3 most common reasons for wrong SaaS metrics

There are a couple of reasons for calculating SaaS metrics, and metrics in general. One is that you want to share them with external people, like investors. The other is that you want to make some strategic business decisions based on those metrics.

Either way you will want to make sure that your calculations are correct. Otherwise you jeopardize the whole effort. I’ve seen quite a few ways of calculating SaaS metrics, and here’s a list of most common reasons for them being incorrect.

Not using reliable data

Gargabe in, gargabe out. This is a well known principle in computer science. When your underlying data is trash, you cannot expect that any calculations based on this will be good, right? This also, unsurprisingly, applies to calculating metrics.

People tend to use all sorts of input data for SaaS metrics. Spreadsheets. Sales data from a CRM. Invoices.

From my experience you want to rely on invoices. Invoices are the most trustworthy source of information, because they are legal documents. This forces quality. It’s not the case with other sources. For example sales data is added by many people, with different assumptions, driven by different goals (like quarterly targets).

Lack of automation

It’s ok to do things by hand when starting. With 10-20 customers you can still manage your data by hand, using spreadsheets. With 50 it’s already getting complicated. 100 is unbearable.

Because of the repetitive nature of this task you will get sloppy. Also if you grow your business, you get even more work. The more customers, the more things to update, the more potential problems. A common mistake here is confusing New MRR with Renewal. This happens because you have to understand the whole history of the customer to correctly attribute your invoice to either of them. It means you have to look back through your whole input file.

Not identifying customers correctly

This can be somewhat a part of previous points, but is a big topic in itself. A customer is a separate entity in terms of metrics. In most cases you can calculate metrics for every customer separately and then just join the results together. This means that if you misidentify some payment, things can get ugly. If you end up with two customers (A & B) representing in reality just a single customer you will see invalid metrics. Churn from customer A and New MRR from customer B (instead of just continuous MRR for a single customer) is a typical problem.

In this example Customer A churns in June and renews in July. Customer B is treated as a new customer (new MRR) in June and churns in July. If the reality is that they are the same customer, the situation would be totally different: continuously $100 MRR and no churn/renewal. Other than that your total number of customers could also change, as well as average revenue per account.


Every business wants to have reliable metrics. It boosts confidence and trust. Empowers your decisions.

The beginnings are ok, because there is not much data. You will sort through it by hand.

Things get tough when you start growing. From my experience crossing 30-50 unique customers is a good trigger to think about this problem in a more structured way. Prepare an input dataset that you trust and automate the task of calculating metrics. It does not have to be anything fancy, but make sure you codify the rules.

Of course using a fully automated solution will make this problem easier. Fully managed solutions like Probe or automated metrics from Spreadsheets via Probe plugin take a lot of burden off your shoulders. All the details and corner-cases (and there are quite a lot of them) are already solved for you. Integrations with common invoicing software will make sure that you can rely on what you see. You get more ways to slice the data. Look at it from different perspectives.

Happy measuring! Michal, Co-founder, Probe