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

by Andrew Chen

In my several years there, it was unusual to ever hear about an aggregate number—like total trips or total active riders—except as a big vanity milestone at a company all-hands.

Those aggregate metrics were regarded as mostly meaningless. Instead, the discussion was always centered on the dynamics of each individual network, which could be nudged up or down independently of each other, with increased marketing budget, incentive spend for either drivers or riders, product improvements, or on-the-ground operational efforts.

Although the networks don’t own their underlying resources, it’s the connection that matters. The entire ecosystem stays on because the value is in bringing everyone together. That’s the magic.

The “effect” part of the network effect describes how value increases as more people start using the product. Sometimes the increasing value manifests as higher engagement, or faster growth.

The questions to ask are simple: First, does the product have a network? Does it connect people with each other, whether for commerce, collaboration, communication, or something else at the core of the experience? And second, does the ability to attract new users, or to become stickier, or to monetize, become even stronger as its network grows larger?

The mechanics of network effects provide a path for new products to break through, as they are often able to attract new users by word of mouth and viral growth, as well as increase engagement and decrease churn as the breadth and density of the network grows.

Most products these days are low technical risk—meaning they won’t fail because the teams can’t execute on the engineering side to build the products—but they are generally also low defensibility. When something works, others can follow—and fast.

Networked products also have strong advantages in attracting new users, by leveraging their users to refer other users—this is critical as the channels to market to potential audiences have become highly competitive.

Network effects are one of the only protective barriers in an industry where competition is fierce, and defensive barriers are weak. While Instagram might be able to copy Snapchat’s features like Stories or ephemeral photo messages in a few months, it’s difficult to change the behavior of millions of consumers to switch over.

The network effect version of this in the technology industry happens when there is “overcrowding” from too many users. For communication apps, you might start to get too many messages. For social products, there might be too much content in feeds, or for marketplaces, too many listings so that finding the right thing becomes a chore. If you don’t apply spam detection, algorithmic feeds, and other ideas, quickly the network becomes unusable. But add the right features to aid discovery, combat spam, and increase relevance within the UI, and you can increase the carrying capacity for users.

Just as a messaging app will grow and grow but eventually saturate its market, so too does the growth of animals slow down as they begin to overpopulate their environment. While the terms are different, the core concepts and the math are the same: The Allee effect → The Network Effect Allee Threshold → Tipping Point Carrying capacity → Saturation

The Escape Velocity stage is all about working furiously to strengthen network effects and to sustain growth. This is where the classical definition of a “network effect” is wrong. I redefine it so that it’s not one singular effect, but rather, three distinct, underlying forces: the Acquisition Effect, which lets products tap into the network to drive low-cost, highly efficient user acquisition via viral growth; the Engagement Effect, which increases interaction between users as networks fill in; and finally, the Economic Effect, which improves monetization levels and conversion rates as the network grows.

Airbnb had to fight off its European competitor by competing on the quality of the network, and scaling its network effects—not via traditional competitive vectors like pricing or features.

A larger network and a smaller network in any given market have distinctly different strategies—think of it as a David strategy versus a Goliath strategy. The upstart has to pick off niche segments within a larger network, and build atomic networks that are highly defensible with key product features, and, when applicable, better economics and engagement. The incumbent, on the other hand, uses its larger size to drive higher monetization and value for its top users, and fast-following any niches that seem to be growing quickly.

The teams that adopted tended to be other startups, for a very good reason: Technology startups adopted us early because they have the belief, whether naive or not, that software can better their lives. These startups were just like us—many of the early teams had fewer than 10 people, just like Slack.

The minimum number of people to be defined as a team, even today at Slack, is three. As long as you have three people, it can be stable.

It’s a myth that network effects are all powerful and positive forces—quite the opposite. Small, sub-scale networks naturally want to self-destruct, because when people show up to a product and none of their friends or coworkers are using it, they will naturally leave.

How many users does your network need before the product experience becomes good? The way to answer this is for companies to do an analysis on the size of their networks (on the X-axis) plotted against a set of important engagement metrics (on the Y-axis).

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. Contrast that to products that are highly asymmetrical, with content creators and viewers, or marketplaces with buyers and sellers—these are likely to require a much bigger number to hit the threshold, and require a much bigger effort to get started. The size of an initial network helps determine a launch strategy.

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. Getting this initial network off the ground is the key, and the key is the “atomic network”—the smallest, stable network from which all other networks can be built.

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.

Embedded within Slack’s strategy, and the strategy of many early-stage networked products, is a series of short-term boosts—often called “growth hacks”—which are important in forming the initial atomic networks. In Slack’s case, it was their incredible buzz within the early-adopter startup community, and their invite-only launch.

The concept of atomic networks is powerful because if you can build one, you can probably build two. Each one often becomes easier, because each network can be intertwined with the next—Slack’s success within one company can help it become successful in another, as employees move about and introduce the product to new workplaces.

A successful new product should be able to answer detailed questions: Who is the hard side of your network, and how will they use the product? What is the unique value proposition to the hard side? (And in turn, the easy side of the network.) How do they first hear about the app, and in what context? For users on the hard side, as the network grows, why will they come back more frequently and become more engaged? What makes them sticky to your network such that when a new network emerges, they will retain on your product?

1% of the user population might start a group (or a thread within a group) 10% of the user population might participate actively, and actually author content whether starting a thread or responding to a thread-in-progress 100% of the user population benefits from the activities of the above groups (lurkers)19

Within these platforms, just ask yourself, “If a piece of content was created, and no one saw it, would the creator be disappointed?” If the answer is yes, then social feedback is a key value. The combination of tools, aggregation of audience, and a networked product is what is needed to unlock the hard side of these networks—it’s all about the content creators.

It’s important to focus on this tiny slice of users so that messaging, product functionality, and business model are all aligned to serve them. Without this group, the atomic network will collapse—a social network can’t exist without its content creators, and a marketplace can’t exist without its sellers.

similar to Instagram and YouTube’s content creators, Wikipedia’s content creators are likely motivated by the community itself. Social feedback, status, and other community dynamics encourage editors to keep creating content. Wikipedians, as they call themselves, can show their expertise in a topic by maintaining comprehensively written pages, and people within the community will thank and appreciate them. This provides status. They can make edits to correct others, which offers another form of status and satisfaction. There’s teamwork and a feeling of camaraderie, which create bonds that retain users over months and years.

Thus the order of operations, at least for most consumer-facing marketplaces, is “supply, demand, supply, supply, supply.” While supply might be easy to get onto the network early on through subsidies, eventually it will become the bottleneck. The hard side of a network is, by definition, hard to scale.

What people are doing on their nights and weekends represents all the underutilized time and energy in the world. If put to good use, this can become the basis of the hard side of an atomic network. Sometimes the army is built on people with excess time, but sometimes it is built on people with underutilized assets as well.

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

Networked products are fundamentally different from the typical product experience—they facilitate experiences that users have with each other, whereas traditional products emphasize how users interact with the software itself. They grow and succeed by adding more users, which create network effects, whereas traditional products grow by building better features and supporting more use cases.

And the most important features in networked products often revolve around how users find and connect with each other, whether that’s photo tagging, sharing permissions, or “People You May Know.”

Uber’s app was initially outsourced to Mexico, so that when later engineers joined the company, they needed to be issued Spanish-to-English dictionaries to understand the comments and source code. In these cases it isn’t until later, as the product hits scale, that the engineering teams are upgraded.

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.

The Magic Moment is a nice concept, but it would be even more useful if you could measure it. The way to best do this might be surprising—you start with the opposite of magic, the moments where the network has broken down, and you start solving the problem from there.

For a workplace collaboration tool, like a wiki, it might be stale or missing documentation needed for work that no one’s filled out.

To begin the discussion of the Tipping Point, I’ll start with a prominent strategy, “Invite-Only,” that is often used to suck in a large network through viral growth. Another method to tip over a market is with a “Come for the Tool, Stay for the Network” strategy.

invite-only launches have been a key feature of many products precisely because for networked products, there are huge advantages. It allows the early network to gel as a community, develop a high density of connections, and grow organically via virality.

In the early days, it was our job each day to make sure there was good content on the front page. We’d post it ourselves, using dozens of dummy accounts. Otherwise the community might dry up.42

Flintstoning can be phased out over time. In that way, it can be thought of as a close cousin to the “come for the tool, stay for the network” approach. The Flintstoning approach focuses on artificially propping up the hard side of the network with highly manual efforts, whereas the “come for the tool” props it up with software.

Travis would regularly say to the product teams, “Product can solve problems, but it’s slow. Ops can do it fast.” As a result, Uber saw itself as an “ops-led” company, and it was this team that best embodied the startup’s entrepreneurial and creative culture.

Only three sourcing strategies account for every B2B company’s very early growth. [These are: Personal network, Seek out customers where they are, Get press.] Thus, your choices are easy, yet limited. Almost every B2B business both hits up their personal network and heads to the places their potential customers were spending time. The question isn’t which of these two routes to pursue, but instead how far your own network will take you before you move on.

To kick off an effort to generate more revenue, a cross-functional Growth and Monetization team was convened. It had a core of quantitative, business-minded product leaders who also teamed up with engineers and designers, led by ChenLi Wang and Jean-Denis Greze. This team would be empowered and given the resources to directly drive growth and monetization opportunities. They would create new insights about Dropbox’s business, identify and prioritize opportunities, and execute against them by shipping new features and updates to the product.

Strong product-driven cultures like Dropbox tended to believe that the only thing that mattered in attracting users was a great product—why waste the talent of skilled engineers working on landing pages or optimizing email notifications when they could be building the next generation of great features? A similar objection often comes from the marketing teams at technology companies who historically owned customer acquisition—why create a duplicate team that overlapped with their work? Yet the track record speaks for itself—growth teams have emerged across the industry as a focused way to scale products toward Escape Velocity.

Not every user is the same, nor is every network the same. Dropbox’s insights on this were profound: some users joined Dropbox as part of the “come for the tool” strategy—but stayed with the tool without increasing their engagement by sharing folders and documents with others. In contrast, the ones who used Dropbox for collaboration and sharing—the network features—became significantly more valuable over time. Dropbox’s users could be divided into High-Value Actives (HVAs) and Low-Value Actives (LVAs), which was useful as a quality indicator. It could be overlaid into the strategies for marketing channels and partnerships to make sure HVAs were being acquired, not LVAs.

Understanding the value of users impacted Dropbox’s growth strategy. When they partnered with one of the largest mobile companies to provide photo backup services, the Dropbox team realized that the partnership generated a lot of new users but they were all LVAs. This created substantial costs in supporting the users but not necessarily future revenue as they were unlikely to upgrade. The HVA versus LVA concept helped the Dropbox team understand and prioritize its various efforts.

The “Acquisition Effect” is the ability for a product to tap into its network to acquire new customers. Any product can buy Facebook or Google advertising, for instance, to attract new users, but only networked products can tap into viral growth—the ability for users in its network to tell others in their own personal networks. This keeps customer acquisition costs low over time, fighting against the natural rise that comes with market saturation and competition.

The types of projects that amplify the Acquisition Effect are oriented around viral growth: referral features that reward users when they invite others, tapping into contacts to create suggestions for who to add to an app, and improving conversion along the key moments in the invitation experience.

The “Engagement Effect” describes how a denser network creates higher stickiness and usage from its users—it is a more specific form of the classic description of network effects that I covered at the beginning of the book, “the more users that join the network, the more useful it gets.”