Don't use revenue multiples to value high growth companies
When people talk about the great 2020-2021 tech bubble and its subsequent implosion, much of the discussion centers around revenue multiples.
Historically (at least in the time I’ve been investing) the standard ARR multiple for a private SaaS company financing was 10x, though perhaps more for a very hot company. My first SaaS deal was priced at 20x current year-end ARR, on the basis that this was “expensive in the short run but cheap in the long run”, as the company was growing at 300% so this really meant 5x next year’s ending ARR. “Expensive” and “cheap” were all relative to this 10x benchmark.
In 2018-2019 the new benchmark for private companies crept up to something closer to 20x. You can see this in the above chart as public multiples over NTM revenue increased from the 5-7x zone to 8-11x territory (private multiples usually are higher than public multiples because private co’s are growing faster). Post-covid19 lockdowns, 100x became the new 20x which was the new 10x:
Now we are back where we started — but what does this mean? Are tech stocks now a bargain? And what does this tell us about how likely we are to make money from investing in any particular company?
The main thing I want you to take away from this post is that although market revenue multiples can tell you a lot about investor sentiment, and a lot about how valuable the market thinks a particular company is (will be), a low multiple for any particular company does not imply that you are getting a bargain, or that something will appreciate in value, or really very much at all about whether your deal was a good one or a bad one.
Problem #1: Revenue multiples assume that valuation scales linearly with revenue
One of the main issues is that once you start contemplating things in revenue multiples, there’s a very strong impulse for thoughtful price discipline to evaporate because “it’s growing so fast it will inevitably grow into its multiple” (this, I am told by investors far more experienced than me, is something that people say during every bubble).
Take a company like Snowflake, a company whose growth and NRR is so mind-bendingly high that it makes a great illustration of these concepts.
source: Public Comps
As of last Friday, Snowflake’s enterprise value was approximately $45B. Analyst estimates are that 2022 revenue (actually fiscal 2023) will be $2B, and that 2023 revenue will be $3.2B. So that’s a multiple of 22.5x this year’s revenue and 14x next year’s revenue.
Here’s where the mind begins to go haywire— “wow, this company is growing so fast, that 14x will become 9x the next year, 6x the following, and so on… actually, this stock is cheap!”
What’s the problem? The problem is that this growth trajectory is already well-known by the market, and already reflected in price. This is especially the case for a company like Snowflake which reports a super robust NRR number that acts as a floor on future growth (at 178% NRR, we can be pretty certain this company will grow by more than 30% next year). These characteristics are why the company has a 22.5x NTM revenue multiple, unlike Google’s which is 5x.
Let’s say Snowflake only hits $1.9B in 2022 revenues. Was value created or destroyed? Well, if the current market price was justified with expectations around hitting a certain minimum level of high growth, you might see the stock price tumble dramatically even as the business has grown 50% or 60%.
Furthermore, to presume that high growth is a free lunch where any sufficiently fast-growing company is guaranteed to compound its way into an improved valuation over time, is to assume that the person trading with you is an idiot. Markets are forward-looking — they are very aware of how the business is “supposed” to perform. If you’re going to play the “well it has to be underpriced relative to 2026 revenues, right?” game, then you’re assuming the market was smart enough to bake in assumptions about 2023 growth but not 2026. On the other side of your trade are a hundred thousand professional traders and AI hedge fund algorithms. Are you sure you want to take that bet?
Problem #2: Revenue multiples can obscure profitability and other measures of value
This is where you can make huge mistakes. Twilio is the poster child for this:
source: Public Comps
Twilio trades at an EV of around $15B. As recently as 6 months ago the EV was closer to $40B. Some boneheads, such as this author (cough cough), started saying things to themselves like “wow, on 10x forward revenues for a 50% grower?! That’s cheap as hell!” Whoops:
Maybe 10x wasn’t so cheap. What was the problem? Well, gross margins are <50%, free cash flow is negative, GAAP losses are very large and growing with scale, competition is catching up, there’s significant dilution from stock-based compensation, and (most importantly) revenue multiples for money-losing SaaS companies completely collapsed over the same time period. If your entire rationale for investing was to compare the revenue multiple to the growth rate, you got hosed.
Is it now cheap at 4x? It may be very cheap (it’s certainly "on sale" compared to 6 or 12 months ago) but I would not recommend looking solely at the multiple to answer this question.
Speaking of that comparison, let’s actually have a look at how the two numbers compare:
source: Public Comps
A few things to notice here:
Yes, of course growth and multiple are correlated (though R-squared is only 0.31, I might have expected higher). It would be pretty peculiar if they weren’t.
Most of the companies at the top-right have high multiples, but the quality of these companies is very good — most generate a lot of cash, have been growing at a sustained clip for a long time, are GAAP profitable or close, and operate in enormous markets with tons of headroom.
This means that companies with high multiples have those multiples for a reason, and growth is only one of many inputs into that multiple. To me this says that revenue multiples are descriptive of assets but not helpful to investors. They tell you how valuable the market perceives a particular asset to be relative to revenue, but they do not tell you if you are getting a bargain by buying them.
Problem #3: Revenue multiples force you to make comparisons to historical prices, which don’t matter
Back to Bill:
Revenue multiples for growth companies are now back to where they were in the 2015-2017 time period. Between 2018 and 2021 they were much higher. Does that mean they are now cheap? We know for sure that they’re “cheaper” but this doesn’t tell us anything about whether they’re cheap. Maybe they’re still expensive — maybe multiples will come down 70% more from here and stay there for a decade… would you still feel comfortable holding the stock if that was the case?
If you’re not a trader or speculator but actually an investor, you can’t get to anything approximating a margin of safety by looking at how multiples have deviated from historicals. You may still make money, but as Gurley says, it’s a “hack proxy” at best.
What to do instead
If you’re an angel investor, or a VC, or a public markets growth investor, I recommend spending very little of your time thinking in terms of revenue multiples.
Instead, spend a lot more time thinking about what kind of profitability this particular business will deliver at scale, how long it will take to get there, and how risky their path is. You don’t have to build a full DCF, but you should develop a viewpoint on where this company is going and whether your expectations differ significantly from the price that’s on offer, without referring to what the multiple on revenue or ARR is. If you can’t complete the thought exercise, then that’s a good indication you’re merely speculating on price.