We posted a note a few months ago that analyzed the marked decline in EV/Sales multiples for software stocks while trying to analyze whether the derating was nearing an end. We noted that projected growth was still at an all-time high, while EBITDA margins had systematically declined from the mid-2000s, resulting in a Rule of 40 that was roughly in flat.

Since that time, expected revenue growth has remained elevated while multiples have continued to collapse, going from 4.7x when we wrote the note in March to 3.2x now for the “median” software company.

Adjusted EBITDA margins have also stayed fairly static meaning the traditional Rule of 40 score for the median software stock has stayed relatively constant at 34%.

It occurred to us that there could be other variables potentially masking the true change in underlying profitability of software companies. To investigate, we looked at two tricks I’m aware of that management teams employ to overstate their EBITDA margins:

1) Capitalization of Research and Development (R&D)

2) Stock-based Comp (SBC)

There is some discretion over how companies classify their R&D expenditures. Some companies report all of their R&D as an operating expense while others capitalize some of their R&D cost and then depreciate it over time. If a company chooses to capitalize the R&D it will never show up in EBITDA because well… EBITDA is earnings before depreciation. For this reason, we often prefer EBITDA - Capex or EBIT when doing apples to apples comparison of companies.

As for SBC, this expense is typically added back in the calculation for Adjusted EBITDA to inflate the number. While it is true that stock compensation is not a cash cost it is still a very real (and consistent) cost to the company as one of two things will ultimately happen:

Dilution will increase as shares are issued to pay employees.
The company will buy back shares to avoid dilution and keep the share count flat, which of course uses cash, effectively making it a direct cash cost.

Depending on the level of the stock comp, this can really dent a company’s underlying margins, and since it’s an annual exercise of granting shares to compensate employees, it’s usually not a “one time” thing that will go away.

So, the question is, have software companies been utilizing this trick more recently to mask underlying profitability behind the guise of Adjusted EBITDA?

Here is the median Capex/Sales for software companies:

Well, so far so good. While we are currently slightly elevated versus the 2% - 2.5% level we saw during the 2000s, we are down from a peak of 5% in 2015-2016.

Now what about stock comp as % of sales:

Now this is interesting. Since 2016, the median software company has increased their stock comp rate as % of sales from 4% to 9%. That’s effectively 500 bps of hidden margin compression.

If we compare the median Adjusted EBITDA margin vs. what we might call the “True” Operating margin (Adjusted EBITDA- Capex - SBC), we can see that not only has there been a persistent decline in both numbers, but the gap has been widening. As illustrated by the gray bars, the gap between Adjusted EBITDA and True Operating margin has been as low as 4% historically, but now sits at a record high of 12%.

Perhaps evem more jarring, the median software company’s underlying True Operating margin is only 1%.

Below we show the Rule of 40 using traditional Adjusted EBITDA vs True Operating margin. You can see the slope of the line is steeper for True Operating margin, implying that even after accounting for growth, there has been underlying deterioration masked by increasingly exorbitant stock comp.

While in a perfect world we would do more analysis around actual new stock grants, as opposed to just SBC expense which is amortized and can be driven by stock price movement, we believe this is likely still directionally correct. Perhaps more importantly, we can identify some of the biggest offenders while highlighting names that have stronger underlying “True” margins.

Who Are the Biggest Offenders?

We took a representative sample of pure software names and show their difference between Adjusted EBITDA margin and “True” EBITDA, which we show below:

One offender is Bill.com. They trade at 14x NTM sales and have a traditional Rule of 40 score around 44%. However, their stock -based comp is 34% of sales, along with capex at 3%, meaning they only have a True Rule of 40 score of 7%.

Zscaler, another high-flier, still trades at ~15x sales with 38% growth and 15% headline EBITDA margins. However, they have 39% of their sales in stock-based comp and 8% on capex, meaning their modified Rule of 40 is only 7%. Should this still get a 16x sales multiple?

Other material offenders include Confluent, Coupa, Okta, SentintelOne, Palantir, and Snowflake, among many others.

Who Still Holds Up?

On the other hand, companies like Datadog stand up reasonably well. Yes, they are at 18x NTM sales, but their 46% projected sales growth is supplemented by a headline 19% EBITDA margin and “only” 17% of sales as SBC, meaning they still have a Modified Rule of 40 of over 44%. Other higher multiple names that still hold up well include Paycom, Veeva Systems, and Qualys.

Conclusion

The EV/Sales multiple is ultimately a short-hand proxy for earnings (assuming some kind of steady state operating margin). The traditional thought is that software companies at some level of scale can hit 30% - 40% operating margins in a lower growth environment, hence the Rule of 40 (trade off from growth to profitability). If you’re trading at 10x sales but you can double your sales in three years and then hit 40% operating margins at 0% growth, you are effectively trading at 12.5x EBIT on a plateaued sales level, not quite as insane as might first appear.

However, what this appears to show is that, on average, there has been an underlying deterioration in growth adjusted profitability in the sector. This in turn should impact what EV/Sales ratio an investor is willing to pay, given the implied lower margins. Ultimately, as investors become more discerning and realize that their operating profits are increasingly and subtly going to employees rather than them, there may be a reckoning on egregious levels of stock comp. If there is a sustained recession, we would expect margins to rise as they did post 2008:

Until then, we believe it’s prudent to be cognizant of the major implied differences in valuations at a stock level and whether we need to be a bit more conservative in assumed EV/Sales multiples as proxy for earnings shorthand.

Voss Capital

Stock Comp in Software