Startups: doing things that don’t scale vs those that do
One of the oft-cited learnings for founders in their early days is to do things that don’t scale. The Airbnb founders famously in their early days rented out their own apartment for early inventory, and sent professional photographers to hosts’ homes to make houses look great. There’s a whole, crowd-sourced collection of early hacks that don’t scale, but help in getting from 0 to 1.
If we were to model this out graphically, doing things that don’t scale would look like this orange dotted line (original is green):
Things that don’t scale are useful in the early days, going from 0 to 1. They shift the starting point of the graph, helping disproportionately in the early days. Doing things that don’t scale multiple times actually has huge impact in the early days, and can be the difference to survival when going from 0 to 1:
Things that Scale
As you zoom out however, these two lines converge in the long run. We see that things that don’t scale don’t matter in the long run.
Let’s model things that scale (often 1 to 100, not 0 to 1):
- Growing and managing a team successfully
- Improving retention & engagement
- Improving conversion rates
Changing things that scale affects the unit-economics of the business overall, and hence impacts each point on the curve. It is a change in gradient (or rate of change), rather than a lateral shift:
Couple of things are obvious here — these scaleable (“things that scale”) changes are infinitesimal in the early days, won’t move the needle. But in the long run, they affect the trajectory of the startup orders-of-magnitude more than things that don’t scale.
As I work on new projects, I often ask myself — is this a lateral shift, or a gradient change in the overall product? In the early days (0–12 mos), its better for founders and early employees to spend more time on lateral changes going from 0 to 1. Beyond that (2yrs+), put the blinders on and focus focus focus only on the things that scale and affect the core fundamentals of your product.