February 11, 2016

Reducing SaaS customer churn – Part I (The problem)

In a previous life, I used to work in CRM at a B-to-B telco which sold voice and data services to multiple market segments, from SMBs to global accounts. The main business metrics were customer acquisition costs (CAC), monthly recurring revenue (MRR), customer life-time value (LTV) and customer churn (which measures the attrition rate of one’s subscriber base). Fast forward 10 years to SaaS (Software as a Service) cloud computing, whose main metrics are – CAC, MRR, LTV and customer churn!

In the first of this three-part series, we’ll examine why customer churn – and its associated metrics of CAC, MRR and LTV – is so critical to the success of SaaS cloud vendors. In part II, Reducing SaaS customer churn – Part II (Metrics and reasons for churn) we’ll review the various types of churn metrics and explore the main reasons why some SaaS customers don’t renew their subscriptions. And in the third and final post, Reducing SaaS customer churn – Part III (Levers for reducing churn), we’ll look at the options available to vendors to reduce customer churn.

For those interested in exploring this fascinating subject in more detail, there’ll be a Further Reading section at the end with links to excellent in-depth studies on SaaS business model profitability based on the four variables of CAC, MRR, LTV and churn.

But first, let’s compare the fundamental business models of on-premises vs SaaS to see why churn is a non-issue in the former and everything in the latter.


Traditional on-premises B-to-B software vendors have a front-loaded revenue model in which the lion’s share of a customer’s life-time value (LTV) is realized at the signing of the contract. From an accounting perspective, the software license revenue is recognized immediately and annual maintenance revenue is recognized monthly. Project implementation services revenue is recognized over time based on percent-complete milestones.

Once a company has implemented the software and integrated it with its own data, processes and infrastructure, it has effectively created vendor lock-in. The option of switching to another vendor is virtually non-existent because of the high sunk costs – both financial and political – and the time and effort it would take to start all over again.

Under such a model, the main business metrics are therefore up-front license sales and on-going maintenance revenue (which can virtually be taken as a given over at least a 5 year period because of the vendor lock-in effect described above). And since 70-75% of total customer value is realized during the first year, customer churn doesn’t even figure on the radar screen, and the concept of customer life-time value (LTV) is largely academic.

Finally, high customer acquisition costs (CAC), mainly in the form of marketing, sales and pre-sales teams interacting with the prospect over a sales cycle of at least 3-6 months, determine the minimum deal size below which it is not profitable to do business (usually 100-300 users).


SaaS is a utility-type rental approach to software, based on standardized on-demand services running on a multi-tenant or shared infrastructure. This model naturally favours volume and scale, resulting in a more efficient use of capital, higher resource productivity and hence lower unit costs.

The standardized nature of SaaS is particularly important for our discussion. Unlike on-premises customers, who each run a highly-customized product on their own infrastructure, SaaS customers all work off the same version of the software running on shared infrastructure (with personalization limited to pre-defined configuration options). This standardization not only keeps unit costs down, it also limits vendor lock-in, making it easier for unhappy customers to leave.

When it comes to customer-vendor interactions, there are multiple models for SaaS. This varies from “no-touch” self-service (ie, web) thru “light-touch” inside sales (ie, telephone) to “high-touch” enterprise sales (ie, feet-on-the-ground and face time). Customer acquisition costs (CAC) of course increase in proportion. For an excellent overview of this scale, see slide #46 in this SaaS Business Models presentation from Matrix Partners, and slide #13 in Selling SaaS Online, from the Kampyle SaaS lecture series.


Regardless of product complexity and customer touch points, the SaaS revenue model is based on monthly recurring revenue (MRR). From an accounting perspective, revenue is recognized monthly – including annual advance payments, which are recognized in 1/12th increments, with the outstanding months accounted for as deferred revenue (a liability on the balance sheet).

So, unlike an on-premises vendor that sits on an ever-increasing pile of cash as front-loaded revenues pour in from soon-to-be captive customers, a SaaS vendor’s financial situation is only as good as next month’s invoicing, because unhappy customers are “free” (up to a point – see next section) to cancel their subscriptions and leave. And when this revenue loss starts to offset significantly the revenue from newly-acquired customers, the net result affects both the top line and the bottom line. Welcome to customer churn!

Under the SaaS model therefore, profitability is directly related to customer life-time value (LTV), whose growth is directly linked to happy customers – who always renew their contracts. So the longer a customer has been on board, the more reliable his revenue stream becomes.  Customers will tend to churn more during their first year of service (especially during the first 3-6 months) than in subsequent years.


On the subject of vendor lock-in and vendor switching costs, this can vary significantly based on the particular SaaS model. At the no-touch, self-service low-end, everyone uses the same canned processes, which leaves data as the main variable. Vendor lock-in is therefore minimal and switching can be relatively easy, for example, moving from one SaaS online survey provider to another.

As we move up the SaaS business model scale though, the impact of customer-specific process configuration, integration to other systems and organizational change management starts to raise the switching stakes. So at the high-touch end of the scale with enterprise-level SaaS, you could very well end up with vendor lock-in. While nowhere near the scale of on-premises lock-in, it can nonetheless be dissuasive enough for the end result to be effectively the same.

For example, if an enterprise SaaS customer is one year into a 1000-seat, multi-country ERP or CRM implementation, then there will most likely be de facto vendor lock-in, even if on paper the customer has no financial commitment beyond next month’s invoice or the annual renewal date (and, of course, there is no outstanding depreciation to write off, since SaaS rental costs are opex).

Such implicit vendor lock-in, however, does not obviate the need to keep enterprise SaaS customers happy. Unhappy locked-in customers who feel they are not realizing the value of their investment are bad for business – just ask the current ERP and CRM on-premises software vendors.

In part II of this article next week, Reducing SaaS customer churn – Part II (Solutions), we’ll review the various types of churn metrics and explore the main reasons why some SaaS customers don’t renew their subscriptions.


Speak Your Mind