effects.

Most founders confuse growth with compounding, and it's an expensive mistake. You can buy users. You can't buy the thing that makes each new user valuable to all the others. That's something you have to design from the beginning.

In 1878, there were exactly 21 telephones in the world. 

Imagine owning one. 

It would have been almost completely useless, you could only call 20 other people, and they probably weren't the people you wanted to talk to. I remember getting my first mobile phone, and there weren't many people to call; I even went on the newspaper to find numbers.

What's fascinating is that with each 22nd phone sold, the value of the preceding 21 phones went up. This is because every new phone didn't just add another device, but unlocked 21 new potential conversations.

A century later, there were half a billion phones. The technology hadn't changed that much, better materials, clearer signals, but fundamentally the same idea. What changed was the network. Every new phone made every existing phone more useful.

That's a network effect. And once you understand them, you start seeing them everywhere.

Why Networks Win

Most products improve linearly. You add a feature, you get a bit more value. You fix a bug, the experience gets slightly better. It's like building with Lego blocks, each piece adds exactly one piece worth of value.

Network effects are different. They improve quadratically, or at least super-linearly. The value isn't in the product itself; it's emergence from the connections between users. It's the difference between building a better hammer and building a city. The hammer is just a hammer. The city becomes something larger than the sum of its parts.

This is why Facebook could beat Friendster despite launching years later with inferior technology. It's why eBay crushed better-designed auction sites. It's why Slack became a verb even though there were plenty of good chat apps before it.

When you have real network effects, you're not just building a product. You're building physics.

The same force that makes network effects unstoppable in the endgame makes them nearly impossible to start.

Who wants the first fax machine? Or the second? Or the tenth? Until you reach some critical mass, the product is genuinely not very useful. This is why so many potentially great network-effect businesses die in infancy. Founders look at their flat engagement numbers and assume they're building something nobody wants.

But flat engagement is the default state before critical mass. Your job is not to get exponential growth from day one. Your job is to survive long enough for the physics to kick in.

The best founders treat this as an engineering problem, not a marketing problem. 

Here are some patterns I've noticed:

Give the product utility even when it's empty. Dropbox synced your files whether or not your friends used it. The collaboration features were a bonus, not the core value proposition.

Start impossibly narrow. Facebook didn't launch to everyone; it launched to Harvard. Then Yale. Then a few more schools. Networks need density more than breadth. Better to be the dominant player in a small pond than invisible in an ocean.

Subsidize one side of the market. Adobe gave away PDF readers so people would buy PDF writers. Credit card companies give consumers rewards so merchants will pay fees. Sometimes you need to pay people to create the conditions for network effects to emerge.

And sometimes, though this feels slightly dishonest, you need to fake it till you make it. The Reddit founders posted under dozens of fake accounts until real users showed up. It's like throwing a party: nobody wants to be the first person there, but everyone wants to go to the party that's already alive.

I won’t advise doing this, but you can get actual volunteers and not fake accounts. 

You'll know when you've crossed the threshold. It doesn't happen gradually—it happens all at once.

Your retention curves suddenly bend upward. Your support tickets change from "This is lonely" to "This is overwhelming." Competitors raise huge rounds and still can't catch up, because your users now get more value from your network than from any feature those competitors can ship.

Marc Andreessen calls this "escape velocity." It's a good metaphor. Before the tipping point, you're fighting gravity. After it, growth feels like falling downhill.

But networks can also rot. I remember reading about Usenet in its early days. It was like having access to the world's smartest people. Then came the spam, the trolls, the endless flame wars. By the end, it was unusable.

The same thing happened to MySpace. Early on, it felt like digital self-expression. Later, profiles became cluttered billboards, and the whole experience turned into visual chaos.

The lesson is that growth isn't always good. Networks need curation. They need governance. You have to treat this as a product problem: reputation systems, algorithmic filtering, invite-only stages, whatever it takes to ensure that the average user experience improves as the network scales.

How to Tell If You Really Have Network Effects

Most founders think they have network effects when they don't. Here's the test: if you could clone your product exactly but with zero users, would it be just as useful?

If the answer is yes, you don't have network effects. You might have a great product—superior technology, better design, lower costs—but you don't have network effects.

Real network effects show up in the data. Time to find a match in your marketplace should decrease as you add users. Retention of newer cohorts should be higher than older ones. The percentage of users who arrive organically should increase over time.

If these metrics aren't improving, you're not building network effects. You're just buying growth and hoping it sticks.

Once you have real network effects, you become very hard to kill. Competitors can copy your features, but they can't copy your users. And users, unlike features, take years to accumulate.

The only way to attack a network effect is usually to change the game entirely. Instagram didn't beat Facebook by building a better Facebook; they built something different around photos. Snapchat didn't beat Instagram by building better photo sharing; they built something different around messages. TikTok didn't beat Snapchat by building better messaging; they built something different around short videos.

Each new platform found a different atomic unit, the fundamental thing that users were connecting around and built network effects from there.

I think network effects are the closest thing in business to perpetual motion. 

Most companies have to keep pushing to stay in the same place. They need constant marketing to acquire users, constant development to stay competitive, constant fundraising to fuel growth.

But when you have real network effects, your users do the work for you. They invite their friends because the product is more useful with their friends on it. They create content because other users consume it. They stay because leaving means losing access to the network.

It's a beautiful thing when it works. Build something that gets better the more people use it, and you won't just grow, you will compound.

first published on My Website