Wing.vc published the Enterprise Tech 30 last week. It’s a coaches poll of the top enterprise startups broken into early, mid and growth stage. Congratulations to all the companies and in particular, the 8 Redpoint companies on the list: Mattermost, Cockroach Labs, LaunchDarkly, Tray.io, AppZen, Snowflake, Hashicorp and Stripe.
Coaches polls are fun because they provide a different perspective on the market. I analyzed the data set and added a few columns to it to see if there are any trends.
I categorized each company by their primary buyer. Engineering dominates with 37%. Many is the next category. These are products that sell across departments: Mattermost, Notion, Zapier, etc. Design is next, followed by Sales and Finance.
These are very different buyers than I expected. A few years ago, I analyzed the public market cap of SaaS companies by buyer. HR and Sales dominated, with IT a close third. Engineering barely made the list. I should re-run that analysis. But this data suggests the distribution will be markedly different in a few years.
5 of the 11 companies selling to engineering are open source - about half. Open source remains a viable customer acquisition strategy despite the threat from the cloud vendors; Amazon in particular.
Next, I divided the top 30 companies by their place in the stack. We think of enterprise companies in three layers. Applications are software used by business users. Platform includes products that help developers to build applications: payment gateways and low code platforms. Infrastructure are core comprises core information technology: databases, monitoring, key management.
Application and Infrastructure companies each represent about 40% with Platform filling out the rest. If we were to weight this by market cap, we would see a different result since Stripe’s most recent round values the company at $22B+.
The data does suggest there are fewer companies in the platform layer, though, which is consistent with our experience.
Last, I bucketed the companies into price segments: SMB, Mid-Market and Enterprise, based on my estimates of their ACVs. As I’ve said many times, there are many ways to build a very successful business. And the data shows that. There’s a broad distribution of companies across each price segment.
It’s always interesting to see how other investors view the enterprise market. Thanks to the team at Wing for pulling together this great survey.
In Rethinking Customer Churn Rate & LTV/CAC, Thibaud Clement illuminates a counter-intuitive concept about churn. The faster you increase your growth rate (acceleration rate), the higher the churn rate.
Consider the same startup under two scenarios: one in which the acceleration rate is 50% and one in which the acceleration rate is 0%. In the 50% scenario, churn will be 67% higher. A surprising result.
Why does this happen? Because the odds of churn decrease with time, particularly for products with monthly billing. If a business acquires many customers in one month, a big chunk of the customer base is at the point of highest churn risk.
We should define acceleration rate. 50% acceleration rate means your growth rate increases 50% per month. In month 1, it’s 7%; month 2, 10.5%; month 3, 15.75%. In other words, acceleration rate is the first derivative of MRR growth.
For most companies, this will be a short-lived phenomenon because it’s very difficult to maintain high acceleration rates for very long. Growing from 10% m/m to 15% m/m is quite possible in the sub-$5m ARR range, but much harder at $20m in ARR. At that scale, acceleration rate is typically negative (i.e., growth slows).
There’s a corollary: this means as growth slows, churn decreases, assuming no changes in customer behavior. A greater fraction of the customer base is mature and much more likely to continue to pay.
If your SaaS business is growing really fast, and you’re wondering why churn is growing despite high NPS (net promoter score) or other metrics, this is why.
When I shared the Redpoint SaaS Metrics Template, I wrote about the difficulty I had identifying key engineering metrics. I was grateful for all the responses from leaders at many startups to share their expertise. I’ve updated the template with a few metrics.
Reliability - percent of application requests that load. 1 minus reliability is the percentage downtime. This measures the durability of the application.
Availability - percent of application requests that load within a certain latency. 99% uptime means for 99% of seconds within a month, the application responded to requests within 5 seconds. Each company should define the acceptable latency. This measures the responsiveness of the application.
Incident rate - support case count divided by active users. This is a proxy for product quality. If the support case count declines on a per user basis, this indicates product quality broadly defined has improved.
I think these three metrics provide a sense of engineering efforts for boards and executives to discuss. I wish there a way to get a sense of engineering cadence or productivity (similar to AE quota attainment), but that remains elusive.
When we published the results of the freemium survey earlier this year, we noticed respondents targeting the enterprise observed higher net dollar retention and lower churn than those startups targeting other segments. I wondered if we could observe any other patterns about enterprise businesses, so I produced this analysis of public companies with ACVs (annual contract values) of $100k or greater.
In the series of charts that follow, the red bars indicate the value of the metric during the year of IPO. And the blue dashed line is the median.
Revenue growth at IPO spans quite a wide range from FireEye’s 148% to Financial Engine’s 19%. The median is 70%, which is consistent across all other modern software companies at IPO.
Gross margins are also span the gamut from 79% to 34%. These distributions indicate that there’s no clustering as a result of a higher price point. However, in this case, the $100K gross margins of 60% are about 10% lower than the entire population sitting at 66%. The p value is 0.1 suggesting the difference has a high likelihood of being a material difference.
There are a few potential hypotheses for this. Gross margins are lower when customer support costs are higher and when hosting costs are higher. If enterprise customers demand more support or if they demand more expensive infrastructure (like single tenant databases), they may impose lower gross margins on their vendors.
Examining net income margin (profitability), we find one outlier in Hortonworks. Again, a broad distribution exists. The median NIM at IPO is -31% identical to -29% across the board.
Next, let’s turn to Sales and Marketing as a Percentage of Revenue. The most efficient here is Veeva, spending only 15% of revenue on sales, just incredible. The median is 54%, which again doesn’t deviate from the population as a whole.
Last, let’s examine payback period. We know churn rates are meaningfully lower for enterprise companies. Does the other side of the customer equation also favor them?
The median estimated payback period is 18 months. Veeva clocked 3.7 months and Hortonworks 29 months. The median across all companies is 15.5 months compared to 18.4 months in the enterprise. But the t test suggests no meaningful difference, with a p value of 0.97, indicating the means are identical.
In conclusion, the only meaningful difference between enterprise companies and the rest of the public software population is a slightly lower gross margin - about a 10% decline - relative to others.
Every six months or so, I take a look at how the public markets are valuing next-generation software companies. There’s been quite a bit of volatility over the last five years, and this update is no exception. As of mid-June, the public markets value software companies at all-time highs.
The chart above shows the total enterprise value (TEV)/forward revenue multiple for the basket of public software companies. Just a quick reminder on these metrics. Enterprise value equals market capitalization plus debt minus cash and short-term equivalents. Forward revenue is the sum of the projected revenues over the next 12 months.
The blue line the chart is the median over the period which is approximately 5.7x. The red line shows the median across the stocks in that particular month. In 2014, the median touched 7.7x forward before falling by about 60% to 3.3x two years later. Since then, we’ve seen an incredible bull run that brought the forward multiples to 9.5x. A correction followed to 7.1x at the end of 2018. Next, a resurgence back to 9.6x forward. In short, we are in the priciest valuation environment of the last seven years.
Let’s break this down by stock. Zoom tops the list at 44x forward, followed by ZScaler and Okta at around 25x. Then Veeva and Atlassian at 21x. These are numbers we’ve never seen before. On the other hand, these are also some incredibly efficient businesses growing at incredible rates in very large markets. And investors understand these businesses at a deeper level than they have in the past because companies disclose key metrics like net dollar retention.
So a question arises: is the market indiscriminate in the way values companies? The answer is no. About half of the stocks have witnessed declines in multiples in the last 12 months, some very significantly. On the other hand, others have seen massive expansion, nearly doubling of multiples.
In other words, the variance of valuations has exploded. The chart above shows that the variance of the forward multiple has increased from about 3x to about 7.6x in the last six months. It’s another way of saying that multiples have expanded at the upper end of the spectrum quite significantly. We’ve always seen some software businesses trade at 3x and 5x, but we’ve rarely seen businesses trade at more than 20x.
The most attractive companies have become much, much, much more expensive relative to their peers. The chart above is a correlation chart between the change in forward multiple over the last 12 months and the current forward multiple. If a stock had a big multiple 12 months ago is very likely that the multiple has expanded by 50% or greater since then. The R squared is 0.83, indicating a very strong relationship.
Just a few months ago, we partnered with Mattermost and led the Series A. We believe that open source applications will be an important part of the future of software because of their security, lower costs of customer acquisition and the flexibility they offer customers.
Today, Mattermost is announcing a $50M Series B from Y Combinator Continuity and Battery. I’m thrilled to partner with Ali Rowghani from Y Combinator and Neeraj Agarwal from Battery for the next leg of the journey. They both share the same vision of open source application software changing the enterprise landscape.
Customer after customer connects Mattermost to their most important systems. Often, these systems invisible to the Internet. Security operations teams create air-gapped networks for incident response, operations teams developing workflows to fetch logs from different systems to remediate outages, and IT teams developing powerful workflows to automate inbound support tickets from their users - all on Mattermost.
Mattermost’s momentum is a result of tremendous developer adoption. Developers love the product because they can customize it. They modify the UI to incorporate key development and deployment metrics, embed Mattermost chat into internal applications, and rich communication support for gifs and emojis.
And because it’s written in Go and compiles to single binary, Mattermost installs and upgrades easily. Just copy a single file to your server. And it scales and scales handling tens of thousands of concurrent requests.
The collaboration market is measured in the tens of billions. Enterprise buyers’ increased desire for data control, compliance and security signifies a substantial chunk of that market is addressable by open source. Mattermost’s Series B is another step forward to fulfilling that ambition. Congratulations to the team on another great milestone.
Over the last decade or so, I’ve compiled a metrics sheet to summarise a SaaS business. While no living document like this is ever perfect, this is currently the best board-level summary of the overall health of a business I have found. I’m sharing it so that others may benefit and improve it. If you have suggestions, please email me.
The template is broken into six sections: People, Bookings & Revenue, Cash, Sales, Marketing, Customer Success.
People is the first section because people are a startup’s most important element. This section covers employee satisfaction, headcount, and recruiting metrics. Indicators of challenges include a spike of non-regretted attrition or a decreasing employee satisfaction score.
Bookings and Revenue illuminates the company’s performance in closing new business (bookings) and recurring revenue. These are pretty straightforward.
Cash works through the cash balance, burn and implied number of operational months.
Sales breaks down the new customer acquisition metrics: total AEs, number of ramping account executives, bookings capacity, quota attainment and so on. It’s important to track the number of leads in the conversion rates to sales generated by sales. Oftentimes, companies won’t make explicit the lead generation responsibilities between sales and marketing, but I think that’s really important to diagnosing potential issues in the go to market.
Marketing digs into lead generation and conversion metrics. Also, unit economics for repayment and lead velocity rate. Months to repay is a powerful composite metric that describes the ultimate efficiency of the marketing lead acquisition portfolio.
Customer Success reviews the net dollar and logo retention, plus churns and expansions.
You’ll notice there is no engineering section. I found it’s really difficult to identify consistent metrics of engineering performance. Lines of code, count of P0 bugs, days of delay in releases. So engineering tends to be more qualitative, if so I hope someone can surface a better way.
The template is meant to provide a high-level overview of a business and identify where areas of excellence and areas worthy of deeper investigation. It’s another tool for your workbench.
Next week, on the 27th of June, Redpoint will host Office Hours with Guillaume Cabane. Guillaume is an exceptional marketer. He built the highly successful growth practices at Segment and Drift. He stands out because of his persistence at the cutting edge. I remember when he told me of how he used the Clearbit Reveal API to change the content of conversations in Drift pop-ups to meaningfully improve conversion. He’s always at the vanguard of using technology to drive awareness and demand for SaaS products.
Typically, Office Hours are hour long fireside chats with a speaker. We host them at our office in San Francisco and we gather questions from a small audience of about 30-40 people in a closed door format.
This time, we’re trying a new format. Over the course of two hours, we’ll invite four startups to conduct one-on-one office hours for 30 minutes with Guillaume and I to discuss go-to-market issues and questions facing their startup.
Apply to attend these office hours here. Demand for these events often exceeds our hosting capacity, so we will email confirmations five days before the event. Reminder: these events are all off the record. Please register if you can attend in person.
Yesterday, Salesforce announced it would acquire Tableau for $15.7B. Tableau sells data visualization software and the team has built an incredible business. We analyzed the S-1 in 2014. The company has grown since its public offering to generate about $1.1B in revenue, growing at 29%. Let’s put this acquisition in context.
First, it’s the third business intelligence related acquisition in the past month. Google announced the Looker acquisition last week. SiSense acquired Periscope Data. And now Salesforce is merging with Tableau. This wave of consolidation in the BI world suggests this is a key area of competition amongst the biggest software companies in the world over the next decade. Collectively, these acquirers have spent $18.4B on these three businesses.
Second, the Salesforce/Tableau acquisition is the third largest software acquisition since 2012, second to Microsoft acquiring LinkedIn and IBM purchasing RedHat.
Third, Salesforce paid 11.2x trailing (or last 12 months’ revenue) for Tableau, which is the lower middle of the multiples across these transactions. I’ve included only companies who have disclosed their financials in this data set.
To get a sense of how this price and multiple compares in clearer terms, lets look at multiples vs growth rate. The 2019 mergers are in red. They include Tableau, SendGrid and Hortonworks. Tableau is the middle red dot.
These acquisition multiples are in line with the two previous years of large acquisitions, suggesting that the pricing in the M&A market for the first six months of 2019 hasn’t changed materially. The $23.9B in acquisitions suggests that 2019 will be similarly active as 2018, which saw $66B of acquisitions.
You’ve found product market fit. You’ve hired a team, including some managers. Your initial, small customer base is very happy. You’ve discovered an initial channel of customer acquisition that’s working. You’ve raised a meaningful round of capital. And then, right then, product innovation decelerates to zero.
The fast pace that characterized the past 12-18 months, when you would germinate an idea and write the code in less than a few days, has evaporated. Suddenly, the product and engineering teams are bogged down. Every innovation requires a Herculean effort to achieve.
Why? Why does this fact pattern evolve in many software companies? Here are the most common reasons I’ve seen.
First, technical debt. The freewheeling, hedonistic days of idea to instantiation in an instant are over. They’ve left you with the hangover of technical debt.
Architectural issues arise that the team didn’t anticipate when you were building features for a single customer or isolated use case. Now, there’s a growing customer base and more complex integrations. It’s time to shore up the initial infrastructure with production ready code. This happens at nearly every company, even Google. Few people are as motivated to refactor code as write new features. Combine technical debt with demoralized people and you get molasses every time.
Second, the founder/CEO’s once singular focus on product is no longer possible. Rewind a year ago, when 90% of their time was spent on product: finding product market fit, understanding customer needs, translating that into a vision and mocks to be coded.
Today, the demands on the CEO’s time have exploded manifold. Fundraising, recruiting, press, hiring, managing, board meetings; the task count and diversity has exploded. Instead of focusing 90% on product, they may have 15% or 20%. Without someone driving the product roadmap, product innovation decelerates.
Third, inertia. As your customer base grows, the product can’t move as quickly as you’d like because each iteration requires existing customer education. Even minor UI tweaks spike inbound customer support queries, which cost real money.
Fourth, testing. There’s a colloquialism for a collection of testing software within the quality assurance world: harness. The idea is to harness the furious efforts of the thoroughbred engineering team into a smooth release process that ensures few errors for customers in production. At this stage, the mot juste isn’t a harness, but a yoke, a heavy wooden cross beam braced across the shoulders of oxen.
Because just as the engineering team has accrued technical debt, so has quality assurance. Few companies have invested the effort during the crusade for product market fit to develop a robust testing suite. But as the business scales, testing becomes another key part of scaling infrastructure that requires significant investment, without any user-facing advances.
Most of these issues cannot be countered. Developing tests before product market fit isn’t worthwhile. You might anticipate inertia, but you won’t be able to meaningfully change customer behavior; we all habituate to software as we learn it.
But startups can focus on these areas when they arise by prioritizing great quality assurance, creating transition plans for existing customers, and finding a way for someone in the organization to run product nearly full time again; either by hiring someone and delegating that responsibility, or by delegating the other parts of the CEO job and focusing on product.