Moving Beyond MRR: Evaluating New SaaS Pricing Models

2839a4f by

Over the past decade people leading and investing in early-stage companies have become obsessed with MRR — monthly recurring revenue. Has this obsession passed its best before date?

Monthly Recurring Revenue. The term comes up in just about every meeting between an early-stage company and its investors. In fact, many VCs will only consider investing in a company once it reaches a certain level of MRR. Valuations are based on some formula determined by unit economics, growth rate and MRR. And business processes are optimized around MRR. How did we get here?

First, let’s take a look at what the components of MRR actually represent.

Monthly – generally it takes three data points to establish a trend. If you focus on quarterly results, as the public equity markets do, it takes most of a year to establish any given trend. You just can’t manage a high growth company this way. You certainly can’t make investment decisions either. So, the focus on looking to monthly numbers makes sense for high-growth companies.

Recurring – this is the key point, the revenue keeps coming back, it is predictable. Investors like predictable. They like to think that if they pour in so many dollars they will get some multiple back. In the words of one VC in Vancouver “I am looking for a debugged sales process.”

Revenue – well, revenue is revenue, and with SaaS the key is to balance pure subscription revenue with professional services. Professional services revenue tends to be discounted by investors as it is typically not recurring. One wonders about the focus on top-line implied by the MRR obsession. Fortunately, investors’ other obsession, unit economics, provides a good governor for this.

No SoftwareSo again, how did the SaaS world become so focused on MRR? and its ‘No Software’ logo are partially to blame. pushed the industry towards the Software as a Service model (SaaS) in which there exists a single, multi-tenant application. User specific needs are supported through configuration or external modules. And today, this model is winning because it is more efficient to develop and (when bandwidth and storage costs are low) to operate. Users pay to access functionality and store their data through a subscription, rather than through a purchased license.

The subscription model has some important advantages over the licensed, on-premise alternative. Commitments are short-term, they generally come with some form of a try before you buy option (freemium or free trial), pricing tends to be simpler and there is the opportunity for new pricing metrics that better track value — consumption-based pricing so to speak.

Modern B2B SaaS offerings are heavily instrumented. The vendor and the business can (or should) know a lot about use, making it possible to constantly improve user engagement (there is even a movement towards building habit forming software, see Nir Eyal’s book Hooked). More advanced companies worry more about customer value than user engagement. Is the software actually creating business value for the company? How do you know? How much value?

So far so good, and the ‘subscription economy’ heralded by companies like Zuora and Aria seems to be real. Are there any problems here?

There is a tension between the desire for consumption-based pricing, the on-demand model, and the need for predictable revenue.

B2B software vendors don’t really want to provide consumption-based pricing, or even value-based pricing. They provide steep discounts (as high as 50% though 20-30% seems to be the normal range) for annual commitments and customer success teams (who have replaced customer support) are generally compensated for maximizing renewals.

It is not just the vendors who want predictability. Corporate buyers need to be able to predict and manage their costs. Only when use of the software directly tracks revenues — or even better, profits — can they really consider consumption-based pricing.

The key message? The goal is predictability. Subscriptions are just a way to get to predictability, for vendors and customers.

Subscriptions are a way to manage risk. SaaS software vendors are giving their customers a risk discount. Companies are not sure how much use or value they will get from a subscription, so they need a risk discount to make a commitment.

Prediction Factors

Are there ways to predict use or, preferably, to predict value? If there are then SaaS vendors can reduce the risk discount and make more money. And companies will be happy to pay these higher prices because they will be taking on less risk in adopting the software.

Predicting use, and then value, is an emerging application of predictive analytics. As these engines get traction and demonstrate that they actually can give insight into future use and value we are likely to see new pricing models emerge, based on quantitative predictions of use and value.

Who is going to deliver this? Possibly one of the big subscription management companies. But, I suspect not. This kind of innovation is more likely to come from a company focused on granular usage tracking, a customer experience optimization or customer success vendor. Or just as likely, a new company, one that I have not yet heard of, is already hard at work on this.

New Call-to-action
  • If you haven’t, I’d suggest reading a short book called B4B. It really speaks to the same ideas you are covering…maybe not from the billing perspective but definitely from changing the supplier/customer relationship side.

    Here’s the link to the book on Amazon

    • Thank you. Have ordered the book (isn’t Amazon wonderful, in some ways). The foundations for this come from the blending of value-based pricing (see Tom Nagle, John Hogan and Joe Zale The Strategy and Tactics of Pricing ) with predictive data. The big dynamic pricing engines like PROS have been doing this for some time for things like air line tickets, hotel rooms, even parking spots in the city. What changes with SaaS (and the IoT is the amount and nature of the data available. Design of instrumentation for SaaS and IoT is going to be a very interesting field over the next decade.

  • Stevan, I think getting the data is the problem. I see two ways this can be applied.

    1. Company doing this internally using only its data

    In this case, prediction unlikely to work. How a company use one software package is not predictive of the other (e.g. accounting SW does not predict how sales SW is used or value added).

    2. Central company collecting data and predicting

    Then data about value for a company and even use might be closely guarded. Most companies will not share the data.

    So it is unlikely to work for above two. Is there a third usecase I missed?

    • We have built a platform that does item 2 –
      Kairos measures customer lifetime value (CLV) using advanced machine learning – so provides critical insight to marketers, but it could easily drive dynamic and predictive pricing models for SaaS … I’ll give this more thought and seek feedback.

  • Steven – check out what we are doing at Kairos Analytics – maybe this is the new innovative company you mention in your last sentence.