Network Effect

Back when products were explained purely by their features, some situations had no answer. A team would ship something faster, cheaper and cleaner than the incumbent, and nobody moved. No comparison table explained why.

Messaging apps are the clearest case. A messenger nobody else has installed is useless, and a messenger where ten people you know are already sitting has an obvious reason to exist. A new one that is half again as good still requires bringing those ten people along. The reason people stayed put had nothing to do with the product itself.

In meetings this always showed up as the same sentence. Somebody asks why people aren't switching when ours is better, nobody has an answer, and the meeting ends with a decision to spend more on ads. The users that money brings in arrive alone, look around alone, and leave. The money is gone, nobody stayed, and the same question comes back next month.

Network effect is the name for the part those explanations kept missing. Once you treat value as depending not only on features but on who else is already inside, the reluctance to switch and the difficulty of catching a leader fall out of the same picture.

Add one user and you add the connections that user can make, which raises the value for everyone who was already there. The whole property sits in that second half: the newcomer isn't the only one who gains.

Counting connections makes the shape obvious. Two users have exactly one possible pair between them, five users have ten, ten users have forty-five. Five times the people, forty-five times the pairs. Each user added raises the value the product delivers by more than the one before did, not by a fixed step.

Connections form in two ways.

1. Users on the same side connect to each other. Messengers and collaborative document tools work this way. A colleague joining raises the value you get immediately.
2. Users on opposite sides connect. This is a two-sided market, a market that only functions when suppliers and buyers are gathered together. More restaurants on a delivery app is better for the people ordering, and more orders pull in more restaurants.

The same property runs in reverse, because it also means there is almost no value when few people are around. That is the cold start problem. Open a secondhand marketplace and the first week has twenty items listed, so buyers leave empty-handed, and with no buyers nobody bothers listing anything.

Which is why products built on this property start small on purpose. Not the whole country, but one neighborhood, one campus, one company first. Ten thousand users scattered nationwide are worth less than three thousand inside a single university, because the scattered ten thousand almost never run into each other. The unit that matters here is density within one group, not the total user count, and that is why early numbers for this kind of product get reported per campus or per city rather than in one national figure.

The most frequent use is deciding whether a product can grow this way at all. When users connect to each other, they don't need a reward to bring others in — inviting people is already in their own interest. A product whose value is unchanged whether you use it alone or not, a budgeting app or a journaling app, has almost none of this property and has to rely on word of mouth or paid acquisition instead.

The second is choosing where to open a market. Density creates value, so the budget goes into one place instead of spreading evenly. The argument for why a company operates in only three districts of one city, when an investor asks, comes from here too.

The third is explaining competitive advantage. Features get copied in a quarter; a group of users already inside is much harder to reproduce. Using that argument requires real connections between those users, though. A competitor can hire away the design and ship the same screens, and still find that the customers it wants are each waiting on twenty colleagues who would have to move at the same time.

"This isn't a product with a network effect. More users doesn't make anything better for the users we already have."

The growth plan has to run on ads or sales rather than leaning on the product itself.

"Let's not open nationwide. Build density inside one campus first."

Spreading thin produces no value anywhere, so narrow the scope.

"Right now the cold start problem comes before the network effect."

Filling the space up to the point where the property starts working is the job first.

At Pagedeck this comes up one team at a time. When three people at a company use it, sharing a document means nobody reads it and everyone drifts back to private notes. When fifteen people on the same team are inside, meeting notes and comments start moving and cancellations nearly stop. So Mia watches the share of people using it within an account before she watches the number of accounts.

Ravi gives the same reason when he works on the invite flow. Spending on ads for new companies returns fewer people who stay than getting colleagues invited inside companies that already signed up.

The most common error is treating a large user base as proof of a network effect. Big numbers feel solid, so the assumption is natural. But the property comes from connections between users, not from how many there are. In a weather app used by ten million people a month, no user gets better weather because of another user. A company that claims the property on user count alone watches those users walk to a competitor the moment one appears, because there were never any connections holding them.

Reading it as permanent safety is the other one. After hearing how hard a leader is to catch, that conclusion is easy to reach. What creates value, though, is density inside one group rather than the total, so a product can be dismantled one group at a time. If the students at one school move to another service together, the value inside that school is gone while the overall number still looks large, and a few quarters of that repeating school by school brings the whole thing down.