Hi, and welcome to the a16z podcast. I'm Lauren Murrow. In normal times, every company operates against some hypothetical growth model — a data-driven framework that describes how your product grows and how you acquire new users.In the fallout from the pandemic, most founders and CEOs are in the process of completely revamping their growth models from the bottom up amid new and unpredictable consumer behavior.
It turns out that there are many, many, many flavors of this. This is the verbal version of it. When you go deeper, you're able to translate this set of hypotheses and ideas into spreadsheets and numerical models for what's actually happening in the business and understand the flows.: Right.
And so, if your whole thing is about: Okay, we need 5X growth and that means that people need to invite each other at a certain rate — and if they're not, then maybe you need to make that up with paid marketing spend; with financial incentives for your users to use the product, whether that's in the form of free subscriptions or in the form of a lower price point; or, if you're a marketplace company, you might give people discounts that are dropped into all the consumers'...
If people just aren't in the mindset to convert and you're seeing headwinds, then the best thing you can do is build up this stored potential that you might be able to convert later, and that's really where to start to invest. And that easy evolution is taking some of the channels that potentially no longer work or some components of the growth model that no longer work, and cutting them and saying, "Look, if it turns out that this period is longer than six months, it's better to get rid of this, and then we can rebuild the function later," whether that's from an expertise or infrastructure perspective. And that's a little bit more evolutionary.
But if we're talking about experiments in the sense of: "I'm trying things and I'm trying to gauge the reaction, either qualitatively or anecdotally," those are a different set of experiments. What you're really looking for is an obvious reaction that something is working or not working, versus statistically significant data.
At least what we've done at Reforge is we've looked at our scenarios on a wider range of outcomes than we would typically consider. Everything from a 20% cut in demand down to an 80% cut in demand over a wider range of time periods. And it's less likely that those extremes happen, but they're more likely than in typical situations. So that's why you have to account for those.
Pretty much any product that's subscription and has really high retention on the subscription side is going to be an interesting state, because you can maybe acquire a lot of these customers for cheaper through paid marketing, through more engagement. And if you're able to then retain a really high degree of renewal after this, then those users will probably keep forever.