Every analyst is asked to produce actionable insights. It’s an expression of this sentiment: “please don’t give me one more inscrutable report with charts, I just want insights on what to do!”.
Here’s the good news: this is a sign of a curious organization, one that wants to learn from data. But, insights aren’t magical items conjured by analysts wielding sql. In this post, I want to talk about a few foundational pieces that enable a strong data or insights-driven culture.
- First, calculating metrics is insufficient. Reports, dashboards, executive or business reviews all contain a hodgepodge of measurements, which reflect what has already happened, and knowing they have happened is not inherently insightful. Besides, metric aggregations deal in sums and averages, which are not insightful either. Embracing segmentation as a first class citizen in the tooling, in the org-wide discussions, in presentations, this is a key evolutionary step. This will result in a desire to slice and cut the data constantly many ways - to disaggregate into component parts, to understand drivers, and to confirm or reject hypotheses.
Some of this work may seem excessive, even wasteful at times, but it is important culturally to establish a perpetual state of dissatisfaction with aggregate metrics and aggregate thinking in general.
- Second, a metric is not an island. There are systemic connections that need to be deeply understood and internalized. A simple example is the often inverse relationship between increasing top-of-the-funnel metrics and downstream retention metrics in any growth equation.
- Thirdly, combining the power of metric disaggregations and metric connections, we need to constantly reshape and rebuild aggregated models of how the business processes work, and what the levers are. There are equations present in all parts of the business - in growth, whether it’s sales or marketing, in operations, in service teams, and especially in finance. Being able to distill the analytical work into higher-level models of business processes and controllable levers, input and output variables to be simplistic, enables action.
And finally, taking action is crucial in an insight-driven culture. If you don’t do anything, then the analytics can only capture is what is already underway - which may be the result of actions taken months ago, or even years ago! It is relatively easier to convert actions into insights than the other way round. So, a bias to action is a very good thing.
In summary, there are four ways in which we can evolve away from the stalemate of having hundreds of reports and business functions still demanding insights - segmentation and dis-aggregation, understanding metric connections, re-aggregation and an understanding of levers, and finally taking action, which completes the loop.
And yes, we absolutely need to evolve the tooling to support this cultural evolution.
