Key: On Valuation and our companies

Thoughts on traditional EV/S, and why EV/S has little to do with our companies.

Some who are new to the board seem almost personally offended by the fact that our stocks usually have EV/S ratios which are very high by traditional EV/S standards.

I don’t have the answer to what is “overvalued,” but I know that traditional EV/S ratios have very little to do with our companies! Our companies are profoundly different than the companies that EV/S was traditionally used for. Why? Here are some reasons:

First of all, a company with 70% to 90% gross margins is worth a much higher EV/S ratio than a company with 30% or 40% gross margins because each million dollars of sales is worth so much more to the company in take home dollars.

Just think about this for a minute. If you have 85% gross margins, a million dollars in sales is worth $850,000 to you. If you have 42% gross margins (still quite acceptable), the same million dollars in sales only brings you $420,000. Now really think about that. How can you put the same million dollars in the denominator of EV/S and expect to get a sensible value? Our company with an 85% gross margin is naturally worth twice the EV/S of a normal company with a 42% gross margin, other things being equal.

And a company with a 28% gross margin (believe me, there are plenty of those too, in the real world) only keeps $280,000 out of that million in revenue. How can you put the same million dollars in revenue in the denominator of EV/S for all three of those companies??? Our company with 85% gross margin is naturally worth three times the EV/S sported by the 28% gross margin company, other things being equal… But… other things aren’t equal!!!

Secondly. For a company that is leasing software that becomes integrated into the core of the customer’s business, and with a subscription model that brings in recurring revenue, each million dollars of sales today is not just for this year. It’s for next year too, and the year after, and the year after that, and…. pretty much forever. No one, simply no one, is going to tear out a system that is core and essential to the smooth running of their business, and that would disrupt their entire business to pull out, to save a few dollars. It ain’t gonna happen folks.

Okay now, you have a million dollars of sales this year that will, for all practical purposes, be there next year, and the year after too, and new sales next year will be an extra bonus added on. When you put that million dollars into the denominator of the EV/S equation, what do you have to multiply that million dollars by to take into account all those future years of recurring revenue? By three? By four? By five? That sure brings down the real EV/S for our SaaS companies, doesn’t it?

Compare it to a clothing manufacturer (just for instance). It sells 100,000 coats this year, but has no idea if it will sell 100,000 coats next year, or even 50,000 (maybe another brand will be in fashion). Recurring revenue on a subscription sure beats the heck out of that, doesn’t it? At first glance that clothing company example may seem irrelevant. But no, the EV/S of maybe 3 or 4 that it carries, has helped to shape the idea in your head of what a EV/S normally is. But if the clothing company’s EV/S is 3, if one of our companies has the same revenue (the same S in the denominator), what should its EV/S be? Four times that? Six times that? Ten times that?

Thirdly. But wait! This year’s customers will spend more next year. Our companies have a dollar-based net retention rate maybe averaging 125% or so. That means that this year’s sale revenue isn’t just going to recur next year, but it will be 25% bigger next year, and bigger the year after that. Well of course a company with a 125% dollar-based net retention rate of recurring and growing revenue will have a higher EV/S ratio, than a normal company with the same revenue, the same S value, down there in the denominator, which may not even be there at all next year … (duh!)

Fourthly. And then there is growth rate! Well, of course a company that is consistently growing revenue at 50% to 80% is going to have very high EV/S ratios, because in just two years a consistent 60% growth rate means they will have more than two and a half times as much revenue as they have now. That’s in just two years! (And several of our companies have 90% to 130% growth!!!)

And in three years, that 60% growth will mean more than four times the revenue they have now!

And in four years, more than six and a half times the revenue they have now! That will sure bring that EV/S ratio down, won’t it? I won’t go beyond four years but that’s enough. (You won’t believe it but a fifth year will bring the revenue to more than ten times what you started with. Obviously they don’t need to keep a 60% growth rate to really push up their revenue! That S in the denominator is going to grow rapidly.)

Fifthly. And these companies are capital non-intensive. To double their revenue they don’t need to build a new factory, and buy new machinery, and hire new factory workers, etc or open new stores, or…whatever. They can just double their revenue in the cloud with pretty much the only expense being S&M to get the new customers. And then, as we wrote above, the customers are there almost forever. Just think what that does to traditional valuation.

Sixthly. And being non-capital intensive means that almost none of our companies have any debt. They have cash and equivalents. Inflation won’t mean they have to pay a lot of interest. It means they will make more interest on their cash.

Seventh, and finally, of course a company that is leasing a software solution that every enterprise on the planet needs, and that the vast majority don’t have yet, and that all those companies will keep indefinitely once they install it, will have a higher EV/S ratio than a company selling a product that anyone can put off getting a new model of, or stop buying for the duration of a recession, etc.

Here’s the key to this: You can live another year with your old cell phone, or computer, or car, or raincoat, or refrigerator, or kindle, or ski jacket, or your old factory, or whatever, without buying a new one next year, but once you lease this software, you keep leasing it indefinitely (no stopping for a year.) If you think about that and understand it, you’ve gotten the message!

And of course, of course, of course, companies that have ALL these features…

Leasing software, not a physical thing that requires capex to expand
70-90% gross margins AND
a subscription model with recurring revenue AND
125% net retention rates AND
growing revenue at 50% to 80%, AND
selling products that all enterprises need
low capital intensive
no debt and plenty of cash …

are going to have very high EV/S rates (…duh), and I don’t know what is high, but I will NEVER sell out just because the price has gone up, and because some people think the EV/S is too high. (Although I will trim if my position size has gotten too large). I just don’t know where these companies will ultimately end up.

My decision about my confidence in a company is based on gross margin, recurring revenue, growth rates, dollar-based net retention rates, necessity to their customers, dominance in their field, my confidence in management, and how all that looks to me for the future. Traditional EV/S enters very very little into the equation.

Best,

Saul

199 Likes

They can just double their revenue in the cloud with pretty much the only expense being S&M to get the new customers.

Well, sort of … but remember that these companies have to maintain and support all these customers, so they do need to grow their support organizations as well. However, there is an advantage here as well compared to the tradition sale of software by an up front purchase plus maintenance. In the traditional model, the software tends to be installed on the customer premises. Often, this means that there are customer-specific modifications so that every thousand customers, one has a thousand different versions of the software to support. This can get very messy if the customer makes its own customizations, often leading to making it too expensive to move these customizations to newer versions of the software. Very messy. With SaaS, one tends to have a single copy of the software and, if difference customers need different features or behaviors, that is controlled by switches and options in a common software base, greatly reducing the maintenance headache.

Back in the 1990s when I was selling my ERP package, while the software was installed on the customer premises, I took the approach of working very hard to integrate customer-specific modifications into a common code base and to do continuous updates so that there were never any big transitions for new software. The only customer specific software were things like invoice formats. These days, one would use something like a pdf form and there would be no need for differences in the software itself.

2 Likes

Our companies are profoundly different than the companies that EV/S was traditionally used for. Why? Here are some reasons:

Secondly…Okay now, you have a million dollars of sales this year that will, for all practical purposes, be there next year, and the year after too

Is this really different from Coca Cola?

When you put that million dollars into the denominator of the EV/S equation, what do you have to multiply that million dollars by to take into account all those future years of recurring revenue? By three? By four? By five? That sure brings down the real EV/S for our SaaS companies, doesn’t it?

And for Coke then too.

So I think at least one of those 7 reasons actually is not different at all — making it clear that it is not about this or that point supposedly being different (or not!) but about the combination.

2 Likes

But how is the market valuing “our companies” now compared to historically? Borrowing from Jamin Ball’s excellent free weekly updates at Clouded Judgement, still very highly. You can subscribe free here:https://cloudedjudgement.substack.com/?utm_campaign=pub&…

Regarding EV/NTM Revenue multiples, Jamin writes last week:
“Looking at high growth software median only - we are still 76% above pre-covid highs, 108% above where we were on Jan 1, 2020, and 39% above the previous peak in September 2019.”

I wrote during the last downturn, that I thought that the market would reward Hypergrowth Stocks with higher multiples than pre-covid, as there’s a better appreciation of the business model. But how much higher? Is 76% higher the new standard?

So while we should not use EV/S to compare our companies to other companies that have these very different business models, I do see value in comparing the high-growth software sector valuation metrics as a group (cohort), across time.

When you do that, despite a huge correction, these companies are still valued much higher than historically had been true. How much of that will stick? If it’s still 2x too high, well then, a doubling of revenue will fix that.

5 Likes

Is this really different from Coca Cola?

Yes. With Coke, the Coke you sold this year is gone next year and you have to sell them again. If there is a big move to Pepsi, there is no obstacle for the consumer making the switch. With SaaS, the software you sold this year is still there next year and any new sales add on to the total revenue.

Perhaps this is clearer if one contrasts it to the traditional software sales. There, suppose one has a product that one sells for $200,000 and whose maintenance is 20% of the purchase price. Year One, one gets $240,000 and all subsequent years one gets $40,000, assuming you keep the customer happy. With SaaS, year one is maybe $80,000 and so are all the other years. Over 5 years the traditional model brings in $400,000 and so does the SaaS model. But, every following year the SaaS model brings in $80,000 and the traditional one only $40,000. So, the sale is easier to make because the year one cost is much lower.

This is nothing like Coke.

I think the usual rule is that the total cost is equal in something like 4 years, so perhaps the amount above should be more like $100,000 so that in 5 years one is $100,000 ahead and gaining at $60,000/year more with SaaS.

Plus, having built the software so that it can be shared across all customers, the cost of implementation is much less.

But, the key point Saul is making here is that the SaaS customer continues to be a significant source of revenue year after year after year with no new sales required, thus additional sales are additive to revenue, not merely replacing what one made last year.

15 Likes

MFChips, if we widen out our scope a bit, the story changes. Here’s the rest of what Jamin Ball wrote. Emphasis mine:

"When looking at overall median cloud software multiples - we are still 12% above pre-covid highs, 35% above where we were on Jan 1, 2020, and 10% above the previous peak in August 2019. However, if you remove companies that went public during the pandemic, the overall median is right back to where we were pre-covid.

One other way to slice the data is to look at the top and bottom quartile of multiples. Currently, the top quartile multiple is about 2.4x as big as the bottom quartile (18.4x vs 7.5x). Interestingly, this is quite similar to where we were pre-covid at 2.4x (16.3x for top quartile vs 6.9x for bottom quartile)."

I have no idea where we end up, but I think these additional statistics help complete the picture.

20 Likes

Static multiples for a company only make sense if compared to a cohort of similar firms. Too often people look at P/S or P/E of a growth firm and compare it to an average metric based on more mature firms. Rather than look at multiples I like to model out what the possible cash flows might be as a rough gauge of value.

With enterprise SaaS companies I’ll model revenue growth going down to 80% of what it was the previous year for each year in succession until it hits more of a steady-state growth rate. If you look at historical examples of older best-in-class SaaS companies this is about what to expect. At the same time, I model margins going up to about where older software/SaaS companies are averaging and what I believe each company can achieve based on its own business model (ex. Monday should be able to get higher end state margins than Snowflake or even CrowdStrike since it has relatively low compute and storage costs).

I’m discounting the estimated cash flows back by a pretty decently high 9.5%, which I believe takes into account a full interest rate cycle. I’m assuming taxes will go up. I think I’m being decently conservative.

What I find when doing this is that there’s good value among many of these enterprise SaaS names right now. I’m not the only one who uses this method. Bert Hochfeld also uses cash flow estimation approaches and sees similar values. Even Morningstar, which has historically been quite conservative when it comes to valuing tech firms, believes Snowflake, CrowdStrike, and Zscaler to be about fairly valued right now, not over-valued.

Mike

28 Likes

we are still 76% above pre-covid highs

Well of course the valuations of our companies are above what they were in 2019. In 2019 most of them were just-IPOed tiny companies with no track record.

In addition we where very early adopters, in that the people on this board had figured out all that stuff I wrote about in the starting post of this thread “On Valuation and our companies” way back when “SaaS” was just two "S"s and two "a"s to most investors. How else do you think we tripled or quadrupled our portfolio values in 2020? The general investing community figured it out too (eventually), and bid our companies’ valuation up to where it should be considering those seven points.

Saul

33 Likes

"When looking at overall median cloud software multiples - we are still 12% above pre-covid highs, 35% above where we were on Jan 1, 2020, and 10% above the previous peak in August 2019. However, if you remove companies that went public during the pandemic, the overall median is right back to where we were pre-covid.

So that is a very important point above, thank you PeterHay.

As I consider this paragraph above, I can see how there’s a valid and useful point to removing companies that IPO’d during this time of change. Essentially, to compare before to now, we must remove the companies that entered the picture during the interim to get a more pure comparison.

Doing so removes many high growth companies including SNOW, MNDY, ASAN (as examples) and leaves behind a higher weighting of companies whose growth has started to slow. So, correspondingly, their valuation metrics would be lower now than before for EV/NTM Rev.

So, it’s not clear if this tells the whole story. Would really love to take the scatter plot data he has shown (EV/NTM Rev Multiple vs NTM Rev Growth) and plot that vs time over the last 2-3 years. Because that would help account for changing growth rates in the cohorts.

1 Like

IMHO anytime you have to use a multiple of Sales instead of Earnings it’s usually about a healthy little seedling that is already credited as having a future state as a giant oak. But lots can go wrong during the period required to reach that future size.

While I agree with many of the points made in the OP, I disagree with the implication that any stock price you pay in the early stage doesn’t matter. It may be true in the very, very few that eventually become an Amazon or Google (giant oaks) but most will never achieve that level of success and the price you pay will matter.

Anytime there are fat profit margins there will be competition and only a very wide moat teeming with hungry crocodiles can ward it off. I don’t think that most Saas providers have such a lasting moat. I work for a giant engineering firm and every few years we switch to another CAD software package because the price and features are compelling enough to endure the inconvenience. I think this will be an ongoing issue for SaaS providers.

11 Likes

I just wanted to re-iterate this key point from Saul.

My decision about my confidence in a company is based on gross margin, recurring revenue, growth rates, dollar-based net retention rates, necessity to their customers, dominance in their field, my confidence in management, and how all that looks to me for the future. Traditional EV/S enters very very little into the equation.

One of the biggest mistakes I see some people make (and one that I used to make all the time!) is that their conviction sometimes goes up because valuation goes down.

“If you liked it at $400 you should like it more at $200.” Yes, but what has changed? And were you right to like it at $400 in the first place? Sometimes you have to admit you were wrong, and the story wasn’t what you thought. Or maybe the story has changed.

At the end of November, Upstart had fallen 49% from it’s all time high, and Datadog had fallen 11% from its ATH. Datadog’s trailing PS was still about 70 and Upstart’s was closer to 30. Many people thought this was an “opportunity” on Upstart, but nobody said the same about Datadog. In December, Datadog is down 8% and Upstart is down 36%. Companies that meet the standards Saul outlines above are expensive for a reason, and they tend to stay that way until something changes with the company.

All this said, I have re-taken a 2% position in Upstart. But my Datadog position is close to 22%. I don’t have a perfect answer to anything, and Saul and the other great investors on this board would probably say the same. I simply urge everyone to be honest with yourself – are you ignoring the wisdom espoused here? There’s a reason no one on the board is touting how they have huge gains from the opportunities they identified with Fastly or Nutanix after the valuation got “cheap.”

Bear

103 Likes

“Sometimes you have to admit you were wrong, and the story wasn’t what you thought. Or maybe the story has changed.”

Or you need to be patient and let the story play out a bit longer.

Let’s say UPSTs earnings were a tad more in line to the expectations on the board, and guidance was a bit closer, or there was a bit more joy in the earnings call, or maybe none of that needed to happen. Let’s just go with, what if the stock price, instead of cratering, went up, and was now at say 600. How much would have all the discussions changed around that scenario.

Maybe judging UPST on one quarter, on one conference call, is a little short sighted. Maybe no one should have had more then 10% of their portfolio in one name without a longer track record. We can do this all day.

Not every stock is going to go straight up, or straight down, and some take time to develop. Hindsight is always 20/20. Will anyone here be surprised, or shocked if UPST signs a major deal, or many deals, that earnings are better then expected, that guidance goes up, that the stock performance over the next one to five years, maybe ten years is nicely up and to the right? I won’t be.

70 Likes

A general rule is that all firms are valued as a sum of future profits; this is true for pretty much any firm. So when you see a firm such as Snowflake trading at a market cap of 100B, it is safe to assume that at some point it should be able to generate $5B in profit.

This is where the margin calculation and growth rates come in. If they are able to get their profit margin (not gross!) to say 30%, they would need to generate $16B - $17B in revenue. Last year they were able to generate just under $600B; but that was up over 123% from the previous year.

Pretty simple to run the numbers from there.


Year      Revenue      Growth Rate
1         $0.6B        100%
2         $1.2B        100%
3         $2.4B        100%
4         $4.8B        100%
5         $9.6B        100%
6         $18.2B       100%

So you can fiddle with the growth rate (and even the profit margin assumptions), but you can see how much future growth is already priced into the stock.

Note: The business model is not super relevant to the end-state; but is very relevant to how they get there. A subscription model will be much easier to grow at high rates than a traditional business, but it doesn’t really change the end-state math.

Maybe SNOW will grow at faster than 100%, maybe they can get their profit margins higher than 30% - feel free to play with the numbers, but make sure you understand what the implications are.

tecmo

21 Likes

Great write up Saul, but I would like to make a couple of points pertinent to your comments.

  1. I think you have clearly demonstrated why EV/S is not a particularly relevant stat for making an investment decision in a company with the SaaS business model. But, EV as a stand alone number is rather important. As the company’s overall size increases, the law of large numbers becomes more and more relevant. Salesforce and Netflix don’t find much love here, though it could be argued that they are both fundamentally SaaS companies (please, don’t start a thread on the validity of this example, I picked on them just to illustrate a point). As you well know, the larger a company becomes, the harder it for that company to double and double again. I know you know this. You’ve make the same point elsewhere. I’m simply restating it for the benefit of folks who may not understand that size makes a difference.

  2. And to your second point that no company is going to replace s/w that’s integral to their operations - well, that’s not correct. They do this if there are compelling reasons. When I worked at Boeing (I retired in 2010) the commercial division ripped out all their mainline engineering and manufacturing business systems and replaced them with new software (this was an enormous multi million dollar, multi year project). Allan Mullaly, who was president of commercial at the time (later, CEO of Ford), described the project as akin to changing the tires on your car while driving 60 MPH on the high way in heavy traffic. I won’t go into details about the rationale for doing this. The point is, there are at times compelling business reasons to change s/w that is highly integrated with the operation of a company. The concept of what actually constitutes a moat is critically important. Let’s look at CRWD and S as an example. Does CRWD have an unassailable moat? I don’t think so. I can’t see anything (this may be due to my ignorance) that would inhibit a company from hosting both applications (for a while) in order to determine which one they will subscribe to on a long term basis. I’ve reduced, but not sold out my CRWD position, and I’ve not initiated an S position, yet. But I’m scratching my head on this one. The point is a previously thought CRWD had a strong and sturdy moat, now, I’m not so certain. Another reason a company might change their s/w if they build a new factory for a new and important product. When Boeing committed to the 787 model, this is exactly what they did. They re-evaluated their entire gamut of business systems and made a lot of very significant changes. I’ll just mention as an aside that I was one of the people who reviewed the 787 s/w plan. I suggested that they had failed to recognize that they had changed their business model to one of being a supply chain management and large system integrator from the prior engineering/manufacturing model. Unfortunately, I was over ruled by the engineering contingency who were in charge of the project.

10 Likes

And to your second point that no company is going to replace s/w that’s integral to their operations - well, that’s not correct. They do this if there are compelling reasons.

It should be noted that, with the traditional software sales model, there are a number of reasons helping this kind of replacement to happen. Worst of all, perhaps, is the over the transom type of sale where the original vendor assumes no responsibility for maintaining the software; everything is done by the purchasing company. And, of course, the purchasing company isn’t expert at software, so they typically don’t do a very good job and after some years it is clear that there are other products available which would support them better.

Even if the vendor continues to provide support, if the purchasing company does a lot of their own modifications to the software, then it is likely after a period of time that the vendor will come out with a new version and the purchasing company is often unwilling to spend the large amount of money necessary to apply all their modifications to the new version and thus they fall behind the state of the art and eventually get to the point where something new from someone else is better than what they have. This is often true because of evolving architectures like moving from server-based to client-server to web architectures.

SaaS tends to avoid this problem as there is typically one copy of software used by everyone and customizations are added controlled by option switches and configuration so that the using company gets the behavior they want, but the software base is not divided. Companies which are really good at this … which presumably our companies are … should thus be less susceptible to replacement.

9 Likes

Bravo tecmo,
Your criminal misrepresentation of numbers begs correction, both for Snowflake’s sake and for the sake of using Snowflake (certainly one of the most “overvalued” in our hypergrowth universe) as a poster child of how quickly top-line hypergrowth can make even a nosebleed multiple like SNOW’s (forget about entirely reasonable multiples like ZI, AMPL, or UPST) come down to earth.

You say “Last year they were able to generate just under $600M [in revenue].” Your post is dated December 20th, 2021, so you must have been referring to the four most-recently reported quarters, which would be Q4 FY20 - Q3 FY21 (Nov 1, 2020 - Oct 31, 2021).

Year   Q1    Q2    Q3   Q4
FY19                    88
FY20  109   133   160  190
FY21  229   272   334

Wait a minute…that can’t be. Because the four most recent quarters (TTM) is actually $1.025B, a far cry from $600M. Ohhh, I see what you did, when you say “last year” you meant a year before a full year ago (Feb 1, 2020 - Jan 31, 2021). Interesting…I’ve never seen that before–it smacks of bias but I won’t jump to conclusions. And for a company like Snowflake, which is more than doubling revenues YoY and accelerating QoQ, that actually makes a huge difference.

So when I look at your table, “Year 1” is actually a full year ago, and as of your post date, we are about 90% of the way through “Year 2.” Let’s make your table again, except now, let’s call the TTM “Year 0” and the NTM “Year 1,” and so on…

Year    Revenue   YoY Gr
Year 0   $1.03B     109%
Year 1   $2.11B     105%
Year 2   $4.01B      90%
Year 3   $7.02B      75%
Year 4  $11.58B      65%
Year 5  $17.37B      50%
Year 6  $25.19B      45%

You’ll see that I not only arrived at a “Year 6” (o/a 2027) revenue level that is quite a bit higher than yours, but I did so with using significant deceleration each and every year along the way–instead of the tongue-in-cheek perpetual doubling that you used. What’s more, such deceleration conservative-ness is not only prudent for Snowflake–some may say it’s pessimistic. I mean, Snowflake’s NRR just clocked in at 173%–and rising. They also just reported a Q3 that saw sequential top-line acceleration from 104% to 110%. Slootman basically browbeats us every quarter reiterating that customers don’t really “spend” until ~9 months after onboarding, so it would seem that such outperformance/acceleration is possibly hitting an upward inflection. And one more thing, Amazon Web Services is 19 years old, is at $45B in revenue, and it just grew top-line by 40%. So I’m modeling (being conservative) that Snowflake hits maturity far faster, and at a far lower level, than AWS.

Alright, let’s fast forward using my above (conservative) estimates. It’s Christmas 2027 and basically the world runs on data and the data runs in the cloud. I’m sitting on my couch watching the Vikings miss the playoffs again when Snowflake reports $26B in revenue for the year and is looking at $35B in 2028 (a slowdown to 35% growth). A 30% net margin keeps food on the table, and is not terribly unreasonable (MSFT, ZM, ADBE, NVDA, and others are in this range or higher). What’s a deserving P/E–well the average of those just mentioned is a mind-numbing 60x (thanks to NVDA). So let’s just use ADBE, it’s growing in the low 20% range and has a 30% net margin you can set your watch to. It’s P/E…55x.

Alrighty, forward revenue $35B x 30% Net Margin = $10.5B Profit x 55 P/E = $577B Market Cap on Christmas Day 2027. A 34% CAGR over six years from today’s $94B…for one of the “overvalued” companies in our universe.

I’ll end with this to hopefully make a broader point. Just today, Zoom (not a company I own but one I admire) was listed on CNBC alongside DraftKings and Peloton as “overvalued.” It’s spoken as matter-of-fact, unchallenged, of course Zoom is overvalued, it’s Tulip-mania. The whipping post of hypergrowth-gone-wrong.

“Overvalued”

ZM today has a lower forward Price-to-Earnings multiple than…(wait for it)…Proctor and Gamble.

What Else is There To Say?

Eric Przybylski, CPA

79 Likes