“The best leading indicator of the success of a B2B SaaS company is user engagement.”
Boston VC in Conversation

User engagement can predict future success. So how should we be measuring engagement? This is a frequent topic of conversation here at TeamFit. We track engagement daily both for the overall platform and for each company. At this time, it is the main thing we are working to optimize.
In their excellent book, Lean Analytics, Alistair Croll and Benjamin Yoskovitz define engagement as ‘daily use.’ OK, that looks like a good place to start. But what does ‘daily use’ actually mean, and what if your application is not intended to measure engagement?

At TeamFit, there are a set of key actions that people can take that indicate engagement and suggest they are getting value. Some of the key things we care about are …

  • Claim a skill
  • Suggest a skill to another person
  • Rate a skill
  • Provide evidence of a skill
  • Start a project
  • Invite people to a project
  • Join a project
  • Interact on a project record
  • Search for a skill
  • Search for a person
  • Search for a project
  • Search for a company

That is just a start. TeamFit is a rich application that will support the full range of interactions around skills and projects. TeamFit will integrate data from many other systems, especially project management, collaboration and communication.

That is a lot of different data points to track and make sense of. Some of us look at these everyday but we have a general consensus that we need a single number to optimize and that this number should not be a simple measure of daily active users.

So what should it be? We are investigating something we call Predictive Engagement, predictive e for short.

Predictive e is a metric that predicts its own future value.


Predictive e is an integrated measure of usage that predicts future usage.

This number will be unique to your solution and how it is calculated will evolve over time. (Yes, we know that makes benchmarking more difficult.) The inputs are all of your critical usage metrics. You can use any of a number of prediction engines to calculate e and then test it against the data for the following week.


Of course it could be that e successfully predicts zero future usage. So you need to design e to measure increasing use. And you will still likely have several flavors of e depending on your business model. You may need to measure e for individuals, for companies or business units, for teams, or for cohorts.

But at least with e you have a number that has predictive power. That is the purpose of a good metric; it helps you shape the future.

At its heart, TeamFit is a prediction platform. We are building it to predict:

  • The probability of project success (and tell you what you need to do to improve the odds of success)
  • What skills a person has
  • What skills a person could have
  • Who will work well together
  • What skills will be in demand in the future

What do you need to predict?

Predictive e is named in honor of the great mathematician Leonard Euler.