SaaS Valuation Redux: the SANE Cloud Posse Rides Again
- We previously analyzed public SaaS companies with two regression models to develop a “rule of thumb” for benchmarking SaaS valuations – we called on investors to ignore a “growth at any price” mentality and focus on companies’ balance between growth and profitability
- While revenue growth is more than twice as important to valuation than EBITDA margin, the trend toward increased importance of profitability continues as the SaaS market matures
- The long run trends previously observed have continued: a historically stable relationship with growth, an increasing importance of profitability, and a modest premium for scale
- Over the last year, the SaaS market has oscillated between being fully valued in late 2015, undervalued in early 2016, and now back to fully valued, albeit with a slightly higher average EBITDA margin than last year – while markets notably sold off in March of 2016, improving fundamentals and several strategic acquisitions in the SaaS universe have buoyed today’s valuations back to 2015’s highs
- We extend our assessment of patterns among the companies that achieve higher/lower-than- predicted multiples (based on the formulas) and discuss implications for investors in the space
Last June, we shared the first of what we hope becomes a series of analytical posts on the SaaS ecosystem. Briefly, we suggested a new way of benchmarking the valuations of high-growth SaaS companies by regressing public market comps along the axes of revenue growth and profitability – a process we termed the “SANE” approach. We felt that this approach was more robust than the limited scope of using revenue growth alone and was less arbitrary than the “Rule of 40%” which weights revenue growth and EBITDA margin equally. We developed both a five-factor model (incorporating realized and predicted revenue growth, both gross and EBITDA margins, and company scale) and a simpler, two-factor model that melds the quantitative approach of the five-factor model with the intuition of the Rule of 40% (while capturing the majority of the variance explained by the five-factor model). As of 12/31/15, the 5-Factor SANE model exhibited an R² of 71%, compared to 43% for the 2-Factor SANE, 33% for Revenue Growth alone, and 21% for the “Rule of 40%” formula.
In our prior analysis, we used Q4 2015 data: Q2 2016 wasn’t yet available and we wanted to observe the differing performance among companies assessed by our model. However, this one quarter window was likely too short to generate any meaningful conclusions – and suffice it to say that since then, a good deal has transpired in the SaaS ecosystem. In this analysis, we (i) update the SANE framework for valuing public SaaS companies with recent data, (ii) assess changes in statistical relationships and what they portend, and finally, (iii) use this framework to explore changes in the public SaaS ecosystem.
2) Updated Model & Data
We updated both the 5-factor and 2-factor models in this analysis. As would be expected due to the inclusion of more variables, the 5-factor model produces a higher R² of 0.84 compared to 0.71 for the 2-factor model. Yet, Adjusted R², which adjusts regression results for the inclusion of additional predictive variables, remains only modestly below Actual R² (0.83 Adjusted vs. 0.84 Actual for the 5-factor model and 0.70 Adjusted vs. 0.71 Actual for the 2-factor model). Similar relationships were found as of 12/31/15 and we remain confident in the validity of both models, as evidenced by both this narrow spread between Actual and Adjusted R² and the overall positive statistical significance of both regressions as tested at the 95% confidence level. The correlation of the 5-factor model improved modestly since our last analysis, while the correlations of the 2-factor model increased dramatically.
Reformulated in today’s qualitatively different market environment, the updated 2-factor equation’s statistical significance is even greater and its coefficients remarkably consistent. The intercept (representing the value of the dependent variable (valuation multiple) when the independent variables (revenue growth and EBITDA margin) are held constant) decreased from 2.6x (accounting for 37% of the regression) as of 12/31/15 to 1.9x (27% of the regression) at present, parallel to an increase in R² from 0.43 to 0.71. In tandem, the coefficients on both revenue growth and profitability rose modestly, with revenue growth now accounting for 57% of the regression (an increase from 49%) and profitability accounting for 16% (an increase from 14%). These changes suggest both that the trend toward balancing growth and profitability has continued and that valuations in the SaaS market over the past year have converged toward their fundamentally-predicted values.
The relative value relationships between companies in the SaaS universe yield the following SANE formula at present:
While we like the “rule of thumb” nature of the 2-factor model, the 5-factor model is more useful for quantitative analysis of SaaS market prices, and experienced a comparable increase in explanatory power. The share of the 5-factor equation driven by profitability remained roughly constant while the share driven by growth rose from 43% to 60%. This increase reflects the same strengthening statistical relationship observed in the 2-factor model, but the magnitude of the difference in the change between both variables is much more clear. While profitability continues to gain in importance, for the SaaS universe is it top-line growth, both achieved and forecast, that truly drives valuation.
Nevertheless, a conclusion that Mr. Market’s priorities have again shifted to growth over profitability would be mistaken. Corroborating our cross-sectional results over the past year with a historical panel regression detailed in the graphs below, we observe a continuation of the trends from 2015, including the continued rise in importance of profitability. Revenue growth has retained primary importance as a determinant of value, with trailing growth increasing from a 0.63 to 0.72 correlation and forward growth diminishing, from 0.77 to 0.67. The two remaining variables held constant, with market cap remaining unchanged and gross margin rising modestly from 0.28 to 0.30. With Q1 2017 financial reporting still inchoate, part of the diminished coefficient on forward revenue growth is likely due to a smaller comp set.
By far, the most important trend in the data has been the continued increase in the importance of EBITDA Margin to valuation, rising from a modest negative correlation in 2015 (-0.20) to slight positive at present (0.10). Not to understate the point, 2016 and 2017 have seen the first reemergence of a positive correlation between valuation and EBITDA Margin in the SaaS market since 2010. While the impact of the trend is muted in the first graph, showing a smoothed moving average, the green line in the second graph below illustrates the recent rise in EBITDA margin’s correlation to valuation – we expect a continuation of this trend in coming quarters as a natural evolution related to the maturing of the SaaS market.
3) SaaS Market Update
2016 was a year of tremendous oscillation in the equity markets, even if they finished the year with returns near the long run average. The NASDAQ composite’s 7.5% annual return trailed the broader-market S&P 500’s 9.5%, and fared more poorly on a risk adjusted basis, with a maximum drawdown of 15% compared to the S&P 500’s 10% in early 2016. Software company valuations largely mirrored public market price changes, but even more dramatically.
Following disappointing guidance from Tableau (DATA) and LinkedIn ([LNKD]), the cloud software sector slumped precipitously, in tandem with generally weak earnings across the major averages. In Q1 2016, the SaaS market was valued at a median 4.0x revenue multiple – a level last seen in the “double dip” pullback of Q3 2011 and substantially below the average of 5.0x since then. By the time we’d published our last results, in Q2 2016, the valuations in the market were already snapping back to equilibrium: increasing to a median of 4.4x in Q2, 5.1x in Q3, 4.3x in Q4, and 4.9x at present (Q1 2017). The same public market volatility that led to a paucity of IPOs in technology led presented a robust landscape for both strategic and financial M&A.
In developing the SANE approach, our goal was to benchmark valuation by focusing on its foundation, suggesting how companies might converge normally over time, or rapidly in times of market rationalization. Focusing on the fifteen companies that were acquired or taken-private between our last publication and the present, we can see this convergence in action. On average, the SANE model was within 35% of takeout value as of 12/31/15, and was able to predict takeouts within a remarkable 10% accuracy if the model were iterated in the same quarter. These companies are listed in the table below, which shows valuation multiples (both the value predicted by SANE and the actual market multiple) as of 12/31/15 and at the time of takeout. For each period, we also calculate the relative over or under-valuation of the actual company multiple with regard to that generated by SANE – higher / lower predicted than actual value would suggest a company is undervalued / overvalued by the market. We then calculate the change in stock price for each company, showing the gain or loss to a long equity position taken on 12/31/15. Finally, in the commentary column, we briefly discuss whether the SANE model or the market priced the company more accurately and how the strength of the SANE valuation estimate improved as the company moved toward exit. The companies are listed below, from most undervalued to most overvalued.
Exhibit 5: Valuation and Performance of Public SaaS M&A Targets (2015-2016)
Buying the cheaper half of the fifteen companies, as determined by the relative difference between predicted and actual multiple as of 12/31/15, would have returned 38% if held through takeout, compared to an 18% return for the richer half, 27% for all fifteen acquired companies, and 16% for the entire comp set.Source: Capital IQ, Catalyst analysis
Five of fifteen acquired companies (1/3rd) went to financial buyers and the remainder to strategics. Conventional wisdom suggests that financial (private equity) buyers are more attracted to value plays, while strategic acquirers are better-positioned to realize the synergies of faster-growing and higher-valued companies. At first glance, this segmentation appears to be borne out in the historical data below, taken from a different historical comp set, which shows a purple cluster of financial acquisitions of low-growth, low- valuation companies.
However, a low valuation in absolute terms doesn’t necessarily make a company a value investment. The above graph plots the relationship between valuation and growth, only incorporating one leg of the SANE stool. Since we’ve found SANE to be a more robust approach to benchmarking company value than a focus on revenue growth alone or the “Rule of 40,” it could be reasonably assumed that financial buyers would focus on companies more undervalued by SANE – those inappropriately punished by the market relative to their underlying financial metrics. Surprisingly, the reverse holds in our sample set. The five companies with financial exits were 18% undervalued both as of 12/31/15 and at exit (an interesting coincidence we’ll return to momentarily), while the ten companies with strategic exists were 43% undervalued as of 12/31/15 and 4% undervalued at exit. For comparison, the group of fifteen acquired companies were 35% undervalued as of 12/31/15 and 9% undervalued at exit, and the comp set as a whole was fairly-valued as of 12/31/15 and slightly (4%) overvalued at exit / present.
In contrast to conventional wisdom, this suggests that financial acquirers target higher-valued companies and pay multiples in line with or at a slight premium to public trading, while strategic acquirers target the most undervalued companies and pay prices that are above-market but in line with the fundamental value drivers illuminated by SANE. The SANE approach enables investors to better predict companies with the potential for outsized performance, particularly that accruing from strategic acquisition. We can offer further insight through exploring the outlier groups described in our previous study: how both investment performance and the convergence between fundamentals and valuations differed between these cohorts.
Market Darlings are companies with actual enterprise value to LTM revenue multiples above 5x and that are valued more than 20% above their predicted multiple based on metrics alone – in other words, high valuations on both an absolute basis and relative to their metrics. The cohort as of 12/31/15 included Xero (XRO), ServiceNow (NOW), Workday (WDAY), Splunk (SPLK), Hubspot (HUBS), Netsuite (N), Ultimate Software (ULTI), athenahealth (ATHN), and Benefitfocus (BNFT). Of the nine companies, one (Netsuite) was acquired, two (Ultimate Software and Workday) remain market darlings, five are now fairly-valued (athenahealth, Benefitfocus, ServiceNow, Splunk, and Xero), and one (HubSpot) has shifted from being overvalued by 28% to being undervalued by 18% (practically making it now a “Prove-It”) on continuing improvement in fundamentals. This Market Darling cohort went from being 23% overvalued in 2015 to being only 2% overvalued (within the range of fairly valued) at present, with the group’s average spread between predicted and actual valuation widening 21%. Compared to an average 20% equity return for the entire comp set, an investment in the nine 2015 Market Darlings would have returned -5%. The current Market Darling cohort includes seven companies (Veeva, Workday, Proofpoint, LogMeIn, Ultimate Software, Ebix, and Demandware). The current Market Darlings trade at an average 8.5x multiple (vs. 5.8x suggested by SANE, or 30% overvalued on average), compared to the overall comp set (including takeouts) which is valued at an average of 5.4x, in line with SANE.
Prove-Its are companies that have strong enough metrics to be valued above 5x revenue, but trade at a value more than 20% below their predicted value. The cohort as of 12/31/15 included Hortonworks (HDP), AppFolio (APPF), Textura (TXTR), Box (BOX), MINDBODY (MB), Broadsoft (BSFT), Xactly (XTLY), and Apigee (APIC). Of the eight companies, two (Apigee and Textura) were acquired, one (Broadsoft) remains a prove-it, four (Box, AppFolio, MINDBODY, and Hortonworks) fell out of the prove-it cohort into the fairly-priced category, and one additional company (Xactly), fell from the Prove-It cohort into the “Underperformers,” primarily driven by a decline in EBITDA margin. This Prove-It cohort went from being 60% undervalued in 2015 to 12% undervalued (roughly within the range of fairly-valued) at present, with the group’s average spread between predicted and actual valuation tightening 48%. Interestingly, both the 2015 cohorts of Market Darlings and Prove-Its experienced substantial convergence toward the fundamental-based valuation range suggested by SANE. Compared to an average 16% equity return for the entire comp set, an investment in the remaining 8 Prove-Its would have returned 39%. Apigee, the most undervalued company in the 2015 cohort according to SANE (183% undervalued) returned 116% from that time through acquisition. The current Prove- It cohort includes seven companies (BroadSoft, HubSpot, RingCentral, SPS Commerce, Salesforce, 8×8, and RealPage) currently trading and one recently acquired (Diligent Corporation). HubSpot (as discussed earlier) was previously a Market Darling and only BroadSoft remained a Prove-It. The current Prove-Its trade at an average 5.2x multiple (vs. 7.0x suggested by SANE, or 29% undervalued on average), compared to the overall comp set which is valued at an average of 5.4x, in line with SANE. The substantial turnover in Prove-It companies between the 2015 and current cohorts is remarkable, and highlights the particularly strong convergence traits of these companies (undervalued on both absolute and relative bases) – and suggests a focal point of future research on how to leverage the SANE approach for investment decision-making.
Underperformers are companies valued less than 5x that trade more than 20% below their predicted value, which is also less than 5x. The cohort as of 12/31/15 included Healthstream (HSTM), inContact (SAAS), Intralinks (IL), Halogen (HGN), Liveperson (LPSN), and Jive Software (JIVE). Of the six companies, two were acquired (inContact and Intralinks). The remaining companies were then valued at 1.7x, compared to a predicted 2.8x (67% undervalued on average); at present, they are valued at 2.2x compared to a predicted 2.6x (21% undervalued on average, or 47% undervalued excluding LivePerson). The average spread between predicted and actual valuation tightened 17% from 12/31/15 to the present. Of these remaining companies, all but one (Liveperson) remain underperformers. Liveperson is now fairly-valued after its fundamentals deteriorated such that its valuation is now in-line with market norms. Compared to an average 16% equity return for the entire comp set, an investment in the remaining 4 underperformers would have returned 23%, though the return is skewed by Halogen’s 74% return – excluding Halogen, the three remaining underperformers returned a modest average of 6%. Remarkably, the current Underperformer cohort includes thirteen companies (athenahealth, Benefitfocus, Castlight Health, ChannelAdvisor, Halogen Software, HealthStream, Jive Software, Quality Systems, Tableua Software, Tangoe, Upland Software, Workiva, and Xactly), a substantial increase from the 2015 cohort’s six companies, three of which (HealthStream, Halogen Software, and Jive Software) remain. The current Underperformers trade at an average 2.7x multiple (vs. 3.9x suggested by SANE, or 56% undervalued on average), compared to the overall comp set which is valued at an average of 5.4x, in line with SANE.
We’ve shown how a nuanced understanding of the drivers of valuation, boiled down to a handy tool for benchmarking company valuations in the SANE approach, can help investors select undervalued companies poised for growth and avoid those with fundamentals struggling to keep pace with market valuations. We conclude this update to our analysis by moving from cohorts to the aggregate comp set, which illuminates the pattern of convergence between actual and predicted multiples over the course of 2016. Based on each company’s actual valuation relative that predicted by SANE as of 12/31/15, an investment in the 19 companies more than 20% undervalued returned an average of 38%, compared to the 12 companies more than 20% overvalued which returned 3%, and the aggregate comp set which returned an average of 20%. Of the 19 undervalued companies, only one (Tangoe) had substantially negative equity performance, and two (Shopify and Apigee) more than doubled in value; of the 12 overvalued companies, five (athenahealth, Ultimate Software, ServiceNow, Marin, and Guidewire) had negative equity performance. Turning from company performance to relative over or under-valuation compared to peers, we focus on the percentage spread between actual company valuation and that predicted by SANE. For companies previously more than 20% overvalued (which averaged 35% overvalued), this spread tightened by 10%, or by 22% excluding the only widening outlier (Marin). These companies saw an average decline in multiple of 24%, from an 8.3x average to 6.8x at present. By contrast, the average spread of the companies previously more than 20% undervalued (which averaged 64% undervalued) tightened by 30%, reflecting the cohort that was undervalued relative to SANE converging toward the fundamental-based market norm. These companies averaged a 25% increase in multiple, from 3.6x to 4.5x. Of the fairly valued companies last time, the spread widened by 4%, parallel to the increase in relative valuation of the comp set as a whole. The median multiple for the entire comp set decreased 4% (from 5.2x to 5.0x) and the average was down a comparable 2% (from 5.5x to 5.4x). Reflecting convergence toward the mean, of companies valued more than their predicted value last time, the average multiple declined by 17% (from 6.8x to 5.8x) and of the companies valued less than their predicted value, the average multiple increased 18% (from 4.4x to 5.2x).
The current list of companies more than 20% overvalued (ignoring takeouts and those with negative predicted values, listed in order of ascending current overvaluation) are: Marin Software, LivePerson, FireEye, Guidewire, Veeva, TrueCar, Amber Road, and Workday. Many of these are expensive companies that either had a decent decline or increase in actual multiple from last time. The current list of companies more than 20% undervalued (ignoring takeouts, listed in order of ascending current valuation) are: Tangoe, Jive Software, ChannelAdvisor, BroadSoft, Castlight Health, Tableua, Xactly, Quality Systems, RingCentral, Halogen Software, SPS Commerce, Benefitfocus, Workiva, athenahealth, HealthStream, salesforce.com, 8×8, and RealPage.
Our data set includes 55 currently publicly-traded SaaS companies. We exclude Twilio (TWLO) from this analysis due to its recent IPO. Earlier dates in the data set include up to a total of 89 companies, reflecting those that were then publicly-traded but have since been acquired or otherwise removed. The inclusion of these companies is a change from the data set employed previously, intended to mitigate a degree of survivorship bias in the data. As previously, the current set of companies also includes some that have transitioned from a more traditional on-premise software model. The selected companies are listed alphabetically below:
Relative to our analysis as of 12/31/15, the additional companies included in this study are listed alphabetically below:
- Ariba (formerly ARBA) – procurement and sourcing software – founded 1996, IPO 1999, acquired by SAP in 2012, included as historical comp
- Atlassian (TEAM) – collaboration software – founded 2002, IPO 2015, included due to post-IPO price stability
- BlackBoard (formerly BBBB) – educational software – founded 1997, IPO 2004, acquired by Providence Equity Partners in 2011, included as historical comp
- ChannelAdvisor (ECOM) – eCommerce software solutions – founded 1996, IPO 1999, included as new comp
- Concur (formerly CNQR) – cloud software – founded 1993, IPO 1998, acquired by SAP in 2014, included as historical comp
- Constant Contact (formerly CTCT) – online marketing services – founded 1995, IPO 2007, acquired by Endurance International in 2015
- Convio – non-profit marketing software – founded 1999, IPO 2010, acquired by Blackbaud in 2012, included as historical comp
- Covisint – cloud business software – founded 2000, acquired by Computware in 2004, IPO 2013, spun-off in 2014, included as new comp
- Demandtec – cloud-based retail merchandising – founded 1999, IPO 2007, acquired by IBM in 2011, included as historical comp
- Diligent Corp – board collaboration software – founded 1994, IPO 2009, acquired by IVP in 2016, included as historical comp
- E2open – supply chain management software – founded 2000, IPO 2012, acquired by IVP in 2015, included as historical comp
- Eloqua – marketing automation – founded 1999, IPO 2012, acquired by Oracle in 2012, included as historical comp
- Epiq Systems – legal software – founded 1964, IPO 1997, acquired by OMERS Private Equity and Harvest Partners in 2016, included as historical comp
- ExactTarget – marketing software – founded 2000, IPO 2012, acquired by Salesforce (CRM) in 2013, included as historical comp
- Guidewire – insurance software – founded 2001, IPO 2012, included as new comp
- Kenexa – human capital management software – founded 1987, IPO 2005, acquired by IBM in 2012, included as historical com
- Omniture – marketing and web analytics software – founded 1996, IPO 2006, acquired by Adobe in 2009, included as historical comp
- Proofpoint – email security software – founded 2002, IPO 2012, included as new comp
- Responsys – marketing automation software – founded 1998, IPO 2011, acquired by Oracle in 2013, included as historical comp
- RightNow – customer service software – founded 1997, IPO 2004, acquired by Oracle in 2011
- Saba Software – human capital management software – founded 1997, IPO 2000, acquired by Vector Capital in 2015, included as historical comp
- com – human capital management software – founded 1987, IPO 2010, acquired by Kenexa in 2010, included as historical comp
- Success Factors – human capital management software – founded 2001, IPO 2007, acquired by SAP in 2011, included as historical comp
- Taleo – human capital management software – founded 1996, IPO 2005, acquired by Oracle in 2012, included as historical comp
- TrueCar – automotive pricing and information software – founded 2005, IPO 2014, included as new comp
- Twilio – cloud communications platform as a service – founded 2008, IPO 2016, excluded from analysis due to post-IPO pricing volatility
- Vocus – public relations software – founded 1992, IPO 2005, acquired by GTCR in 2014
Our current five-factor and two-factor regression outputs are shown below, respectively