Network Effects, explained.
Network effects occur when a product or service's usefulness depends on how many other people participate—and, often, which people they are.
Why it happens
A communication network has more possible connections as membership grows. But possible pairs are not actual conversations. Relevance, compatibility, and participation determine whether those potential connections create value.
A communication tool can become more useful when more of the people you want to reach use it. Each additional member introduces new possible pairs. With n members there are n(n − 1)/2 distinct undirected pairs.
The experiment separates that possibility from actual connections. Each pair connects independently with the probability you choose. Hollow circles have no direct connections in the sampled network.
Read the result
Change membership and connection probability separately. Adding people increases possible pairs; low connection probability can still leave many isolated members. Read potential connectivity and realized connectivity together.
A worked example
A class chooses a messaging app
One student installs an excellent app that no classmates use.
A less elaborate app becomes more useful to that student if their study group already participates there.
The relevant network is the people the student wants to reach, not simply the platform's worldwide user count.
OPTIONAL DEEPER DETAILGo deeper: inside the model
Inside this model
This is a small undirected random graph with no self-links or duplicate edges. At 0% connection chance there are no edges. At 100%, every possible pair is connected.
Connections are only an illustrative proxy for opportunity. The simulation does not assign economic value, and it does not assume that indirect connections are as useful as direct ones.
Where this idea is useful
Adoption and usefulness can reinforce one another when people gain from sharing a network. But a member count alone leaves out who can connect and why they would want to.
A common misconception
“Every extra user makes a network better.”
Congestion, spam, incompatible needs, and irrelevant participants can reduce value. Network effects need a mechanism, not just a growing number.
What this explanation leaves out
- Real relationships are not independent or equally likely. Clusters, geography, compatibility, and personal preferences matter.
- Congestion, spam, moderation costs, and poor-quality interactions can offset benefits. More possible pairs do not guarantee more value.
How are network effects different from economies of scale?
Network effects change user value through participation. Economies of scale change production costs as output grows. A business can have either, both, or neither.
Whose participation would make a tool more useful to you, and whose would make little difference?
Further reading
Easley and Kleinberg distinguish the direct benefits of network participation from copying others for information.