Publications

By Tim Graettinger and Bill Lazarus

Anecdotes, stories, and folklore are very useful. They are compact ways to communicate complex topics and to inform day-to-day decisions. But stories age with time; what once was true may no longer be so. And as Mark Twain famously said, "It ain�t what you don�t know that gets you in trouble. It�s what you know for sure that just ain�t so."

In this article, we share real-life examples where various bits of corporate folklore and anecdotes - what people "knew for sure" - were seemingly at odds. Data mining, visualization, and predictive analytics effectively put these stories in a broader context. The results were fresh insights and fresh approaches to solving the problems that these companies faced.

Our first example focuses on a large medical imaging company. Their delivery network, largely built through acquisition, includes thousands of patient contact sites nationwide. Growth had been painless, initially. Then, problems began to surface, and so did the anecdotes. "This site is terribly underutilized." "That site is stretched beyond capacity." "Dr. Smith�s patients are going too far for service." "Patients coming to this site drive 15 minutes or less." The company folklore appeared inadequate to respond. Maxims like "Open more locations", and "Closing a site is risky" once ruled the day, but now there seemed to be plenty of locations and some certainly were inadequate. Could the delivery network be optimized and still perform for all the stakeholders?

To address these growing pains, we changed the perspective to a patient-centered one. The viewpoints expressed in the anecdotes above are biased subtly along departmental lines: a site-centered perspective from operations or a physician-centered perspective from sales and marketing. A critical step was building a new patient-centered analytical data set (not a data warehouse). We combined data elements from departmental "silos" with third party demographic and geographic data. And then we asked a patient-centered question, "Where should I go for service?"

Through data mining and visualization, including geo-spatial analysis and mapping, a clear picture emerged. Patients were making rational decisions based on where they lived. The previous service-site-based analyses only looked at patients that already came to a site, without considering the ones who might have come, but didn�t. Previous physician-based analysis only looked at patients who visited a particular doctor, without considering other patients in the same market area. The new patient-centered view clearly identified those sites that were underperforming and that could be closed. It also pinpointed underserved areas where further growth and expansion could occur.

Read more.