The reason I don't use the world exponential, is that superlinear growth in a product is generally not exponential.
It is usually at best quadratic because there is some kind of network effect at play - the number of edges in a complete graph with `n` vertices is `n(n-1)/2`.
The point is that the value of your product to your users should increase with the number of users OR the marginal cost to you of maintaining the product should decrease with the number of users.
"Superlinear" captures that more effectively than "exponential".
True viral growth is exponential: each of your users gets you N>1 additional users, and so the growth of your userbase is proportional to your current scale.
Superlinear but subexponential growth can still be pretty impressive, though, especially if it's fast (who cares about 0.1% weekly growth?)
One thing to add - in my experience, you need to spend money to go from merely superlinear to (locally) exponential.
This is the original reason that VCs exist.
If you go to an intelligent decisionmaker at a VC firm with proof of sustained superlinear growth and plan for how their capital will allow you to increase the rate (and higher derivatives) of your growth, they will almost certainly invest.
The reason that VCs have to specialize by industry or look for signals like Stanford/Harvard/whatever is that it's very difficult to validate or invalidate proof of superlinear growth. This requires specialization or, if lazy, social signals like the university you graduated from.
The reason I don't use the world exponential, is that superlinear growth in a product is generally not exponential.
It is usually at best quadratic because there is some kind of network effect at play - the number of edges in a complete graph with `n` vertices is `n(n-1)/2`.
The point is that the value of your product to your users should increase with the number of users OR the marginal cost to you of maintaining the product should decrease with the number of users.
"Superlinear" captures that more effectively than "exponential".
True viral growth is exponential: each of your users gets you N>1 additional users, and so the growth of your userbase is proportional to your current scale.
Superlinear but subexponential growth can still be pretty impressive, though, especially if it's fast (who cares about 0.1% weekly growth?)
Very good point.
One thing to add - in my experience, you need to spend money to go from merely superlinear to (locally) exponential.
This is the original reason that VCs exist.
If you go to an intelligent decisionmaker at a VC firm with proof of sustained superlinear growth and plan for how their capital will allow you to increase the rate (and higher derivatives) of your growth, they will almost certainly invest.
The reason that VCs have to specialize by industry or look for signals like Stanford/Harvard/whatever is that it's very difficult to validate or invalidate proof of superlinear growth. This requires specialization or, if lazy, social signals like the university you graduated from.
Less specific than exponential which also makes it more achievable. More positive than nonlinear.