Tuesday, December 16, 2014
Earlier this fall, we co-hosted an event with the Partnership for Public Service that featured three federal executives leading innovative analytic efforts in their agencies. Over the last few weeks, we extended the conversation by blogging on seven gover

In this final installment, we provide highlights from these federal leaders on the most important ingredients for a successful analytics program. (You can watch the video of the panel discussion and listen to each of the seven podcast interviews too.) The executives profiled complex programs in several agencies that have a wide impact on citizens, who benefit greatly from leveraging data as a strategic asset in program operations. What follows are some highlights from those executives on salient take-aways for government and stakeholder groups who are implementing key data-driven programs. Steve Beltz, Recovery Accountability and Transparency Board: “You have to have good data, good analysts and good tools, what I refer to as a three-legged stool approach. If you’re missing just one of those components, you’re going to sell yourself very short on the program and not be able to do a full analysis. Where I see most agencies fall is with the analysts. They can’t shortchange themselves on that. You can have the best data and tools in the world but if you don’t have the right person who knows how to ask the right questions, you’ll get nothing. Somebody has to know how to understand the answer and then dig deeper.” Malcom Bertoni, Food and Drug Administration: “You need to have champions both on the analytical side and on the program side—some data junkies who really love measuring and understanding and analyzing how an organization ticks, and some program managers on the front lines who get it, who are willing to embrace it and work with the analysts and improve their organization to make that part of their organizational culture.” Lisa Danzig, Office of Management and Budget: “Engage a set of people who think this could be worthwhile and/or already have a problem or goal they’re trying to achieve and that you could apply this to. This helps you avoid that cycle of collecting hundreds of metrics that aren’t relevant to the problem. It helps tie together the people who ultimately are going to be the advocates, who are the people with the problems and the goals.” Carter Hewgley, Federal Emergency Management Agency: “Find a champion at the leadership level in your organization. Assess the culture as it is. If people are not into it, you’re going to have to have a different strategy than if they are already on board. Then, you’ve got to demonstrate a quick win early. Pick a problem they care about and show them really quickly that, ‘Hey, if you did this differently you could save money or you could improve the quality of outcome for the people you’re trying to serve, or you could just make people’s lives easier.’” Gerald Ray, Social Security Administration: “The key thing is to put the data scientists with subject matter experts. You have to have someone who is very knowledgeable about your program so they can help the data scientists map through the issues you need to analyze. The data scientists are generally very good at doing the analysis themselves. But you also need the subject matter expert to tell them what part of what they’re finding is relevant and what’s not, and to guide them and change the direction to get it more on task and more appropriate for what you need.” Dean Silverman, Internal Revenue Service: “Agency leaders and analytics leaders need to learn how to experiment, or use what I would call test-and-learn techniques. I would aim at the hard problems. It sounds counterintuitive, but don’t be afraid to point combined operating IT and data analytic teams at big issues and keep them on a short development cycle. I’d make everyone focus on and measure outcomes, not outputs. Lastly, I’d own analytics at the highest level of the organization, especially if you want to create change.” Lori Walsh, Securities and Exchange Commission: “There are three fundamental pieces. First is having the right data available. Analytics can help fix holes in data but fundamentally, analytics requires good data. The second piece is the right computing infrastructure and tools, and more sophisticated processing of data. If you’re a nationwide program, you need a good network of computing capabilities so people can work together seamlessly as if they were next door to each other. The third piece of a good analytics program is subject matter expertise. You can do all the analytics in the world on all the data you want, but if you don’t have a focus on what you’re trying to find, you won’t be successful.” The IBM Center hopes that the results of our collaboration with the Partnership has helped agencies continue to enhance their ability to leverage analytics in a way that improves mission results.