The Cold Start Problem: How to Start and Scale Network Effects

by Andrew Chen

Eventually, the difference between two minutes to get a car or one minute to get a car becomes diminishing returns.

2 min shoukd be our goal

Chris Nakutis Taylor, one of the early general managers at Uber, describes their importance: ETAs were always terrible to start. +15 min in some areas, especially in the suburbs. There was another key metric. Get ETAs down under 3 min on average as quickly as possible while covering the whole city. If you could get ETAs, unfulfilleds, and surge down quickly, you’ll have a healthy market.

The target ETA for uber to ensure network effect

For new products, it’s important to have a hypothesis for the size of your network even before you begin. Communication apps can be 1:1, so the network is small, and you can plan accordingly.

What is the minimal size for Bolt?

The solution to the Cold Start Problem starts by understanding how to add a small group of the right people, at the same time, using the product in the right way.

Find the right ppl, right density, etc

the key is the “atomic network”—the smallest, stable network from which all other networks can be built.

Key term: Atomic Network

The networked product should be launched in its simplest possible form—not fully featured—so that it has a dead simple value proposition. The target should be on building a tiny, atomic network—the smallest that could possibly make sense—and focus on building density, ignoring the objection of “market size.” And finally, the attitude in executing the launch should be “do whatever it takes”—even if it’s unscalable or unprofitable—to get momentum, without worrying about how to scale.

Atomic network value prop

The networked product should be launched in its simplest possible form—not fully featured—so that it has a dead simple value proposition. The target should be on building a tiny, atomic network—the smallest that could possibly make sense—and focus on building density, ignoring the objection of “market size.” And finally, the attitude in executing the launch should be “do whatever it takes”—even if it’s unscalable or unprofitable—to get momentum, without worrying about how to scale.

Atomic network value prop

The hard side of a network is important to understand, not just for Wikipedia, but for any new product seeking to launch its atomic network. Without this critical group, an atomic network will struggle to get off the ground.

Hard side kof the network

However, the trick is to look closer—to segment the hard side of the network and figure out who is being underserved.

In other words, the ideal product to drive network effects combines both factors: The product idea itself should be as simple as possible—easily understandable by anyone as soon as they encounter it. And at the same time, it should simultaneously bring together a rich, complex, infinite network of users that is impossible to copy by competitors.

Ideal product

At Uber, we called these moments Zeroes. A zero at Uber was the worst experience you could have, when a rider opens the Uber app with the intent to pick an address and pick up a ride—but there aren’t any drivers in the area! This is a zero.

Zero moment

I encourage product teams to develop their own form of this metric, laid out as a dashboard of networks—whether that’s divided by geography, product category, or whatever else makes sense. Within each, it can be useful to track the percentage of consumers that are seeing zeroes.

Shouldd we develop zero dshboards?

The early work on scurvy gave us the foundational tools. But instead of providing citrus and measuring for malnutrition, tech companies can reason by analogy: create user cohorts by levels of engagement, and analyze what differentiates high value users from lower value ones. These start out as correlations, so use A/B testing to prove causality—once the best levers are found, test many variations of these ideas. Rinse and repeat,

How to identify correlation and then causation in engagement

new users might start to seek another network where they can be successful. They want a more level playing field, and to get that, they will constantly try new products from competitors—not

Network density beats total size, a theme we’ve seen throughout the examples of this book.

By picking the right entry points, these new startups were spring-loaded to quickly reach an atomic network quickly, and then scale up with multiple network effects.

On cherrypicking

But counterintuitively, for networked products, this is often a trap. It’s exactly the wrong way to build a network, because a wide launch creates many, many weak networks that aren’t stable on their own.

Why big bang doesn't work in network products

You’d rather have a smaller set of atomic networks that are denser and more engaged than a large number of networks that aren’t there.

But here’s the paradox: To build a massive successful network effect, I argue that you must start with a smaller, atomic network.

However, the rideshare competition also shows the fallacy in believing in winner-take-all markets—instead, products compete as networks of networks, so that even when Uber’s network was bigger in aggregate, it was only 50/50 in cities like San Francisco and Los Angeles against Lyft.

Uber bundled Uber Eats across many touchpoints within its rideshare app, but still fell behind in food delivery versus DoorDash. Bundling hasn’t been a silver bullet, as much as the giants in the industry hope it is.

Cross pollination