Hi ,
Data science projects are notorious for their high failure rates and most data science models fail to deploy.
So, what if businesses gave up on data science entirely...
And replaced their data science teams with "metrics teams"?
The concept of a "metrics team" was first suggested by CFO Advisor Lauren
Pearl, in the context of finance, but applies equally well here.
Many data science teams spend more than 80% of their time on model building, but less than 20% of their time on telling business leaders what they really need to know - what actions they need to take to achieve their goals.
A metrics team, though, would turn that upside down.
Based on Lauren's idea (with a few tweaks to allow for the context shift), the metrics team would:
- Start with the business goals and associated metrics, such as revenue, profit and expenses.
- Determine the variables or actions
that are expected to drive those metrics.
- Use data to identify high-value opportunities by understanding the causal relationships that exist between the drivers and the metrics.
The metrics team could spend 20% of their time on ML model building if they believed there was value in that.
But by devoting 80% of their time to identifying actionable insights, the team's focus would be on the highest-impact tasks.
And of course, the data science skill set makes data scientists a shoo-in for metrics team roles - if they can just make the necessary mindset and priority shifts.
Metrics teams are something that doesn't yet exist, but that doesn't mean you can't work this way.
Imagine how your impact as a data scientist would change if you allocated 80% of your efforts to enabling better decision-making.
Regardless of what you call yourself, this shift in focus could be the difference between failure and success.
After all, what's in a name?
Talk again soon,
Dr Genevieve Hayes.
