- Consume Our Internet
- Posts
- Filling a leaky bucket: what determines the importance of churn
Filling a leaky bucket: what determines the importance of churn
On unit economics streams and TAM reservoirs
Thanks for reading! If you enjoyed this, share it with a friend or colleague, or you can follow me on Twitter @SashaKaletsky.
Final note before we begin: Caspar wrote a great post about the regulation of influencer marketing a couple of weeks ago, so if you haven’t checked it out yet, it’s here.
If you spend a lot of time with venture capitalists, there’s a good chance you have heard the phrase “leaky bucket”. In case you are fortunate enough not to, I can explain it now: a leaky bucket is a business with high churn, which since it cannot retain its customers, will always struggle to grow. Even as you fill up the bucket, the water (revenue) just keeps leaking out.
The uninvestability of “leaky buckets” is received wisdom in venture capital. Many VCs say they will simply never invest in a business with high churn (let’s say, for example, <80% net revenue retention). No ifs, no buts.
The point of this article is to challenge this orthodoxy. Obviously low churn is always better than high churn, but there are absolutely circumstances when high churn is more OK than others. Or looking at it the other way, there are some circumstances where churn can be more disastrous. But nobody seems to talk about this.
So let’s dive straight into it. When is high churn more acceptable?
The two circumstances (and you need both), when churn is decidedly less likely to kill your business, are as follows:
Fast payback on marketing costs. We can call this the “powerful stream”
Large market size. We can call this the “infinite reservoir”
So there it is. That’s the message of this article. At some times, churn is less of an issue than others. And we’re going to talk about those times.
Hopefully this article will be helpful for founders, investors and startup employees, and not misinterpreted by all of the above.
The evidence
So first of all, is it true that high churn businesses can’t get big? Not always.
In consumer, many of the largest businesses of the world have very low retention, from Duolingo, Airbnb and Candy Crush Saga to products that are quite literally one-time purchases like Louis Vuitton handbags and Tesla cars.
It is not difficult to demonstrate empirically that you can build a huge consumer company without enterprise-grade >100% NRR.
But what people talk about less is why. When exactly does churn matter more? And least discussed of all, when does it matter less?
Two factors that determine how much churn matters
In enterprise software, churn matters a lot. When sales cycles take months or years, contract values are in the millions, and there are only a small number of potential blue-chip customers in the world, when you lose one it is a big problem.
This is obvious, but anecdotal. What can we take away from this which is actually useful? What framework can we build around it?
To answer this, we can return to the leaky bucket analogy. So, you’ve got a leaky bucket. You’ve tried everything to block up the holes but you can’t. Are you doomed? What are the only ways you can ensure to still fill it up to the brim?
First, make sure there’s a powerful stream above it. Second, make sure you’ve got enough water supply to service the stream. If you don’t have both of those, you’re in trouble.
In business language, we can translate that to:
Speed to pay back marketing costs (the powerful stream): Churn matters less when CAC payback is instant or acquisition is consistently organic.
Size of market (the infinite reservoir): Churn matters less when the TAM is near-infinite.
And you need both.
Let’s dig into each, and how they are related.
Variable 1: Speed to pay back marketing costs (the powerful stream)
Marketing can sometimes be helpfully seen as analogous to a capital expenditure. A business puts its money to work in marketing, and customer revenue comes out the other side. But there’s a delay. The money the company makes from the customer will generally be after the money it spends on acquiring them.
The difference in time between this cost and revenue (or if we’re being accurate here, incremental gross profit) is called the customer acquisition cost (CAC) payback period.
The easiest way to explain this is with an example.
Consider two different companies:
Company A: A customer costs $100 to acquire and you make $8 from them per month.
Implied CAC payback: 12 months.
Company B: A customer costs $10 to acquire, and you make $10 from them per month.
Implied CAC payback: almost immediate.
(Word of warning: CAC payback often degrades over time)
For Company A, each churn event is a catastrophe, because it means if they churned within the first year, the company actually lost money! And then they need to wait another year before they get profitable on the next customer which replaces the churning one.
For Company B, churn matters less. The CAC payback is immediate, so when a customer churns, they've still been profitable, and most importantly they can simply continue their marketing to find another customer to immediately replace them.
In other words, the importance of churn is a lot higher when it breaks your unit economics. When CAC payback is fast, this is often not the case.
(This effect applies even more so when acquisition is organic, meaning there’s no CAC to pay back at all! This can eventually happen for consumer market leaders that “become a verb”: Uber, Airbnb, Google, and many others.)
From a leaky bucket perspective, rapid CAC payback means a powerful stream. Again, it’s better if there are no leaks in the bucket. But if your bucket is leaky, your only hope of filling it up is if there’s a more powerful stream of water flowing into it than the leaks are letting out.
"Ah!", you might interrupt, "but what about when Company B runs out of customers? What about when the powerful stream runs out?"
That leads us into the second variable…
Variable 2: Size of market (the infinite reservoir)
Everyone would agree that big markets are better than small ones. But this is a slightly subtler point: the size of the market affects the importance of the retention.
Again, this is best illustrated with some examples.
Let’s start with some markets that have very limited pools of customers.
Scale AI is a good example here: they provide data labelling for LLM, self-driving, and other AI companies. Let's assume for the sake of argument that this core labelling product has only 100 major addressable customers.
This means that each time a customer churns, they lose 1% of their long-term revenue potential.
And if you're selling to the Fortune 500... well... there's only 500 of them.
So even if CAC payback is instant, i.e. there’s a powerful tap filling your leaky bucket, high churn will still kill your business because the water will run out.
Now consider the inverse scenario, where TAM is near-infinite. In the language learning app space, for example, where we've invested, there are over 1.5 billion language learners: no app will ever exhaust them. It would theoretically be possible to keep churning through them forever, and then ultimately market to their children and grandchildren decades later, and still have more to come. In the meantime you can create many billions of enterprise value as Duolingo have. All the while with incredibly high churn, and yes, a leaky bucket.
The same goes for many of the examples above: luxury handbags, automobiles, and mobile gaming: the churn matters less because the TAM is so damn big.
You can churn through users forever.
We can call this the infinite reservoir. If you’re filling your leaky bucket with a fast-flowing tap from a vast pool of water which you can be certain you’ll never run out of, your churn matters a lot less.
Why does any of this matter?
Churn is obviously still a Very Bad Thing, for any business. It creates challenges for growth, unit economics, morale, product, and many more issues. CAC payback can deteriorate, and competitors can eat up TAM.
So to all the haters shouting at your screen: yes, I know.
In a vacuum, obviously, low churn is always better than high churn.
But the point here is that many investors miss the nuance, perhaps because of a bias from enterprise software, where (i) CAC paybacks are generally long and (ii) customer pools are generally limited. This has led many to over-apply the same B2B SaaS retention frameworks to consumer without thinking through the nuance.
So, I urge that next time you are looking at founding, investing in or joining a business in a consumer category with high churn, and your VC friend tells you it's a bad business because of their churn rate, please, ask yourself two questions:
How powerful is their stream of acquisition economics?
How infinite is their reservoir of market size?
Then, and only then, will you fully understand how much of a chance they have of filling up their bucket, leaks and all.
Reply