What’s a data person to do when looking at the data simply isn’t enough? That question came to my mind recently while working on a project. We were asked by a client to develop a model to inform them on how long they might face their current problem.
In the midst of our work, we received some anecdotal information from counsel about the situation we were analyzing — information that gave us pause. Up to that point, the analysis had looked at various data points over time, correlations of variables to settlement amounts, dismissal rates, defense costs, and times to resolve cases. However, the information that counsel shared was not observable in the data available because of the data points and level of detail that had been tracked over the years. (Side note: Always track data in the most granular way feasible, as it can always be put back together but not always taken apart.)
Upon reflection of the new information, we realized that our model, while statistically sound and backed up by “data”, was not accurately reflecting the reality facing our client.
We needed a model that would be a useful tool for our client — not just statistically sound. But we aren’t in the business of making things up. So after learning all we could from counsel about what they observe “on the ground” so to speak, and looking at the data again through that lens, we were able to generate additional scenarios. The new scenarios took some of the counsel observations as inputs into the model. Studying the results of the various scenarios and working together with counsel and the client, we were able to incorporate the qualitative assumptions together with the quantitative assumptions.
We didn’t toss the initial model in the recycling bin. Instead, our additional models that incorporated this softer side of the analysis — when combined with a sensitivity analysis around all the models and variables — resulted in a more robust tool for the client.
The Moral of the Story
Data is everywhere and all of us are generating it constantly – blogs, emails, text, purchasing data from brick-and-mortar stores and online, social media, etc. Advances in technology are making it easier to link those data points into a “story”. Understanding what all that data is telling us, or making use of that data in a business decision context, is part art and part science. Thorough analysis of the data available, in conjunction with communications with the people generating the data as well as those who want to use that data for some purpose — the data consumers if you will — is imperative to providing the appropriate context and maximizing the data’s value.