What Companies Get Wrong About the Equation of the Business
I attended the YC Series A program (representing Mozart Data), and one of the heads of the program said, “You need to know the equation of your business.” This is essentially how your business makes money, but expressed as a mathematical equation (where the components can be measured).
A lot of what your early data team does revolves around this equation. They understand the equation of the business and then implement dashboards that monitor that equation or equations.
With the right equation in place, data acts as a barometer to tell you what's working and what's not. But a lot of businesses either end up not taking action based on what their equation tells them or they rely on their equation too much. Here’s how you can turn your equation into actions that steer the business.
Understanding the equation of your business
I worked at a company called Playdom where we made freemium video games. Most of our players played for free and a small fraction of players would upgrade to premium features. That was how we made money.
To understand whether a game was working, we would write down the equation of the business: the number of users times the revenue per user, which is the revenue of the game. This was our North Star at first.
But it’s hard to forecast how many users you’re going to have, so we then narrowed our view. Rather than thinking about total users (which had massive heterogeneity), we started to think about daily active users and our equation became daily active users times revenue per daily active. That was a more accurate equation for predicting how successful a game would be.
How most businesses get it right
At Mozart Data, we offer a free trial and during that free trial we want to know: did a company set up a transform in its first week? Companies that hook up multiple data sources and set up multiple transforms are much more likely to convert to paying customers. We know this from monitoring the equation of our business, which is usage times revenue.
The equation of your business helps you identify those early signals that the business is doing well or not well because you know what good looks like. You don’t have to wait months or years to know whether you’re making good or bad decisions.
The most important thing in data analysis is not some sort of statistical understanding. It's not writing SQL or Python, and it's not being able to build the most predictive model.
The most important thing in data analysis is to first understand the business and what drives the business and then be able to count those things. That's what it means to write down and understand the equation of business. Writing it down forces you to turn the business into numbers and that's a big part of what a data analyst does — they translate the business into math.
That’s what many businesses get right. But that's not good enough.
Where businesses go wrong
After many iterations of your equation, you’ll get a rich, nuanced view of your business that tells you what’s going on (with some early signal) and then you’ll set up dashboards. But dashboards aren’t the solution. This is the first thing that businesses get wrong about using the equation of their business.
At Mozart Data, we help you get the data you need to set up key dashboards. These dashboards aren’t your enemy, but they also aren’t your end zone. The job isn’t done there. Even reacting to changes in the time series and dashboards you set up isn’t good enough. Most of the time, the changes are a pipeline error or something exogenous that is innocuous – like a holiday (no need to worry). You need to go a level deeper to understand what’s causing what.
The other thing companies get wrong about the equation of the business is when it becomes a crutch. It’s a mistake to use the time series as a full picture of the health of the business when, in practice, it’s misleading more often than it is indicative.
Take the freemium games that we used to create at Playdom as an example. If you see monetization decline for a game, most of the time, the problem is not something you’ve done with the monetization itself. It’s something upstream of monetization. If you’re measuring monetization to be revenue per user and you’ve brought in a bunch of low value users, your denominator has gotten worse – even though you’re mentally biased to assume it’s the numerator.
And this goes back to why it’s important to understand what’s causing what. You can’t know why a certain number is changing just from looking at your time series (a close sibling to “correlation is not causation”).
Taking action based on the equation of your business
When I worked at Yammer, one of the key dashboards of the business was daily active users. Every weekend, the daily actives would fall off a cliff because Yammer is a tool that people use at work.
You expect traffic to bounce back on Monday, but maybe that Monday is Memorial Day and you don’t see the traffic bounce back. Most people wouldn't freak out in that situation. They'd say, “I know it's a holiday.”
Now you're starting to apply logic to your time series. You're not just asking if the time series is deviating from normal, but you’re also applying logic to understand why it is. Going back to my Yammer example, you know what’s happening isn’t an issue so there’s no action to take.
The flip of this, and this is important, is you can never prove or disprove the null hypothesis. You can’t prove nothing is wrong, because by that logic you'd have to test every single dimension of what could be wrong.
You have to have a strategic game plan and a set of prior beliefs to walk through in order to effectively debug what’s going on. The most obvious is it may be a data pipeline issue. In fourth grade, I took the classic test where the first instruction was to read through everything and the last instruction was to skip to the last question – instead, I toiled and tried to answer everything. So check upstream dependencies, and then look at your populations and then subpopulations. Maybe you have a theory that the issue is driven by certain customer segments, so look at your data cut by that customer segment. By diving into these things sequentially, you’re able to see the signals your data is giving.
Dashboards are just a starting point for this kind of data work.
Whether you’re a start-up founder or analyst, knowing the equation of your business is critical to steering the company in the right direction and making adjustments as necessary. And this becomes easier when you pair your equation with the tools needed to easily calculate metrics, create a single source of truth for the business, and automate dashboards — the modern data stack.
Learn how a modern data stack helps you understand what’s happening in your business and make the right data-driven decisions.