"We have network effects" is one of the most overused phrases in startup pitches. Most products that claim network effects have something weaker: a positive feedback loop, or a product that gets slightly better with more users but isn't defensible in the way real network effects imply.
Understanding which type of network effect you have — or could have — is the difference between building a defensible product and building one that's always one funded competitor away from losing its position.
Type 1: Direct Network Effects
The value of the product increases directly for each user when more users join. This is the original network effect, described by Metcalfe's Law: the value of a network grows proportionally to the square of its connected users.
Classic examples: Telephone networks, WhatsApp, Slack.
A telephone is worthless if you're the only one with one. WhatsApp's value to each user is a direct function of how many of their contacts are on WhatsApp. When all your contacts move to WhatsApp, the switching cost becomes enormous — you'd have to convince your entire network to follow you to a competitor.
Building direct network effects: Your product needs to require communication or interaction between users. Single-player tools don't have direct network effects; multiplayer tools do.
Type 2: Indirect Network Effects
The product gets more valuable as more users join not because users interact directly, but because more users attract more complements (tools, integrations, content, services).
Classic examples: Windows, iOS, Shopify.
Each new app developer who builds for iOS makes iOS more valuable to consumers, which attracts more consumers, which attracts more developers. The users (consumers and developers) don't interact — but each group makes the other's experience better.
Building indirect network effects: Create a platform that others can build on. An API, a plugin system, a marketplace for third-party services. As the platform grows, the ecosystem grows, which makes the platform more valuable.
Type 3: Data Network Effects
The product gets smarter (better recommendations, better predictions, better automation) as more users generate more data.
Classic examples: Waze, Spotify, Google Maps.
Waze's routing is better in cities with millions of users sending real-time traffic data than in cities with 1,000 users. Spotify's recommendations improve as more users train the recommendation algorithm with their listening behavior.
The nuance: Data network effects are often overstated. The data advantage saturates — at some point, more data doesn't meaningfully improve model performance. And data advantages are increasingly hard to defend as model architectures improve and foundation models reduce the cost of building AI on top of commoditized models.
True data network effects require not just more data but more unique, proprietary data that competitors can't replicate.
Type 4: Social Network Effects
Users recruit other users because the product delivers more value within a defined social group.
Classic examples: LinkedIn, Facebook, Notion (within organizations).
LinkedIn's value is tied to your professional network existing on LinkedIn. When a company's HR team, managers, and employees all use Notion, the value of the shared workspace creates switching costs at the organizational level.
Building social network effects: Design for groups, not individuals. Features that create shared state — shared documents, team views, shared histories — make the product more valuable when a user's group is using it.
Type 5: Two-Sided Network Effects (Marketplace)
Two distinct user groups — buyers and sellers, drivers and riders, hosts and guests — each attract the other. More buyers attract more sellers; more sellers attract more buyers.
Classic examples: Airbnb, Uber, Stripe.
Marketplace network effects are powerful but require solving the cold start problem on both sides simultaneously. A marketplace with buyers but no sellers (or sellers but no buyers) has zero value. This is why marketplace cold starts are harder than single-sided products.
Building two-sided network effects: You need a minimum viable supply and demand density before the network effects kick in. Most successful marketplaces started hyper-local (one city, one niche) to achieve density before expanding.
The Hierarchy
Not all network effects are equally defensible:
Most defensible: Direct network effects (Metcalfe's Law compounds; switching costs are enormous when your whole network is on the platform)
Highly defensible: Two-sided marketplace effects (liquidity moats are hard to build and hard to attack)
Moderately defensible: Indirect (ecosystem) network effects (can be disrupted if a new platform offers better economics to developers)
Least defensible: Data network effects (increasingly commoditized as AI infrastructure matures)
What Most Products Actually Have
Most SaaS products don't have strong network effects. They have:
- —Switching costs: it's painful to leave because of data accumulation or workflow integration
- —Habit: users have built behavior around the product
- —Product quality: the product is genuinely better than alternatives
These create retention and competitive advantage, but they don't compound the way network effects do. Being honest about this is important for understanding how your product will perform against a well-funded competitor.
FAQ
Can you add network effects to a product that doesn't have them?
Sometimes. Adding collaboration features to a single-player tool can create direct network effects. Adding a marketplace or partner program can create indirect effects. But network effects that are bolted on rather than core to the product design are usually weak.
How do you measure network effect strength?
Look at retention rates by cohort size. If users in larger networks (more teammates, more connections) retain significantly better than users in smaller networks, you have evidence of network effects. The strength is proportional to the retention differential.
What if I'm building in a market where the network effect winner already exists?
Compete on a different network boundary. If LinkedIn owns the global professional network, build a deeper network for a specific vertical (legal professionals, engineers in the UK, fintech founders). Vertical networks can out-compete horizontal ones on depth even if they lose on breadth.
Written by
Ross
Founder & Strategy Lead, Greta Agency
Ross has spent 10+ years building growth engines for companies from seed to Series C. He founded Greta Agency to prove that great software can ship in days, not months.