One thing that amazes me is when I hear people talk about how they analyze churn in a cloud, software as a service (SaaS), or other recurring revenue business.
You hear things like:
“17% of our churn comes from emerging small business (ESB) segment, which is normal because small businesses are inherently unstable.”
“22% of our churn comes from companies in the $1B+ revenue range, indicating that we may have a problem meeting enterprise needs.”
“40% of the customers in the residential mortgage business churned, indicating there is something wrong our product for that vertical.”
There are three fallacies at work here.
The first is assumed causes. If you that 17% of your churn comes from the ESB segment, you know one and only one thing: that 17% of your churn comes from the ESB segment. Asserting small business stability as the cause is pure speculation. Maybe they did go out of business or get bought. Or maybe they didn’t like your product. Or maybe they did like your product, but decided it was overkill for their needs. If you want to how much of your churn came from a given segment, ask a finance person. If you want to know why a customer churned, ask them. Companies with relatively small customer bases can do it via a phone. Customers with big bases can use an online survey. It’s not hard. Use metrics to figure out where your churn comes from. Use surveys to figure out why.
The second is not looking at propensities and the broader customer base. If I said that 22% of your annual recurring revenue (ARR) comes from $1B+ companies, then you shouldn’t be surprised that 22% of your churn comes from them as well. If I said that 50% of your ARR comes from $1B+ companies (and they were your core target market), then you’d be thrilled that only 22% of your churn comes from them. The point isn’t how much of your churn comes from a given segment: it’s how much of your churn comes from a given segment relative to how much of your overall business comes from that segment. Put differently, what is the propensity of someone to churn in one segment versus another.
And you can’t perform that analysis without getting a full data set — of both customers who did churn and customers who didn’t. That’s why I say you can’t analyze churn by analyzing churn. Too many people, when tasked with churn analysis: say, “quick, get me a list of all the customers who churned in the past 6 months and we’ll look for patterns.” At that instant you are doomed. All you can do is decompose churn into buckets, but know nothing of propensities.
For example, if you noticed that in one country that a stunning 99% of churn came from customers with blue eyes, you might be prompted to launch an immediate inquiry into how your product UI somehow fails for blue-eyed customers. Unless, of course, the country was Estonia where 99% of the population has blue eyes, and ergo 99% of your customers do. Bucketing churn buys you nothing without knowing propensities.
The last is correlation vs. causation. Knowing that a large percentage of customers in the residential mortgage segment churned (or even have higher propensity to churn) doesn’t tell you why they are churning. Perhaps your product does lack functionality that is important in that segment. Or perhaps it’s 2008, the real estate crisis is in full bloom, and those customers aren’t buying anything from anybody. The root cause is the mortgage crisis, not your product. Yes, there is a high correlation between customers in that vertical and their churn rate. But the cause isn’t a poor product fit for that vertical, it’s that the vertical itself is imploding.
A better, and more fun, example comes from The Halo Effect, which tells the story that a famous statistician once showed a precise correlation between the increase in the number of Baptist preachers and the increase in arrests for public drunkenness during the 19th Century. Do we assume that one caused the other? No. In fact, the underlying driver was the general increase in the population — with which both were correlated.
So, remember these two things before starting your next churn analysis
If you want to know why someone churned, ask them.
If you want to analyze churn, don’t just look at who churned — compare who churned to who didn’t
Perhaps it was the $168M FY14 operating loss. Maybe it was the $380M in financing raised during the last three years. Or the average quarterly burn rate of $23M. But somehow, I got sucked in.
I just had to know their CAC ratio. Of course, it’s not going to be easy to calculate. While they give us quarterly S&M expense, that’s only half the equation; we’re going to have a figure out –as best we can — quarterly new annual recurring revenue (ARR).
Billings as a Sales Metric
While many SaaS companies don’t disclose “billings,” Box does — but on an annual basis only — in their S-1.
[Click on the images to see full size.]
Billings is an attempt to triangulate on new sales (or bookings) in a SaaS company. The standard way to calculate billings is to add revenue plus change in deferred revenue.
The idea is that if you want to know how “sales” went during a given period, then revenue is not a great indicator because, in a SaaS company, revenue is an indicator of how much you sold in prior periods, not the current one. So you look at deferred revenue trying to pick up the volume of new orders. The problem is that things quickly get very complicated because (1) deferred revenue is moving both down (as past deals convert into revenue) and up (as new deals are signed) and (2) deferred revenue itself is limited only to deals that are prepaid — if a company does a constant business volume but suddenly starts doing a lot of two-year prepaids, then deferred revenue will skyrocket and if, for example, hard economic times drive loyal customers to ask for bi-annual billing, then deferred revenue will plummet, all without any “real” change in underlying subscription business. In addition, multi-year non-prepaid deals are invisible from a deferred revenue perspective (because there’s nothing, i.e., no cash prepayment, to defer).
In short, any metric built upon deferred revenue is only as a good as deferred revenue at reflecting the business.
To demonstrate the relationship between billings and new ARR, I built a model which assumes a SaaS company that starts from scratch, increases new ARR added each quarter by $500K (i.e., $500K in its first quarter, $1M in its second, $1.5M in its third), does only one-year prepaid deals, and has a 90% renewal rate. Here’s what happens.
(You can download the spreadsheet with Box financial summary and the full version of the model here. Be sure to download as an Excel file, not a PDF.)
While in year one, billings is equivalent to new ARR, as you build up the renewals base, it contributes more to revenue and muddies thing up. For a company of the above size, growth, and renewal rate, the ratio of new ARR to billings ends up 0.4.
When you take this same model and (manually) force fit the new ARR numbers to try approximate Box’s revenue and billings from 2012-2014, you get:
A CAC of ~1.6
In this case (and given my assumption set) you end up with a new-ARR/billings ratio of 0.6. To make life easier, I also calculated a new-ARR/revenue ratio (see the full sheet), which ends up around 0.8. I’ll use to this number to calculate my CAC, which comes out to between 1.5 and 1.8. While not quite an idyllic 1.o to 1.2, it’s well below 2.0 and helps explain why Box has been able to raise so much money: their growth has been deemed scalable.
Billings = Ending ARR
In reviewing my models, it’s hard not to notice that billings for a period equals ending ARR for that period. This turns out to be true under my assumption set of subscription-only (no services), one-year deals only, and everything pre-paid. Why? Because for any deal taken at any point during the year, we will recognize some percent of it (X) and the rest (Y) will go to deferred revenue. The difference between X and Y changes across the year but X+Y= the deal size at all times.
This is not true when you have consulting or do multi-year prepaid deals (which can make billings > ending ARR). It’s also not true when you do semi-annual billing (which can make billings < ending ARR).
If you assume for any given company that these factors are roughly constant, then even though uniformly inaccurate, it does provide a simple way to approximate new ARR: take the difference in ending ARR two periods, add a churn assumption, and bang you have new ARR during the period.
Key Metrics, Cashflow, and the P&L
Here are some summarized key metrics (using yellow to highlight points of interest).
Year over year growth, while high at 97% is slowly decelerating.
Gross margins are nice at nearly 80%
Operating expenses are massive: 278% of sales in 1Q12 down to “only” 182% in 4Q14.
S&M expense are a seemingly very high 121% of revenues. This looks bad, but to really know what’s going on we need to examine the CAC, which looks pretty good.
Return on sales is -112%
That burn rate sure grabs you: $22M per quarter
In many ways you see a typical “go big or go home” cloud computing firm, burning boatloads of cash but acquiring customers in a reasonably efficient manner and doing a nice job with retention/cross-sell/up-sell as judged by their retention numbers. When you look big picture, I believe they see themselves in a winner-take-all battle vs. DropBox and in this case, the strategy — while amazingly cash consumptive — does make sense.
Here is a look at cashflow and billings:
And last, but certainly not least, here is the P&L:
I’m always amazed by the R&D spend of seemingly simple consumer services. They spent $46M in R&D last year … on what?
The $171M in S&M expense sure grabs your attention
Since it is the season of predictions, I thought I’d offer up a few of my own for 2014, based on my nearly three decades of experience working in enterprise software with databases, BI tools, and enterprise applications.
See the bottom for my disclaimer, and off we go. Here are my ten predictions for 2014.
Despite various ominous comparisons to 1914 made by The Economist, I think 2014 is going to be a good year for Silicon Valley. I think the tech IPO market will continue to be strong. While some Bubble 2.0 anxiety is understandable, remember that while some valuations today may seem high, that the IPO bar is much higher today (at around $50M TTM revenues) than it was 13 years ago, when you could go public on $0 to $5M in revenues. In addition, remember that most enterprise software companies (and many Internet companies) today rely on subscription revenue models (i.e., SaaS) which are much more reliable than the perpetual license streams of the past. Not all exuberance is irrational.
Cloud computing will continue to explode. IDC predicts that aggregate cloud spending will exceed $100B in 2014 with amazing growth, given the scale, of 25%. Those are big numbers, but think about this: some 15 years after Salesforce.com was founded, its head pin category, sales force automation (SFA), is still only around 40% penetrated by the cloud. ERP is less than 10% in the cloud. EPM is less than 5% in the cloud. As Bill Gates once said about prognostication, “we always overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten.” IT is going to the cloud, inexorably, but change in IT never happens overnight.
Big Data hype will peak. I remember the first time I heard the term “big data” (in about 2008 when I was on the board of Aster Data) and thinking: “wow, that’s good.” Turns out my marketing instincts were spot on. Every company today that actually is — or isn’t — a Big Data play is dressing up as one, which creates a big problem because the term quickly starts to lose meaning. As a result, Big Data today is nearing the peak of Gartner’s hype cycle. As a term it will start to fall off, but real Big Data technologies such as NoSQL databases and predictive analytics will continue to face a bright future.
The market will be unable to supply sufficient Data Science talent. If someone remade The Graduate today, they’d change Mr. McGuire’s line about “plastics” to “data science.” Our ability to amass data and create analytics technology is quickly surpassing our ability to use it. Job postings for data scientists were up 15,000% in 2012 over 2011. Colleges are starting to offer data science degrees (for example, Berkeley and Northwestern). There’s even an a startup, Udacity, specifically targeting the need for data science education. Because of the scarcity of data science talent, the specialization required to correctly use it, and the lack of required scale to build data science teams, data science consultancies like Palantir and Mu Sigma will continue to flourish.
Privacy will remain center stage. Trust in “Don’t Be Evil” Google and Facebook has never been particularly high. Nevertheless, it seems like the average person has historically felt “you can do whatever you want with my personal data if you want to pitch me an advertisement” — but, thanks to Edward Snowden — we now know we can add, “and if the government wants to use that data to stop a terrorist attack, then back off.” It’s an odd asymmetry. These are complex questions, but in a world where the cost of data collection will converge to free, will the privacy violation be in collecting the data or in analyzing it? In a world where one trusted the government to adequately control the querying and access (i.e., where it took a warrant from a non-secret court), I’d argue the query standard might be good enough. Regardless, the debate sparked thus far will continue to burn in 2014 and tech companies will very much remain in the center of it.
Mobile will continue to drive consumer companies like Dropbox and Evernote, but also enterprise companies like Box, Clari, Expensify, and MobileIron. Turns out the enterprise killer app for mobile was less about getting enterprise applications to run on mobile devices and more about device proliferation, uniform access to content, and eventually security and management. (And since I’m primarily an enterprise blogger, I won’t even mention social à la SnapChat or mobile gaming). As one VC recently told me over dinner, “God bless mobile.” Amen in 2014.
Social becomes a feature, not an app. When I first saw Foursquare in 2010, I thought it should be the example in the venture capital dictionary for “feature, not company.” Location-awareness has definitely become a feature and these days I do more check-in’s on Facebook than Foursquare. I felt the same way when I worked at Salesforce.com and we were neck deep in the “social enteprise” vision. When I saw Chatter, I thought “cool, but who needs yet another communications platform.” Then I realized you could follow a lead, a case, or an opportunity and I was hooked. But those are all feature use-cases, not application or company use-cases. Given the pace of Salesforce, they fell in love with, married, and divorced social faster than most vendors could figure out their product strategy. In the end, social should be an important feature of an enterprise application, almost a fabric built across modules. I think that vision ends up getting implemented in 2014. (Particularly if Microsoft ends up putting in David Sacks as its next CEO as some speculate.)
SAP’s HANA strategy actually works. I was one of relatively few people who was absolutely convinced that SAP’s $5.8B purchase of Sybase in 2010 was more about databases than mobile. SAP is clearly crafting a strategy to move both analytics and transactional database processing onto HANA and they have been doggedly consistent about HANA and its importance to the firm going forward. They have been trying for decades to eliminate their dependency on Oracle — e.g., the 1997 Adabas D acquisition from Software AG — and I believe this time they will finally succeed. In addition, they will succeed — quite ironically — with their ingredient-branding strategy around HANA using a database to differentiate an application suite, something that they themselves would have seen as heresy 20 years ago.
Good Data goes public. Cloud-based BI tools have had a tough slog over the years. Some good companies were too early to market and failed (e.g., LucidEra). Birst, another early entrant, certainly hasn’t had an easy time over its ten-year history. Personally, while I was always a fan of cloud-based applications (having become a big Salesforce customer in 2003), I always worried that with cloud-based BI tools, you’d have too much of the nothing-to-analyze problem. Good Data got around that problem early on by adopting a Crystal-like OEM strategy, licensing their tools through SaaS applications vendors. They later evolved to a general cloud-based BI platform and applications strategy. The company was founded in 2007, has raised $75M in VC, is reportedly doing very well, and an IPO seems a likely event in its future. I’m calling 2014.
Adaptive Planning gets acquired by NetSuite. Adaptive Planning was founded in 2003 as a cloud-based planning company and — despite both aspirations and claims to the contrary — in my estimation continues to play the role of the low-priced, cheap-and-cheerful planning solution for small and medium businesses. That market position, combined with an existing, long-term strategic relationship whereby NetSuite resells Adaptive as NetSuite Financial Planning, makes me believe that 2014 will be the year that NetSuite finally pulls the trigger and acquires Adaptive Planning. I think this deal could go down one of two ways. If Adaptive continues to perform as they claim, then a potential S-1 filing could serve as a trigger for NetSuite (much as Crystal Decisions’ S-1 served as a trigger for Business Objects). Or, if Adaptive hits rough road in 2014 for any reason (including the curse of the new headquarters) then that could trigger NetSuite with a value-shopper impulse leading to the same conclusion.
I should end with a bonus prediction (#11) that Host Analytics, our customers, and my colleagues will enjoy a successful 2014, continuing to execute on our cloud strategy to put the E back in EPM — focus and leadership in the enterprise segment of the market — and that we will continue to acquire both high-growth companies who want an EPM solution with which they can scale and liberate enterprises from costly and painful Hyperion implementations and upgrades.
Finally, let me conclude by wishing everyone a Happy New Year and great business success in 2014.
See my FAQ to understand my various allegiances and disclaimers.
Remember I am the CEO of Host Analytics so I have a de facto pro-Host Analytics viewpoint.
Predictions are opinion: I have mine; yours may differ.
Please note that this post has been superseded by A Fresh Look at How to Measure SaaS Churn Rates. I’m leaving it posted to protect in-bound links and because I think it’s still worth reading, but if you want my latest thoughts on how to calculate these rates, see the above post.
I love cloud computing. I love metrics. And I love renewals. So when I went looking on the Web for a great discussion of SaaS renewals and metrics I was surprised not to find much. Certainly, I found the two classics on SaaS metrics:
The Bessemer Venture Partners 10 Laws of Cloud Computing white paper, which I highly recommend despite its increasing pollution with portfolio-company marketing.
The Four Factors
While the above articles are all great, I was surprised that no one really dug into the nitty-gritty of renewals at an enterprise SaaS company, where I believe there are four independent factors at work:
Timing. When a contracted is renewed. For example, how to handle when a contract is renewed early or late.
Duration. The length of the renewed contract. For example, how to handle when a one-year customer renews for three years, and receives a multi-year discount in the process (for either pre-payment or the contractual commitment itself). 
Expansion/shrinkage. The expansion or shrinkage of the contract’s value compared to the original contract. For example, how to handle customers adding or dropping seats or products, and/or price increases or decreases.
The count metric. What do we wish to count (e.g., bookings, ARR, seats, or customers) and what does it mean when we count one thing versus another.
Particularly in a world where companies are increasingly marketing “negative churn” rates and renewal rates well in excess of 100%, I think it’s worth digging into this and offering some rigor.
A Simple Example
Let’s take a concrete example. Imagine a customer who buys 100 seats of product A at $1,200/seat/year on 7/30/12, with a contractual provision that says the price cannot increase by more than 3% per year [1a].
Imagine that customer renews on 6/30/13, buying 80 seats of product A for $1,225, and adding 40 seats of product B at $1,200/seat/year, and who receives a 15% discount for making a prepaid three-year commitment.
Hang on. While I know you want to run away right now, don’t. This is all real-life stuff in a SaaS company. Bear with me, and download the spreadsheet here (as an Excel file, not a PDF) that shows the supporting math.
A few questions are easy:
What were the bookings on the initial order? Answer: $120,000.
What was the annual recurring revenue (ARR) of the initial order? Answer: $120,000.
What were the bookings on the renewal order? Answer: $372,300.
What was the ARR of the renewal order? Answer: $124,100. 
Calculating Churn: Leaky Bucket Analysis
So far, so good. Now let’s talk about churn. Because, as you will see, renewal rates alone are complicated enough, I have adopted a convention where:
When it comes to renewals, I look only at rates
When it comes to churn, I look only at dollars/values
I know this is a completely arbitrary decision, but doing this lets me remember one set of formulas instead of two, reduces rat-hole conversations about definitions, and — most importantly – lets me look at one area in percentages and the other in dollars, helping me to avoid the “percent trap” where you can lose all perspective of absolute scale. 
I define churn with an equation that I call “leaky bucket analysis.” 
Starting ARR + new ARR – churn ARR = ending ARR
So, some questions:
Was there any churn associated with this renewal? Answer: Yes.
Why? Answer: Despite a small price increase on product A, there was a 15% multi-year discount and a loss of 20 seats which more than offset it.
How much new ARR was added? Answer: $40,800. The after-discount value of the product B subscriptions.
What is ending ARR? 124,100 = 120,000 + 40,800 – 36,700.
How many customers churned? Answer: 0.
How many seats churned? Answer: 20.
Note that ARR, seats, and customers are all snapshot (or, point-in-time) metrics that lend themselves to leaky bucket analysis. Period-metrics, like bookings, do not. Bookings happen within a period. There is no concept of starting bookings + new bookings – churn bookings = ending bookings. That’s not how it works. So, when you define churn through leaky bucket analysis, measuring bookings churn doesn’t work.
We can, however, calculate bookings churn as the difference between what was up for renewal and what we renewed. In this case, $120,000 – $372,300 = ($252,300), showing one way to generate a negative churn number. The example makes somewhat more sense in the other direction: if we had a three-year $372,300 contract up for renewal and only renewed $120,000 them we might argue that $252,300 in bookings were churned. From a cash collections perspective, this makes sense .
But from a customer value perspective it does not. Unless the customer has plans to discontinue using the service, by dropping from a three-year to a one-year contract we will actually collect more money from them over the next 3 years if they continue to renew ($438,000 vs. $372,300) . So the bookings churn that looks bad for year-one cash actually results in superior ARR and three-year cash collections.
The lesson here is that different metrics are suited for measuring different things. In this case, we can see that bookings churn is useful primarily for analyzing short-term cash collections and not, say, for customer lifetime value or customer satisfaction.
Renewal Rates and Timing
Now that we’re warmed up let’s have some fun. Let’s answer some questions on renewals:
From a bookings perspective, when should we count the renewal order? Answer: the order was received on 6/30/13 so it’s a 2Q13 booking.
From a renewal rate perspective, when should we count this order? Answer: while debatable, to me it’s a renewal of a 3Q contract, so I would count it in 3Q from a renewal rate perspective. 
When would we count the booking if it were late and arrived on 10/30/13? Answer: From a bookings perspective, it would be a 4Q13 booking. From a renewal rate perspective, it’s the renewal of a 3Q contract, so I would count it in 3Q. 
On a customer-count basis, how do we count this renewal? Answer: 100%. We had one logo before and we have one logo after, so 100%. 
Here it’s going to get a little dicey.
On an ARR basis, how do we measure this renewal? Answer: this begs the question of whether we should include expansion ARR due to new seats, new products, and price increases. Since I am worried that expansion may hide shrinkage, I want to see this both ways. Hence, I will define “gross” to mean including expansion and “net” to mean excluding expansion.
What is the gross ARR-based renewal rate? Answer: 103%. 
What is the net ARR-based renewal rate? Answer: 69%. Now you understand why I want to see it both ways. The net rate is showing that we lost real ARR on product A due to reduced seats and the multi-year discount. The upsell of product B hides shrinkage, producing an innocuous 103% number that might evoke a very different scenario in the mind’s eye (e.g., renewing the original deal for one year with a 3% price hike).
What is the gross bookings-based renewal rate? Answer: 310%. We took a $120,000 order and renewed it at $372,000. (But we transformed it greatly in the process.)
What is the net bookings-based renewal rate? 208%. We took a $120,000 order for product A and turned it into a $249,000 order for product A. But we dropped ARR about 33% in the process (from $120,000 to $83,300) through lost seats and the multi-year discount.
What is the gross seat-count renewal rate? 120%
What is the net seat-count renewal rate? 80%
What is the customer-count renewal rate? 100%
Identifying the Best Renewal-Related Metrics
So, what is the renewal rate then anyway? 69%, 80%, 100%, 103%, 120%, 208%, or 310%?
I’d say the answer depends on what you want to measure. Having nearly drowned you in the renewal-rate swamp, let me now drain it. Here are the metrics that I think matter most:
Leaky bucket analysis is important because ARR growth is the single most important driver of value for a SaaS company.
Churn ARR shows you, viscerally, how much extra you had to sell just to make up for leaks . Rates seem sterile by comparison.
The customer count-based renewal rate is the best indicator of overall customer satisfaction: what percent of your customers want to keep doing business with you, regardless of whether they change their configuration, product mix, seat mix, contract duration, etc.
The gross seat-based based renewal rate shows you how effective you are at driving adoption of your services. Think: land and expand (in terms of seats).
The gross ARR-based renewal rate shows you, overall, how effective you are at increasing your customers’ annual commitment. However, it says nothing about how you do that (i.e., which type of expansion ARR) or the extent to which expansion ARR in one area is offsetting shrinkage in another.
The net ARR-based renewal rate shows you how much of ARR you renew without relying on expansion. This is a very conservative metric designed to unmask problems that can be hidden by expansion ARR.
The gross bookings-based renewal rate is the best predictor of future cashflows. If we know that, on average, we take an order of 100 units and turn it into an order of 175 units – through whatever means – then we should use this metric to predict cashflows. Note that, as we’ve seen, there are trade-offs between ARR and bookings, but the consequences of those can be revealed by other metrics.
Revision 6/25/14, New Definition of Simple Churn, Timing Issues on Gross ARR Renewal Rate
While I generally like and stick with my “show churn in dollars and renewal rates in percents” mentality, I have found that a lot of people still ask about churn as a rate.
To answer, I use one of two different metrics:
“Simple churn” which = (net change in ARR from existing customers / starting-period-ARR) * 4. This is, I believe, what most companies present as their churn rate, includes the effects of both shrinkage and expansion ARR, and is arguably optimistic because it implicitly includes multi-year deals in the starting ARR.
“Simple net churn” which = (churn ARR / starting-period-ARR) * 4. This presents churn net-of (i.e., exclusive of) expansion ARR.
I have discovered that there are timing issues with the gross ARR renewal rate, defined above. For companies that do multi-year deals, you will end up including expansion ARR in your ARR base as it is sold along the way, but only reflecting it in the renewal rate when the contract renews, in effect deferring good news until renewal time, and seemingly failing to take credit along the way.
 Note that in a multi-year prepaid contract that bookings (order value) equals total contract value (TCV). When multi-year contracts are not prepaid, bookings are only the first-year portion of TCV.
[1a] Some purists would argue that having the right to raise the price 3% should set the denominator of subsequent renewal rate calculations to 1.03 * original-value. While I get the idea, I nevertheless disagree.
 The renewal order is for three years, so to calculate the ARR we need to divide the bookings value by three.
 Saying our “churn rate was 10%” makes things sound OK, but saying we churned $2M in ARR is, to me, somehow more visceral. That is, we had to sell an extra $2M in ARR just to make up for existing business that we lost.
 A leaky bucket starts at one water level, during a period new water is added, some water leaks out, and the net change establish the ending water level. (Note that in leaky bucket analysis, definitionally, leaks are never negative.)
 Now might be a good time to download the spreadsheet accompanying this post so you can see my calculations. In this case, the churn is the difference between the total value from product A on the original order versus the renewals order.
 Subscription bookings typically turn into cash within 90 days.
 In reality, we should both uplift the price in years 2 and 3 and discount by the renewal rate to get a better expected cash collections figure. (There is nearly endless detail in analyzing this subject but I will make simplifying assumptions at times.)
 Otherwise, it would juice 2Q renewal rates and depress 3Q renewal rates, making both less meaningful.
 Bonus question: how would you handle the late-renewal scenario at the 7/20/13 board meeting? Answer: I would publish provisional renewal rates that exclude the transaction, letting the board know we have an outstanding renewal in process. Then once it closed, I would revise the 3Q renewal rates accordingly.
 Which then begs the question of how you count customers. For example, while GE has one logo, they have numerous very independent divisions in a large number of countries.
 Note that purist might argue that since we had the right to raise prices up to 3% that we should put 103% of the ARR in denominator in this and all similar calculations, thus dropping the resulting renewal rate here to 100%. While I believe annual increases are important, I still believe renewing someone to 103K in ARR who was at 100K in ARR is a 103% renewal. Tab 3 of the supporting spreadsheet plays with some numbers in this regard.
 It is a good idea to divide churn into 3 buckets to describe the reason: owner change (including bankruptcy), leadership change, and customer dissatisfaction.
I’m Dave Kellogg, advisor, director, consultant, angel investor, and blogger focused on enterprise software startups. I am an executive-in-residence (EIR) at Balderton Capital and principal of my own eponymous consulting business.
I bring an uncommon perspective to startup challenges having 10 years’ experience at each of the CEO, CMO, and independent director levels across 10+ companies ranging in size from zero to over $1B in revenues.
From 2012 to 2018, I was CEO of cloud EPM vendor Host Analytics, where we quintupled ARR while halving customer acquisition costs in a competitive market, ultimately selling the company in a private equity transaction.
Previously, I was SVP/GM of the $500M Service Cloud business at Salesforce; CEO of NoSQL database provider MarkLogic, which we grew from zero to $80M over 6 years; and CMO at Business Objects for nearly a decade as we grew from $30M to over $1B in revenues. I started my career in technical and product marketing positions at Ingres and Versant.
I love disruption, startups, and Silicon Valley and have had the pleasure of working in varied capacities with companies including Bluecore, Cyral, FloQast, GainSight, MongoDB, Pigment, Recorded Future, and Tableau.