As an investor in publicly traded SaaS companies, valuation was a primary concern. Again and again, I met with companies with "stories" that would make my counterparts covering other sectors drool: secular growth, legacy competition lacking the financial flexibility or will to adapt, accurate guidance based on contracted recurring revenue, corporate cultures built for attracting top talent and cruise ship-esque working conditions to retain that talent. The rub for both myself and portfolio managers was getting comfortable with the valuation: I often described valuing a SaaS company as akin to catching a butterfly with a pair of tweezers, and for portfolio managers with memories of market downturns companies not valued on near-term earnings/cash flow will always be cause for some concern. They know all too well that when investors' greed turns to fear, cash flow is re-crowned king and growth heads off to exile. That said, after being a skeptic at first I've come to appreciate that SaaS companies often deserve the valuations the market gives them and then some, even as I am convinced that EV/Revenue is the wrong way to value them. In this post, I'll cover the background of why we've taken to using such a silly technique in the first place, as a precursor to covering some of its key flaws and some ideas for improving how we think about valuing these companies in a separate post.
The challenge can be tied to the fact that the age-old tools of academically minded value investors, discounted cash flow valuations and P/E ratios, are practically useless when faced with a subscription revenue model. To see why, I'll discuss briefly how DCFs work in practice and why we abstract them to multiples, then look at how theory and practice clash and why SaaS companies are valued using EV/Rev instead.
How DCFs work in practice:
As a refresher, a DCF is for practical purposes an accounting identity: investors enter an expected stream of cash flows, and then supply the discount rate, or that rate of return they expect in exchange for the risk level of the investment. With those two variables, one can pull all of the future cash flows forward into a value today, at which the investor should be indifferent between the future stream and today's lump sum payout. Companies, in theory, are worth this point of indifference: the Mr. Market is effectively saying "over the price you trade at today, I'd rather hold cash or something else, but under it I'm willing to buy."
In practice, investors don't model the cash flows of a company into perpetuity because it is inaccurate and (appropriately) silly feeling. Instead, they often model over a period of 5-10 years and then approximate the "terminal value" or what all of the cash flows beyond the final year are worth. If they did value the company out for say a hundred years, then they'd have a valuation as accurate as their discount rate and estimates (which is to say precise but not very accurate), and avoid the methodological concerns I'm about to raise that arise from abstraction.
Where multiples come along:
Almost no one builds hundred year models- instead, they resort to multiples for the terminal value, a way of condensing the information content of the unbuilt portion of the DCF (cash flows and discount rates and the present value of all those case flows) into a single number, expressed as a coefficient on a number we actually have with higher accuracy. This higher accuracy number number is either cash flow or net income from the last year that is modeled, since this is likely the best indicator of what future free cash flow will look like. Investors then put a multiple on it which generally ranges from 8x to 25x, with high growth/quality companies at the high end and slow growing or shrinking companies at the low end. Adding the result of that (discounted of course) to the 5-10 annual cash flows that were modeled annually yields a decent estimate of the present value of the company. Humans don't do math in our heads very well so this is a necessarily significant abstraction: few investors have a strong sense of how "15x earnings" maps into cash flows/discount rates for any given company, but they know that it is a rough approximation that works. I could write pages and pages of the linkages between this convention and actual DCFs underlying it, but unless you're a true finance wonk that will get boring fast (if this hasn't already!).
In the real world, because DCFs are annoying and hard to talk about/compare (see me struggling here?) investors often resort to just using plain multiples and dispensing with even a 5-10 year forecast of cash flows in favor of a few years of forward projections valued using just a single multiple. So we often hear about company valuations in phrases like "at just 8 times earnings, Exxon Mobil is cheap" or "LinkedIn is terribly expensive at 900 times earnings." This is a useful simplification in that it trades the false accuracy of a hundred year model for ease of use, standardization and comparability. It is by far the most common type of valuation work I discussed and grappled with during my time on the buy side for that reason.
Unfortunately, both the DCF approach and its earnings-multiple cousin collapse for subscription models. The challenge is that SaaS companies don't play well with conventional thinking on earnings or cash flow, which is that more earnings/free cash flow in any given period is better than less. Specifically, the faster a subscription revenue company is able to grow over time, the better the company is doing, but the less profitable it will be in the meantime as it spends up front to acquire that revenue. As investors, we really don't want SaaS companies breaking to profitability too soon, because all else equal that means that the company doesn't have as many opportunities for efficient investment of sales and marketing dollars as we first thought. Ideally, SaaS companies will stay unprofitable or near breakeven well into the future, as every incremental dollar of revenue goes right back into acquiring more future revenue streams. This confounds an earnings multiple approach, because we don't want up front earnings per se (and most SaaS companies are smart enough not to maximize for that). It also confounds a DCF approach, unless we build out one hundred years, because a more bullish view of a company's prospects will involve a longer duration of growth and so less cash flow generation over any specified time period.
To get comfortable with this dynamic when I started evaluating SaaS companies, I built a model for a fake company called SaaSco with solid (if not inspiring) Saas metrics. A acquisition cost of 2-3x new subscription revenue, 85% subscription gross margins and 7% net dollar churn. As you can see from the chart below in which each of the 21 years I modeled is represented by a red dot, SaaSco doesn't break to positive GAAP operating margins until revenue growth slips into the mid-teens. Growth rate, not time or scale, is the variable that drives profitability, and a free cash flow multiple doesn't become usefully indicative of the true value of the company until it grows less than 10%.
Looking down from space: EV/Revenue multiples
We've covered how several factors push investors to use multiples instead of the more theoretically rigorous valuation approach represented by a DCF, but I've also explained how SaaS companies don't play nicely with DCFs or the basic multiples investors use for most "normal" companies, especially when they are growing quickly. None of this was particularly insightful: we value SaaS companies on EV/Revenue, everyone knows that. But all that exposition had a purpose in that it shows us the process of elimination through which better valuation techniques fall by the wayside and we're left with a "least worst" approach. That is what EV/Revenue multiples are, and it is crucial to remember is where they came from and that they are ultimately a bastardized version of a DCF valuation. If an DCF is an on the ground perspective and an earnings multiple is a bird's eye view of reality, an EV/Revenue approach is like looking down from space. It both introduces huge inaccuracy into the system and can give the false impressive that software companies are super-expensive with multiples that need to compress over time before "sensible, financially savvy" investors find them attractive. In a follow-up post, I'll discuss the most common inaccuracies, explain how the market determines what multiple to put on a company and how I've gotten comfortable with the underlying financial reality behind the high multiples SaaS companies garner from investors.