I am somewhat odd, in that I actually started my professional career as a computer programmer/analyst in the Consumer Price Index. The old personnel line in the federal government was "we can't afford techies, so we train them." Apparently my background as a failed PhD candidate in economics and editor of my college newspaper gave me a nice combination of left-brain/right-brain aptitude to make me worth training and quickly promoting: I have to say that I learned and did more in database analytics in two short years than I ever did again in many years of association-related work.
What does this have to do with associations? Plenty, I should think. The CPI gave me a great tool set and a perspective that works to some extent with every association I worked with as staff.
- At CPI, infrastructure, mission, and support network all contributed to and benefited from data mining. Our explicit mission was to collect and data on product characteristics and their prices to develop and report a series of monthly indices, so our small army trooped around retail outlets to collect 200,000 price quotes a month; forty of my colleagues would review their work through exception reports; a smaller team of us were engaged in ongoing quality initiatives to ensure that we were measuring inflation properly. On any given day there would be 3 or 4 staff accessing the central database running simulations of alternative methodologies or writing academic-quality papers, and many more reviewing the individual data points to question odd changes or to accept/reject new models and other substitutes in our sample.
- At NAHB I/we did large dataset analysis in support of their Housing Economics journal as we monitored patterns in housing sales, design trends, etc. but the thought of applying this brainpower to marketing didn’t really work for our very large annual meeting or our grassroots/federated structure of membership.
- At ASHP, I worked with many pharmacists on staff who grew up reading the professional literature and many were editors for our AHFS drug information product—a several thousand page annual subscription service that served as a guide for drug use in most hospitals in the world. Yet I was the first person to convert their aging AS/400 database into something we could use to segment the audience to drive marketing plans, budgeting, results and penetration analysis. With a database of 180,000 names but only 30,000 members and maybe 15,000 other customers, we needed this service—we just didn't have it.
- At CRS, we used Pledgemaker to manage a database with 350,000 active donors; another 700,000 former donors; millions of historical donations; and 40 million contact records created by our acquisition program. Unlike membership organizations, in disaster relief operations, there are donors who lie dormant for years but when a tsunami or earthquake or colossal famine strikes, these donors are assets who don't find it strange to be contacted and who generously give again if you're among the first to contact them. Once we hired staff, purchased FirstLogic and SAS, created a duplicate donor file for analysis, we were able to do far more, saving $250,000 a year on data processing bills, spending some of that money to collaborate with an outside firm to create scoring models for each acquisition campaign to save several million more and improve performance.
In future posts I will do much more to explain our methods, philosophies, and illustrate key examples to help others do their own in-house data mining, but for now I wanted to begin with a history lesson...