Network effects create one of the most seductive growth stories in product. More users create more value. More value attracts more users. Repeat. Sometimes that is exactly what happens. Sometimes the next thousand users make the product worse. Search becomes noisy. The marketplace fills with low quality supply. The community gets harder to trust. The best creators get buried. Matching gets slower. Moderation cost grows. The network is bigger. The interaction is worse.
That is why I think network size is an input. Interaction quality is the product.
Membership is not the value
A professional network does not become useful because millions of profiles exist. It becomes useful when the right people can find, evaluate, and interact with one another. A marketplace is not valuable because it has a lot of buyers and sellers. It is valuable when buyers can reliably find acceptable supply and sellers can reliably find demand. A developer ecosystem is not useful because thousands of integrations exist. It is useful when the integrations are trustworthy, discoverable, maintained, and relevant.
The network effect lives in the improvement of the core interaction. That is the thing growth should strengthen in practice, not just in aggregate.
More can create congestion
There are several ways network growth can destroy value. The obvious one is congestion. Too many participants compete for limited attention, inventory, space, or opportunity. The second is pollution. More content, listings, messages, or participants increase the cost of finding what is good. Then there is composition. A two sided marketplace can add thousands of suppliers and still get worse if buyer demand does not grow with them. A community can grow overall while becoming less useful to the niche that made it valuable in the first place.
A collaboration product can add more people while increasing interruption and coordination cost faster than useful connection. A network can even show more aggregate activity while each useful match takes more searching, filtering, or attention to produce. Growth is not one variable. Who joins matters. Where they join matters. What they do matters. Who else is already there matters.
Network health is often local
This is another reason global user counts can mislead. A marketplace may have millions of users and still be broken in Atlanta at 2 p.m. on a Tuesday. A social product may be vibrant overall and dead inside a specific interest graph. A B2B network may be healthy for large enterprises and nearly useless for small companies. A collaboration tool may be valuable in one team because participation is dense and irrelevant in another because only two people use it.
Andrew Chen’s work on the cold start problem makes this local nature of network effects especially clear. The relevant question is not always, “How large is the network?” It is, “Is there enough of the right activity in the local network where this interaction needs to happen?” That changes how Product measures growth.
At some point the job moves from filling to shaping
Early in a network product, quantity can be the constraint. There are not enough drivers. Not enough buyers. Not enough creators. Not enough developers. Not enough colleagues using the tool. The product job is density. But once density improves, a different problem can become limiting. Now Product has to shape the network. Ranking. Matching. Curation. Reputation. Moderation. Segmentation. Local balancing. Quality thresholds. Sometimes even limits on participation, messaging, supply, or choice.
That can feel strange because the growth team spent years removing barriers to entry. Now the product may need selective friction to protect the interaction that made growth valuable in the first place.
MAU can go up while the network gets worse
This is why aggregate growth metrics are not enough. If monthly active users rise 20 percent, what should improve if the network is actually healthier? Time to successful match? Fill rate? Response rate? Repeat interaction? Quality adjusted supply? Trust signals? Core action retention? The right metric depends on the product. The principle does not. Measure whether more participation improves the useful interaction, not only whether more people arrived.
Be careful with “quality”
There is a real danger here too. Quality can become a polite word for incumbency. Platforms can over curate and make it impossible for new creators, sellers, developers, or communities to break through. A ranking system can preserve the people who already have attention. A reputation system can lock newcomers out before they have a chance to earn trust. A marketplace can optimize short term conversion so aggressively that it kills variety and future supply.
Shaping the network is not permission to freeze it. The job is to protect useful interaction while preserving enough openness for the network to renew itself.
Ask what the next participant improves
Before celebrating another growth milestone, define the core interaction your network exists to create. Then ask:
If 10,000 new participants arrive next month, what should become better for somebody else?
If the answer is only “the network will be larger,” keep going. Maybe you still need density. Maybe the next user really does make the product better. But eventually the constraint changes. At that point, adding users is no longer the whole growth strategy. The product has to make sure growth still creates value inside the network. Bigger is not the same as better. For a network product, the difference lives in the interaction.