A useful lens for products: primary vs secondary use cases.

I wrote this up in an old public-but-not-publicized essay.

A primary use case is one whose expected value for a user exceeds their expected cost.

It's "expected" because it's based on users priors for how the feature will work, based on:

How similar tools have worked

What their friends have told them

Their own prior uses of this tool

"Cost" here means both literal money but also friction, uncertainty, etc.

If for a user the expected value is greater than the expected cost, then there is some activation gradient; the tool is viable and over time will diffuse along that gradient and become adopted.

The steeper the gradient, the faster the diffusion.

The precise expected value and expected cost differs for different users, and can change over time.

For example as there's more word of mouth diffusing information, or the product quality changes.

If you don't have a primary use case to be your wedge, the product is not viable.

However it's not all primary use cases. There are also secondary use cases.

A secondary use case is one that is not strong enough to be a primary use case, but is a nice bonus.

If a user is already using the tool for a primary use case, and they see out of the corner of their eye a secondary use case that is relevant, they might use it.

Secondary use cases can be activated in people's peripheral vision.

Typically the secondary use cases are minor and hard to activate.

But sometimes the secondary use cases have a network effect.

The more that people use them, the more useful they get.

In these cases, even if only a very small number of people use the secondary use case, that trickle makes it higher quality, which pulls in more people: a compounding loop.

Some secondary use cases have such a strong network effect that over time they grow to eclipse the primary use case and become their own primary.

The original primary use case is like the shade that protects the secondary use case until it grows strong enough on its own.

Imagine if you had a tool that had a few good primary use cases, but secondary use cases with hyper-viral dynamics (e.g. multiple interlocking network effects), where it gets better and better with usage at a huge rate.

As long as there's a viable primary use case to get the whole thing going, the overall system could become massive faster than you might think.