Beware the enticing call of solutions offering to “easily” calculate your SaaS metrics with no effort. These metrics are key performance indicators for your business and essential to helping guide your executive team to make critical decisions. Is a simple, cost effective, effortless dashboard estimating these metrics good enough? Ease of use is always preferred, but not when the data can potentially mislead or misinform you.
For a small business just starting out, simple estimations might make sense. Priorities at this stage tend to lean towards focusing on in-the-now, versus planning for the future. However, as a company matures, having generalized metrics may not be enough. For SaaS metrics to be useful at scale, you cannot just use a one-size-fits-all solution. Every business is different, and therefore should individually tailor each metric to their needs. These calculations should be complex, and adequate time should be spent on refining them to get proper insight into the business.
Let’s take a look at a few examples.
Churn
Without a doubt, Churn is one of the most important metrics for a SaaS company to monitor. One popular generally accepted calculation is:
Churn Rate = Customers Churned in Period / Total Customers at the Beginning of Period
Straightforward, right? But let’s look deeper look into this calculation. How do you define what a churned customer is? Off the top of my head, here are a few types:
- Involuntary — Credit card expires, person championing your product leaves company, etc.
- Voluntary — On-boarding/losing interest in the first month, realizing the solution does not solve their problems later
- Temporary — Pausing the subscription with plans to come back eventually
We could make a strong argument that a temporarily churned customer should not be factored into the calculation as the purpose of measuring churn is to gauge the ability of the business/product to retain customers (and in this scenario, the customer is technically retained, despite the pause). We could even extend this argument to some involuntary cases as well. A customer could be completely happy with the product, and have high engagement, yet due to forces outside their control, leaves. This muddles churn analysis, as it has nothing to do with your product’s ability to provide value to its customers.
Now let’s focus on the churn calculation itself in a high growth company.
| Month 1 | Month 2 | |
|---|---|---|
| Existing Customers | 100 | 140 |
| Existing Customer Churn (5%) | 5 | 7 |
| New Customers | 50 | 50 |
| New Customer Churn | 5 | 5 |
| Total Churns | 10 | 12 |
| Churn Rate | 10% | 8.57% |
Here, when we use the exact same churn inputs month over month, the results differ by 1.43%. Probably not the most reliable calculation. This happens because the numerator in the equation is over a time period, while the denominator represents a snapshot at the beginning of the period. Large gains in customers every month will heavily affect the calculation.
Let’s refine it by measuring on churns per customer day instead (with the assumption both months have 31 days). We will also assume new subscriptions and churn come in at a constant rate throughout the month to calculate customer days. So we’ll multiply by 0.5 to simulate this with the equation (Existing Customers * 31) + (0.5 * 31 * Customer Net Gain). We’ll extend the above table with the following:
| Month 1 | Month 2 | |
|---|---|---|
| Customer Net Gain | 40 | 38 |
| Customer Days | 3720 | 4929 |
| Total Churns per Customer Day | 0.27% | 0.24% |
| Monthly Churn Rate | 8.33% | 7.55% |
Now the difference is 0.79%, so definitely an improvement! However, what about measuring over different time periods? By evenly spreading out the new subscriptions and churns, measuring by quarter we would get misleading results as well. We could refine this even more by getting rid of our linear net gain assumption and just calculate the exact customer days for each individual customer. Obviously this would require an extreme amount of work, but it will also be that much more accurate for analysis.
Shopify has a much more detailed explanation of this particular problem
Another possible solution is to segment your churn by new customers and existing customers, but this adds complexity as well since you now have two metrics to analyze for churn.
MRR
MRR may seem more straightforward than churn, but can also get somewhat complex. One particular challenge is properly identifying what to actually include in your calculation. For instance, a strong case can be made for discounts being included as it directly affects the subscription contract (recurring basis). What about credits though? Refunds? Add-ons? Variable charges that could be justified as “recurring”? The list goes on and on. This metric is highly customizable to address different types and needs of businesses.
Deciding what should or should not be included in the calculation for MRR should be agreed upon with the executive team and potentially board, to properly set expectations across the company. Otherwise it will only lead to confusion and ultimately mistrust of one of the key metrics for the business.
Expose the real stories behind your analytics
Plug and play solutions that deliver generic estimates on metrics can be dangerous to rely on for measuring critical performance metrics for your business because they may hide important news. This is especially true once you dive into the unit economics of your customers, such as LTV and CAC. These calculations are vastly more complex, include critical data from the cost side of your business, and can be infinitely fine-tuned to the needs of each unique finance organization in a company.
I highly encourage businesses to think through how these metrics apply to their own organization first, before attempting to find solutions to help derive them. When deciding on a solution, make sure that it provides the easiest way to get the exact information your business needs without sacrifice in accuracy and quality.
