Video of My SaaStr 2020 Presentation: Churn is Dead, Long Live Net Dollar Retention

Thanks to everyone who attended my SaaStr 2020 presentation and thanks to those who provided me with great feedback and questions on the content of the session.  The slides from the presentation are available here.  The purpose of this post is to share the video of the session, courtesy of the folks at SaaStr.  Enjoy!


17 responses to “Video of My SaaStr 2020 Presentation: Churn is Dead, Long Live Net Dollar Retention

  1. Thanks for sharing Dave! Going through the video and I think I will have to replay it a few times as it is information dense as usual :)

    Quick question on a different topic: how do you see Deferred Revenues as part of valuing a SaaS business? are they fully/partially added to Net Working Capital or excluded? Could you consider a series of posts on working capital in a SaaS business from an operational & valuation/exit perspective? Thanks/!

    • Deferred revenue (particularly long-term deferred revenue) represents prepayments on multi-year contracts. In effect, they are more of a financing strategy than a reflection of operations and while I can’t remember the details, I think ASC 606 does treat them as a financing event. I’d say investors don’t like them and neither do acquirers because they represent a liability — work that the company / service that the company needs to deliver but for which it’s already been paid. I know examples of where PE firms wanted to decrease their valuation by the amount of long-term deferred revenues because of this. So net/net, they can be a reasonable financing strategy but if and only if you don’t want to get acquired by either a PE firm or a public SaaS company. Ergo, in general, I’d say no. Do multi-year deals with annual payments and use RPO to track your backlog.

  2. Hi Dave- enjoy your work.
    To calculate RPO it seems like you would sum your Deferred Revenue balance (which is generated by what you’ve invoiced, not necessarily been paid- minor quibble) plus any contracted-but-uninvoiced agreements. What would that miss?

  3. Great presentation, thanks for sharing.

    Quick question on NDR, which I agree is a more consistently calculated metric than churn rate.

    You highlighted 115% is a pretty good NDR target, and that 100% – 105% is pretty blah. The latest 1H20 “private” company benchmarks we see are closer to 103% – 104%, and the latest data I saw from KeyBanc was 102%. I assume 115% NDR benchmark attributed primarily to/from public companies, interesting to see the delta in NDR between public/private companies if that is the primary variable impacting the difference

    The other aspect that I found interesting is the impact of NDR on Enterprise Value. Software Equity Group just published their Net Retention Wave which essentially says in private company acquisitions, an increase in NDR, holding top line growth rate equal, had a material impact on EV/Rev multiples. It is interesting to see the impact of NDR rates on valuation appear to be different on private vs public company valuations.

    Thanks again for sharing the presentation!

    • I think you’re right that I may have been optimistic in setting expectations for all companies based on the public company segment. While you can see I was talking about the median in the Meritech Comps data set, and I’ve not yet seen KeyBanc 2020 (if it’s even released). If you have links to either of your studies, please post them here. I was frankly surprised to see the low NDR correlation in EV multiples. I was kind of assuming that investors were basically saying “just grow, we don’t care how” more than we can where the growth comes from but to the extent expansion ARR has a lower CAC you might see it in CAC as predictor.

      • Net Dollar Retention Rate and private transaction EV/Revenue multiple impact, as provided by Software Equity Group research was part of a LinkedIn post I made in July, which can be seen at:

        Original source data on NDR impact on EV/Rev multiple from SEG is at:

        Regarding KeyBanc, the latest “publicly” available data is from 2019 – Page 53 that highlights a 102.7% Net Dollar Retention Rate (Median) can be found at:

        The 2020 KeyBanc SaaS Metrics research – COVID edition has only been pre-released, and is not yet publicly available. Dave Spitz, KeyBanc MD will be presenting the latest research results publicly tomorrow, September 9th with VentureBeat and Sage Intaact. Registration can be found at:

        Hopefully this information is helpful.

      • It sure is! Thanks for sharing.

      • Net is I now agree I was probably too aggressive on NDRs based on the high-flying public company comp set; I think I might set 102-104% as the bar that you need to be over in this light. Anything below 100% is definitely unacceptable. Above 120% is still epic. 103% is OK, not great. But OK.

  4. Thanks for sharing this, Dave–very helpful. Would you mind commenting on usage-based pricing models? When a similar question was raised in the chat, you said it was hard and that you would return to it if time permitted (which it didn’t). I understand why ARR is the gold standard metric for subscription-based pricing, but I’m struggling to wrap my mind around ARR for a usage-based model. Consider, for example, Stripe Payments (pay-as-you-go percentage of transaction value). There is no pre-defined contract value, so (it would seem to me) ARR breaks down? If so, what other metrics might we substitute?

    • Hi Sean,

      It’s hard (for me) partly because I don’t have enough experience with it and I’m pretty drilled in SaaS when it’s about annual contracts typically for numbers of users using some combination of products. So, frankly, consumption-based pricing is on my to-do list for a deep dive. I’ll try right now with some quick thoughts. Let’s say I’m on a consumption-based model and I use 100 units in year 1. I suppose we could call it churn if I only use 10 units in year 2, but I won’t know that until the end of year 2. With annual SaaS contracts, I’d know that at the start of year 1. So churn appears to be a lagging indicator, and ergo one I should look at more frequently than annually. So I could look at monthly “churn” and report that (the NDR for last month’s cohort of customers) but I should probably annualize it to be comparable to SaaS companies. The real thing that scares me is seasonality — e.g., my last company sold financial planning systems that got used a ton during “budgeting season” but only intermittently in the off season. So, if I’m usage-based, I’m going to look terribly in February through August, then God-like in August through September. You could, I suppose seasonalize the rates but they we’re doing so much math, with so little history/data, that I’m nervous. If you wanted to comparable to SaaS companies, I’d say you’d need to look at NDR the same way SaaS companies do (another advantage, I realize, of NDR over churn) which is to look at year-ago cohort analysis, but again it’s one-year lagged because we need to be at the end of year 2 before we can compare year 2 usage to year 1 usage. Now, I’m assuming these companies are not purely usage-based and they may have some minimums so you might compare forward-looking minimums to last year’s actuals. Or forward-looking forecasts (#danger) to last year’s actuals. But in reality if you want to look at actual performance you need to know the actual usage and the actual price paid (presumably on some declining curve). This brings up the idea of usage-churn vs. dollar-churn (similar to user-churn vs. ARR-churn) as something else to look at. This stream of consciousness tends to confirm my default nature to be very careful in using a system designed for one thing (annual contracts) to analyze another. Interestingly, NDR seems to work either way. RPO would work far less well, btw, as I think the whole point of usage-based pricing is to be on usage and not on future committed contracts. I’ll take a look at Snowflake and Twilio when I get time and get back to you.

  5. Hi Dave – fantastic and informative video. Point noted on NDR survivorship basis done by some public companies (choosing to start with this period and look back instead of isolating prior period’s cohort). I am familiarizing myself with SaaS companies and would love your thoughts on this public disclosure as a learning exercise. Is Fastly’s DBNER the standard way to calculate NDR?:

    “We calculate Dollar Based Net Expansion Rate by dividing the revenue for a given period from customers who remained customers as of the last day of the given period (the “current” period) by the revenue from the same customers for the same period measured one year prior (the “base” period). The revenue included in the current period excludes revenue from (i) customers that churned after the end of the base period and (ii) new customers that entered into a customer agreement after the end of the base period”

    Fastly and Cloudflare have very different DBNERs, and I wanted to curious whether it was due to performance or definition.

    • Sounds dodgy on a quick skim. Seems survivor-biased. Seems to include people not in the denominator. Uses revenue, not ARR (are they subscription model?)

      • Hi Dave – Fastly is usage-based. For general platform customers, billed monthly, but little point to note is there is also $50/mo minimum in spend. For enterprise customers, also billed monthly but depending on product, there might be one-time fees. That said, generally yes, measured in gigabytes and requests (i.e. usage) primarily. This is one of the clear differences with its competitor Cloudflare which charges $20/mo for Pro and $200/mo Business subscriptions. Enterprise contracts have to be quoted for Cloudflare but it is still a mix of subscription and usage. Generally more predictable for Cloudflare

  6. Pingback: Kellblog 2021 Predictions | Kellblog

  7. Pingback: Pulse 2021 Slides: Net Dollar Retention (NDR) Benchmarks and Thoughts | Kellblog

  8. Pingback: Come to my SaaStr Annual Presentation: The Strategic Use and Abuse of SaaS Metrics – Kellblog

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.