Cold Start
This is a book about the common patterns and principles behind cold starts.
First, it argues that the way to solve the cold-start problem is through network effects.
What are network effects? It's the idea that people connecting with people increases product stickiness, attracting and retaining users.
The classic example is the telephone. The telephone also had a cold start. Early on, you could only call one fixed person—like a walkie-talkie—so few people wanted to use it.
It only became truly popular once it formed a phone network. The more people who joined the network, the more likely they were to stay.
In short: your product needs a feature that lets users connect with each other.
Second, when you launch a new product, it should come with an "atomic network" built in.
An atomic network is the smallest usable user network—the minimum number of users required to make the network function kick in.
Whether you recruit family and friends or pay people to join, you need to assemble an atomic network so that the moment a new user signs up, they immediately feel an active network around them.
Third, the size of the atomic network varies by product. The rule of thumb is that it should let a new user stay engaged for at least three minutes.
If it's shorter than three minutes, the network effect is probably too weak to retain users.
1. Uber: For ride-hailing, the atomic network should include 15–20 cars so that a new user can get a ride within three minutes.
2. Airbnb: For home-sharing, the atomic network should include 300 listings—that's 300 hosts—so users have enough options to choose from.
3. Reddit: For a community platform, the atomic network should have 1,000 subreddits so different users can all find communities they care about.
4. Slack: For a group-chat product, the atomic network should have three people and already contain at least 2,000 messages.
Fourth, what if your product is purely a tool and lacks any social-network element? Andrew Chen argues that if your product doesn't have network properties, you need to add them.
Users come for the tool and stay for the network.
Instagram originally launched as a photo-filter app, and it couldn't retain users—people used it a few times and left.
Then it pivoted into a photo-sharing network, adding a feed of photos from people you follow so you see friends' updates the moment you open the app.
That change is part of what led Facebook to acquire Instagram for $1 billion, just 18 months after launch.
Other examples: Yelp started as a local-business directory tool and became a business-review network. LinkedIn started as an online résumé tool and became a professional-networking platform.
The bottom line: if you want network effects, you have to become a network.
Fifth, once a user network in a niche reaches 20,000 people, it can grow on its own and eventually cover the entire market.
In other words, 20,000 users is the threshold for a single market. Once you break through that number, you've crossed the cold-start cliff.