Demand for data-driven models supplies agencies with decision making power

Agencies’ move toward making decisions based on data is turning the concept of supply and demand on its head. The more information senior leaders receive, the more data they want to make strategic decisions.

Take the U.S. Agency for International Development. It jumped into the data-driven decision-making pool more than 25 years ago when famine struck Ethiopia and the Sudan.

USAID relied on what many would say now were rudimentary satellite images to figure out exactly where it should send food and other aid to these nations, and why the famine occurred in the first place.

Gary Eilerts, the program manager for USAID’s Famine Early Warning Systems Network, said it took the agency years to get really good at using the data to make decisions. But now that it has, the agency understands its value.

“If there’s been one abiding feature of our program it’s that we’ve gotten better at asking questions. I think that is really an extremely difficult thing to do at the beginning. We didn’t know what questions we could ask,” Eilerts said during a recent panel discussion sponsored by the Partnership for Public Service. “We really only got fairly good at it about 10 years ago, so 19 years struggling to ask our questions. About 10 years ago, we figured that there is a better way to do this and it is by asking our people to help us solve our problems directly, rather than them delivering the products that they thought would help us.”


It’s that idea of asking good questions that is the lynchpin for changing the culture of the government. More and more agencies over the last decade have moved toward a culture of measuring program performance through data, and using those measurements to influence budget decisions.

As part of our special report, A New Era in Technology, Federal News Radio explores how data-driven decision making is getting to be the driving force behind spending and program strategies.

Getting better all the time

The concept of using data to make decisions isn’t new. Both the George W. Bush administration and the Barack Obama administration have spent a lot of energy on collecting, understanding and using data.

The Bush administration launched the Performance Assessment Ratings Tool (PART) to focus on large scale initiatives such as technology or human resources.

The Obama administration pared down the approach, asking agencies to focus on 3- to-8 high priority goals, and only introduced cross agency goals in the last few years.

But the goal of all of these efforts was to use data to make better programmatic and back-office decisions.

Dustin Brown, the acting associate director for performance and personnel at the Office of Management and Budget, said USAID’s experience with using data is becoming more common across the government.

“I think in the past a lot of the efforts focused on increasing the supply of performance information. What we really tried to do is operate from a somewhat different premise that really the key aspect to get right was increasing the demand for the use of data to inform decision-making,” he said. “If you get that in place first, if you get the leaders engaged on a regular basis, then the access to data and analytical tools really will come along. Essentially that’s what we are hearing from agencies now. The leadership is engaged and the challenge really has become making sure we get the metrics right and effectively for decision making.”

Brown said a real example of this change is OMB initially encouraged agencies to hold quarterly data-driven reviews. He said today about one-third of all agencies are holding monthly reviews of their programmatic information.

Interest coming from program side

Along with the top down push for more decisions based on data, other evidence shows there’s also a bottom up push from the program side.

The Partnership for Public Service recently pushed a second report looking at how agencies are using data to make decisions. Working with the IBM Center for the Business of Government, the Partnership highlighted case studies from five agencies, including USAID.

“One of the interesting commonalities among the five is they really are being driven by the program folks who have a problem they can see get better results by using data analytics,” said Max Stier, the president of the Partnership for Public Service. “That’s distinct from someone on top of the government thou shall use analytics. It’s really home-grown. I think these five are successful for that reason. The analytics are directly connected to a real problem that program folks have, and therefore they are making their lives better, they’re improving the mission’s impact and they’re using these tools because it has real effect on what they are trying to achieve.”

The Centers for Disease Control and Prevention and the Treasury Department are two examples of the bottom up and top down approaches. Both realized having more information was good, but being able to make sense of it was even better.

Capt. Christopher Braden, the director of the Division of Foodborne, Waterborne and Environmental Diseases, said the CDC moved to a data driven model in the 1990s during an e-coli outbreak in the Northwest United States.

Braden said the CDC started small but now collects data from 87 state health labs, and through new technologies, researchers look at microbes and can see the connections among the bacteria.

“This type of system allows us to identify a response situation, to identify it earlier and we can scale up depending on what type of organism it is and what kind of illness is occurring in the people. That’s a very proximate use of this type of data,” Braden said. “There are some other types of uses that are a bit more broad. For instance, if we do identify a type of food that is making people ill because this system identified the outbreak, there may be a gap in the nation’s food safety system that needs to be fixed. We can identify some of those gaps that may not be apparent or known to the food industry and that then has a real preventive effect on even the decision makers outside of government.”

Compare, contrast progress

Treasury applied a similar approach to the employee viewpoint survey data from the Office of Personnel Management. Nani Coloretti, the assistant secretary for management, said Treasury takes the information around employee satisfaction and uses it to make improvements across the department.

“We plow that into our management suite of metrics that we track. We basically have a conversation with our Treasury bureaus every quarter about that data. So we use that and there are several other indices that indicate how well a bureau and Treasurywide we are doing on job satisfaction, leadership, performance, talent management, innovation, employee engagement and diversity and inclusion,” she said.

Coloretti said Treasury is benefiting from the way OPM breaks down the data more broadly. A few years ago, OPM would give agencies a few hundred breakouts across agencies and the government at-large. Now OPM provides more than 13,000 different data views.

“OPM has gotten really good at standardizing some of these measurements so we can look at it for Treasury, by bureau and we can compare it to the federal government and other similarly sized agencies,” Coloretti said. “That has been very helpful to benchmark the data and then ask ourselves some hard questions about what we are doing to do about it. That comes out through an action planning process.”

Experts say there are several reasons why USAID, CDC, Treasury and a growing number of agencies are taking advantage of the data to make decisions.

One major reason is the data itself. It’s more plentiful and more easily accessible.

The second is the data analysis tools. Over the past decade, the market and tools have matured.

Understanding the fusion of data, tools

The Partnership’s Stier said the third reason is the people.

“What you see across agencies where this stuff is working is because once again, you have program people who are able to pull in technology as opposed to the technologist driving the actual changes in behavior,” he said. “You really want to have a more integrated approach here where you have the folks that are responsible for the programs becoming more sophisticated about the opportunities that technology provides them, and not just locating the technological expertise in a small number of folks that are not deeply integrated with the program work of the agency.”

Eilerts said USAID had to fight back the urge to depend almost solely on the data.

“There are a lot of people who would like to put all of our data sets, the satellite imagery, the price data, the nutritional data, the crop information, the weather, they’d like to put that into an algorithm because they think that is a very efficient way to do things,” Eilerts said. “We have tried to do that at various levels, at various time frames in the past, but it has never really produced the type of analysis that we think is sensible. So we have not gone into that as a primary means of making our analyses.”

Instead, USAID is developing a system that depends on the analysts looking at all the datasets at the same way in order to make decisions based on all information available.

OMB will continue to institutionalize the data driven approach through the people.

Under the Government Performance and Results Modernization Act (GPRA), agencies must develop strategic plans and update their strategic objectives.

Brown said agencies are in the early stages of that effort now.

“Over the next year we will be working with agencies a review process around those in a more systematic basis than what occurred in the past to inform budget decision making and really make sure we are bringing evidence, evaluation, findings and other sources of data and information to understand the extent to which we are really having an impact on those strategic objectives,” he said. “We are working with our Performance Improvement Council right now to put in place something that really is going to be meaningful and useful to influence that.”

Brown said the approach will be flexible based on the agencies’ needs. But OMB wants departments to use a set of central planning principles in developing the strategic plans and objectives.

He said agencies will do their initial assessments this spring, and OMB likely will make the plans public later in 2014.

Working group grows to 100

Another way OMB is seeing this approach take hold is through the PIO Council. Brown said the PIO Council created an internal agency reviews working group to share best practices and ideas for data-driven reviews.

“What we are seeing is a lot of interest in sharing those from one to the other. Even having agency staff sit in on each other’s reviews has been something that has been extraordinarily influential in the progress that we’ve been able to make,” he said. “Those sorts of tools are pretty adaptable from one agency to the next. We’ve seen a lot more similarities than differences in the ability to take some of these tools that are being used at one agency and apply them to the other.”

Brown said the working group started out with about 10 agency PIO staff members, but it has grown to more than 100 from 30 agencies. The working group meets on a monthly basis to share different approaches on performance improvement.

And as budgets continue to tighten, long-time feds stream out the door and the demand rises to prove a program works, experts agreed the reliance on data-driven decision making will only increase.


Could revival of influential tech office bridge Congress’ IT knowledge gap?

Major reforms needed to effectively manage federal IT

Panel: Digital technologies changing how agencies communicate — both inside and out

Mobile, telework create potential savings, increase productivity at agencies