5 other ways to think about Net Revenue Retention (NRR)
As the frontier of knowledge about subscription companies expands, net revenue retention (NRR) has become the predominant metric in assessing the viability & value of companies in the category. To state the obvious: higher is better. Higher NRR means greater expansion or less churn (likely both) which means a stickier user base that compounds revenue quicker and more cheaply.
But “more = better” is just the beginning of the analysis. There are a lot of other things that a company’s NRR data can teach us. Here are some concepts from the diligence we’ve performed at Toba on many SaaS deals over many years.
1) How does net retention compare to gross retention?
I wrote about this at greater length here. Net retention measures the total impact on recurring ARR from churn, expansion, and downgrades, whereas gross retention (and its cousin, logo retention) ignores expansion and merely focuses on how many customers decide to stick around after a year. This is important because we want to understand how “sticky” a solution is, and getting customers to “stick around” is a prerequisite for eventually upselling them.
Looking at net retention and gross retention together tells a more complete picture about a business. High net retention with low gross retention means the company is doing a phenomenal job of expanding happy clients but is wasting resources on a whole other cohort of accounts that don’t retain at all. The opposite, low net retention with high gross retention, means that clients are sticking but you still haven’t developed an upsell story yet. Depending on where a company is in its lifecycle, you might actually prefer the latter scenario to the former — it’s often easier to sell additional products to a captive user base than it is to fix a gross churn issue even though many customers are happily upgrading.
Some of our most successful portfolio companies consistently showed gross retention rates in the 90’s but only developed additional products that pushed NRR over 120% and beyond once they were well beyond $10m in ARR. This can be a very strong foundation to build out from.
2) How does net retention compare to overall growth in revenues and logos?
An interesting case is the companies that have extremely strong NRR but may not be showing the overall growth rate that you’d expect with such a strong net expansion tailwind.
I often think about one company in particular (which I won’t name, obviously) that became a buzzword within Toba for this particular set of attributes. This company had best in class NRR, something like 155% (with terrific gross retention, well into the 90’s), but it was only growing ARR by about ~60% annually. Once we dove into the numbers the problem became incredibly clear — the company was doing a fabulous job of retaining and expanding their existing clientele, but they had almost no success at closing new business. While the stickiness and the NRR made for a very efficient company (I recall they were generating positive cash flows while growing at the aforementioned 60% rate, so definitely well above the Rule of 40), as a venture investor you’re betting that not only will a business be profitable but it will also be huge, and it’s tough to invest in a company that isn’t showing signs of significant accumulation of new customers.
Sometimes a very high NRR paired with a medium-ish growth rate can be just fine. One example I really like is UIPath. UIPath reports (in their Q1FY2023 earnings slides) a 50% ARR growth rate alongside a 138% net revenue retention rate. Does that mean that the rate of new logo acquisition is anemic (“they’d only be growing at 12% if not for clients expanding”)? In this case, the answer is “no”, and looking closely at the customer counts by segment tells the best story:
Overall customers grew by 22%, customers over $100k ARR grew by 42%, customers over $1M ARR grew by 62% (and remember, NRR is 138%). What does this tell us? It tells us that UIPath is doing a terrific job of expansion, but also they are adding lots of small customers each quarter that eventually expand into larger customers. This shows up in the total customer count even though the impact on ARR growth is minimal (for now). It’s OK if the impact of new customers on the overall ARR growth rate is relatively small (compared to expansion) as long as those new customers can reliably expand next year and the year after.
The holy trifecta is a company with beautiful NRR, very high gross retention, and an efficient and scalable sales motion for acquiring lots of net new logos (whether large or small).
3) Is the company gaming the NRR metric?
Once the investing public figured out that NRR was a better predictor of future performance (and therefore a determinant of valuation) than growth rate alone, it was only a matter of time before clever investors and managers and bankers began to game it. To accomplish this, all you really have to do is find a way to artificially depress your customers’ first year revenues or ARR. Some ways to do this are:
Offering low first year or pilot pricing that normalizes in year 2 or beyond
In a business that has a transaction or payments based revenue component, starting the “clock” artificially early before the variable revenue is realized (e.g. booking a client in January that didn’t start processing payments until May, but treating January-December as the “year 1” period for calculating NRR, rather than May-April).
Excluding churn from the “net expansion” calculation by looking backwards (of our 2022 paying clients, how much were they paying in 2021?) rather than forwards (of our 2021 paying clients, how much are they paying in 2022?). Bad, Twilio, bad!
Showing a very high NRR because the company’s largest customer or channel partner expanded in scope dramatically, in a one-off deal that has little or nothing to do with the expected expansion behavior of the rest of the clientele.
Not all of these are necessarily deceitful or unethical things to do, in fact some (say, offering cheap introductory first year pricing) might be smart business sense. But anyone analyzing these businesses should have a good grasp of what’s actually driving the net expansion from year 1 to year 2 — discounted first-year pricing that eventually normalizes in line with the customers’ expectations when they signed the contract is a dramatically different scenario from having clients triple their commitments because they were unexpectedly blown away by the value delivered.
4) Is churn (and NRR) inclusive of cancelations? How broad is the definition of cancelations?
In general, churn is when an account that is fully implemented and paying for the platform decides to leave. Cancelations are when a booked account never went live and changes their mind before they get onboarded and start paying. Breaking these out into separate categories is helpful, because cancelations are less costly to the business (they don’t have to expend resources to fully onboard the client, and often they claw back commission from the sales rep), though in both cases the business has to eat the CAC that that ended up being unproductive.
Problems can start when cancelations start to rise beyond what would be typical, for instance because there’s actually a serious onboarding problem that is not being reflected in churn or NRR. A good example of this was Q2 of 2020 for many companies — if they had many “booked but not yet live” clients in their implementation backlog, a high percentage of those ended up canceling during the uncertainty of the Covid-19 lockdowns (particularly in SMB), but this would not show up in the net retention calculations. “Typical” cancelation rates can vary but something in the 5-7% range is what I see most frequently.
Another issue can be an overly broad definition of cancelations. One company I know well defined cancelations as any customers who did not make it to 6 months on the platform (customer contracts were for 12 months, so this was not as unreasonable as it sounds — almost every client who didn’t make it to 6 months never got onboarded in the first place). The problem wasn’t that this was sneaky, but that it was unexpected — new prospective investors would take a look at the company, calculate churn, and then be surprised that there was a significant cohort of canceled customers on top of the churned customers they had already counted, that had to be included back into the calculation. This put a bad taste in many investors’ mouths, which impacted the company’s ability to fundraise.
5) How much of the churn was avoidable vs. structural?
In many businesses, particularly those serving SMBs, we have the concept of “structural churn” — this is the amount of churn that can be expected to happen merely due to the characteristics of the market, regardless of whether the vendor has been doing a good job of satisfying their customers.
A good example is the real estate agent industry — something like 1/4 of all American real estate agents leave the field every year, and so any software business serving a large cross-section of agents is going to deal with tons and tons of churn that’s not really their “fault”. This is one reason that technologies supporting the real estate industry have been a graveyard of startup failures.
But just because this structural churn isn’t the vendor’s “fault”, it doesn’t mean that it isn’t their problem to deal with. This is where segmentation becomes very important — is there a subset of your industry you can target that has far lower structural churn? Can you target that segment exclusively? Our portfolio company Luxury Presence is a great example in real estate — they have very modest churn, and the reason is that they target the top 20-25% of real estate agents by sales volume, which have a much higher chance of staying in business from year to year (and of course, more willingness to spend to grow their businesses). Among SMB SaaS companies, we’ve found that serving the premium tier of your market is a great way to sidestep such structural churn.
Isolating structural vs. avoidable churn is also important for ascertaining how much a company is delighting customers. Another one of our portfolio companies, Small Door Veterinary has very good customer retention, but among the modest churn that they actually do experience, the vast majority of that churn comes from structural factors — in this case, a pet (sadly) passing away, or the pet parent moving away from New York City (which, for now, is where all of Small Door’s practices are). Once you remove the structural/unavoidable factors, Small Door’s gross customer retention enters the 98-99% range, which might be tops in our entire portfolio. Which is unsurprising, because Small Door also happens to have the highest net promoter score out of any company in our portfolio.
To conclude…
As you can tell, a lot of these analyses are less about separating “good” from “bad” and more about driving to a more precise narrative about a business. Knowing a company’s NRR rate alone does not tell you a whole lot. But marrying an understanding of the NRR with the gross retention rate, the overall company growth rate, the pace of logo acquisition, the NPS, and the rate of structural churn in the industry … this lenses allow you to develop a deeper understanding of how a company functions and delivers value.
If you have any other methods that you use to think through the nuances of net retention, please don’t hesitate to leave a comment and share!
Makes sense for pure subscription models. How do you think about defining NRR in the context of consumption models? There are a lot of products that have such a large number of plans, it can make sense to have an intermediate category that is a little "consumption-y" and a little "subscription-y".