Experienced analytics or data science teams know that two key challenges for analytics projects are making sure you solve the real business problem (framing the problem) and making sure you can operationalize the result(deployment). In this post I am going to talk about framing the problem and I’ll follow-up with another on deployment.
One of the best ways to think about framing is to consider the questions the analytic team should get answers to before they start building a model:
- How will we measure success?
What are the BUSINESS metrics that define success? Not the statistical or analytic measures, but the actual business measures that sponsors care about. - What decisions does the organization make that have an impact on this measure?
While most teams manage 1, they trip up here. They jump straight from measures of improvement to analytics without considering the business decisons that will have to be made differently – analytically – of results are to be improved. - What question has to be answered to make a decision?
Understanding how each decision involved is supposed to be made is critical. Defining questions and allowed answers as well as building a model of decision making make sure everyone is on the same page. - Who makes these decisions and when?
Are you embedding your analytics in an automated system, persuading a knowledge worker or instructing a front-line employee? You need to know if you are to build the right analytic. - Which decision(s) are we going to change?
Where’s the pressure point? Where can analytics make a real difference? - What are the side effects of changing these decisions?
Last question. If we change this decision to be analytic, what side effects might there be?
All these questions help direct an analytic or data science team, create what CRISP-DM calls Business Understanding, and frame the problem so it can be effectively solved.
If you are trying to make sure that analytics are effective, learn more about framing in our webinar Analytics Teams: 6 Questions to Ask Your Business Partner Before You Model (updated: the webinar has taken place. The link will access the recording ).