One Company's Net Dollar Expansion Is Somebody Else's Lower Gross Margin

Plus! PE can afford to be subpar, live data on the China lockdowns, and DTC metrics explained

One Company's Net Dollar Expansion Is Somebody Else's Lower Gross Margin

When a small B2B company raises money, there's a short list of companies it wants to be "The Next..." AWS, Stripe, Salesforce, Twilio, Shopify--these companies have incredible economics (and incredible valuations) in part because they grow with their customers. When a Shopify customer doubles their sales, Shopify's revenues rise right along with it. Same for Stripe and Twilio. AWS usage tends to scale with growth as well, sometimes linearly, sometimes sublinearly (if a company needs to run a bunch of upfront compute to produce something it can sell repeatedly), sometimes, as I mentioned last issue, linearly or worse.

A Twilio investor can tolerate high sales and marketing expense because Twilio's net dollar retention is running in the 120s right now; if they never closed another customer, they'd be growing 24%.

There's an underappreciated flipside to this, though. Since investors have a strong preference for companies with net retention rates over 100%, they're building a software ecosystem where everybody has a higher cost of goods sold. And the entire SaaS model is built around high upfront costs for product development and sales, followed by high incremental margins.

This gets particularly scary because every SaaS company gets valued based on what its mature steady-state would look like. A big company gets valued this way explicitly; out-year estimates for a company like Salesforce tend to assume a graceful deceleration in sales growth coupled with improving margins. And newer companies get valued this way implicitly, because they get comped to more mature businesses.
And yet, there aren't that many examples of mature SaaS companies. Salesforce will do about $17bn in revenue this year, and their operating margin last year was 4%. If a $10bn+ business isn't mature yet, either every SaaS founder plans on making absolutely dynastic sums of money or most of these companies will disappoint.

Of course, you could argue that Salesforce is a special case, and you could point to AWS's mid-20s operating margin as a counterexample. But it's just one counterexample, and AWS a) was early, b) is a really exceptional suite of services, and c) benefits from economies of scale in the most rapidly-scaling category of spending in the world. AWS is also a complement to Amazon's much less profitable retail operation. If there's an industry with one huge outlier, and that outlier is also the only company in the industry with a particular hard-to-copy trait, it's safest to assume that this trait is why they're an outlier.

But AWS is also notable because they're the number-one culprit in the A16Z AI bear case I linked to last week. If one company can support good economics in part by causing an entire category of companies to have bad economics, we really shouldn't think of all these firms as being in the same business.

Here's how it will shake out: some companies with high dollar retention will retain their exceptional economics, and continue to have high-margin growth and low costs of capital. Their cheap capital and (eventual) operating profits will allow them to build in-house tools that duplicate the functionality they'd otherwise pay for. In some cases, those tools will themselves become independent businesses, though this will be pretty rare: as Google figured out a long time ago, once you have a monopoly, any adjacent product that could be a good business works even better as a complement to that monopoly.

For the SaaS companies that don't achieve this escape velocity, there are two possible outcomes:

The usual pattern is that they grow spending roughly in line with revenue, and when revenue growth decelerates they don't cut spending fast enough. The result is a business that bleeds cash and requires a painful restructuring.

The better exit is to accept that scaling won't save margins, and transition to a consulting model. This will be very disappointing for investors, but a high likelihood of an okay return is better than zero chance at a great return (for which the alternative is... going to zero).

This means that the open question for SaaS dollar-retention champions is: how much of their revenue is dependent on venture-backed companies? I don't know the answer (and they don't have any incentive to disclose it). But the dollar retention numbers are suspicious: if a company can maintain 20%+ net retention for a long period, it implies that their customers are growing revenues at about that pace. And the vast majority of companies grow a lot more slowly than that. In fact, there's only one slice of companies that I'd expect to generally grow that quickly: they're not SMBs, they're not public. They're venture-backed.


A new report from Bain shows that US private equity returns for the last ten years were +15.3% Y/Y, compared to +15.5% Y/Y for the S&P. This is a striking addition to the long list of claims that investors should just invest in indices instead of paying someone 2 & 20. But there are several reasons to think it won't really matter:

  1. The last ten years were great for growth companies, the exact sorts of companies that private equity under-indexes on.
  2. A well-timed report from the smart folks at Verdad points to another reason: PE activity declines significantly during times of financial stress. (Why? Because PE funds have to call capital to do deals, and when times are very bad, their investors will strongly suggest that they not ask for any cash. This makes PE investors structurally poor market-timers. If you want a private equity vehicle that can buy during the next depression, it needs permanent capital.)
  3. PE investors are, in part, paying PE firms not to mark their assets to market. I wrote more about this dynamic here.

Everyone is deeply freaked out about Coronavirus. I felt like a goofy prepper when I ordered masks a few weeks ago, but yesterday the market absolutely freaked out about coronavirus spreading to Iran and Italy. Financial analysts are pattern-matchers, and diseases that spread from China to Italy are, historically, bad news. But a reader sent over a very interesting datapoint on the situation in China: TomTom has real-time congestion tracking for cities around the world, and traffic in China is finally picking up. In Shenzhen, the morning commute on Monday was actually busier than average.

The standard Coronavirus trades are 1) sell everything but gold, 2) sell everything in China, or 3) buy Zoom (up 53% YTD, vs the Nasdaq 100 up 2%). Is Zoom a ridiculously speculative bet? If the world ends, we won't need videoconferencing; if it doesn't end, Zoom gets one or two good quarters before things settle down. There's a third possibility, though: bad news only becomes a crisis when we weren't prepared, and big companies don't like to suffer from the same crisis twice, except in a very abstract sense. (For example: Coronavirus is a crisis because so many of the world's electronics supply chain pass through Shenzhen; 2008 was a crisis because so many of the world's dollar-funding supply chain passed through financial institutions that were levered long complex structured products.) The bull case on Zoom is not a temporary bump in sales, but a long-term shift in behavior.

A little old, but "DTC Metrics, Explained" is a fantastic primer on the math of unit economics. Unit economics are all simple arithmetic, but it's very easy to miss the importance of compounding gains through a complex sales funnel.

If you want to think about unit economics in an even more elaborate way: