Data Delusion
Organizations are failing data functions, and data functions are failing organizations. This is the data delusion. Having data or data functions doesn’t equal effective operational management. Without clear strategy and alignment with organizational goals, data functions become a liability.
If commercial leaders were a historic army commander, many would be like General Hooker at Chancellorsville in Virginia, 1863.
General Lee outmaneuvered General Hooker due to superior information. Lee knew of unguarded paths in Virginia’s dense forests, enabling Stonewall Jackson to flank Hooker’s forces. This information asymmetry led to a decisive Confederate victory. These lesser-known paths exemplify exploitable local minima in operational terrain.
To avoid Hooker’s fate, understand data functions and their role in operational strategy. Data functions are more than just analysts; they’re cartographers and surveyors combined. They’re strategists, artists, and engineers who can map your current position, desired destination, and the path between.
Think of data functions as lieutenants in your operational army, best equipped to understand your operational terrain. They need a clear objective function to optimize towards - the metric defining success for your organization (e.g., customer retention, revenue growth, operational efficiency).
If you don’t know your objective function, start by mapping your terrain. Let’s assume you know you’re losing customers faster than acquiring new ones. Your objective function becomes “improving customer retention.”
In this scenario, surveyors provide factual representation of your terrain. The cartographer then analyzes this information, providing a representation of your operational system. For example, they might identify two primary reasons for churn: user engagement and the rate of new feature releases.
The cartographers draw up a slice of the operating terrain:
The terrain contains both local and global minima. Local minima are areas where customer retention improves due to certain engagement and feature release velocity levels. The global minimum is where customer retention is highest, optimizing both engagement and feature release velocity. Our objective is to reach this point.
Assuming we can plot the current state of churn relative to engagement and feature release velocity, we can see our position on the map and potential improvements given the system’s current state. Clear valleys in the terrain represent alternative system states, or local minima, that we can optimize towards.
This approach aligns with gradient descent algorithms, where we locate global and local minima (or maxima, depending on the objective function) to find the point of maximum customer retention.
Effective data functions can provide this analysis for any given objective function, showing your current position, potential improvements within system constraints, and optimization strategies. However, they need a clear objective function to work with.
The best can optimize for multiple objectives simultaneously, finding a global peak or valley that maximizes multiple metrics
Think of your organization as a blank parchment and data functions as cartographers and surveyors skilled in measurement and representation. They map your operations, highlighting peaks and valleys, and guide you from your current position to your desired destination but only if you provide a clear objective function.
To avoid the “data delusion” that befell leaders like Generals Hooker, understand your data functions’ role in your operational strategy and clearly define your objective function. If you’re unsure of your objective function, start mapping your terrain to understand your current position, desired state, and the path between them. Your data functions are your lieutenants in your operational army, and every great leader needs a map to guide them.