Why premature scaling fails : By Andrew Chen
Sometimes you build a product that’s just doing okay. Not great, but not bad either. Users show up, and some of them stick around – but not many. You can quantify this with cohort retention curves. For a social app, this might be when your D30 is 10%, not 20%+. For a subscription product, this might be when your annual retention is <20%. It sort of works, but not really.
So what do you do if it’s going OK, but not quite working?
The blessing of a small user base
In the early days this works. Let’s use a real example — let’s say you start with 1,000 active users, and then you want to 3x. The good news is, 3x with 1,000 users just means you need to buy 2,000 users. To grow it further, you just repeat again, right?
Well, not quite. Even if you’ve bought 2,000 users, you might only retain 500 of them longer term. (So that you have 1000 actives + 500 new actives). The problem is that a large % of these users burn off. That’s fine, maybe just replace them, by buying more users? To 3x again, you have to replace:
The % you lost
Plus, buy another whopping amount just to grow.
The treadmill arrives
The traction treadmill eventually arrives once the numbers get big. This is when you lose a % of users fast, but then just have the budget and funding to replace them- but then can’t keep grow on top. Eventually on a base of millions of users, you might churn a million users, and need to replace them — and buy more, to grow. So then you’re starting to talk about very large outlays, just to stay afloat, much less to consistently grow.
You start running hard and just staying in one place.
You might increase budgets. Pay up on CAC/LTV. Optimize and run experiments. And that works a little.. until the treadmill reappears.
Conclusion
In the end, this story isn’t about polishing your product forever. I’m certainly not encouraging that. But I do think you need to understand where your product stands relative to other successes (and failures in the market). The benchmarking is important. And if you try to scale and fix at the same time, be prepared for the treadmill to show up. By then, your nimble speedboat of a product will have grown into an aircraft carrier, and it’ll be hard to turn things around. This is when scale becomes the enemy to iteration.
