It’s one part of a larger initiative in the IRS Procurement Office to promote data-driven decision-making.
Officials within the Internal Revenue Service's Office of the Chief Procurement Officer recently produced an interactive web application that employs a predictive model to foresee when contracts will be signed.
This new dashboard—deemed the Projected Contract Award Date—is already showing promising results at the IRS, which has roughly $2.5 billion in contracts each year, according to those who helped make it. And the technique the tool offers is applicable to other agencies.
“One of the biggest questions that program executives ask is when their contracts will be in place. Being able to predict these dates helps with operational planning,” an IRS spokesperson told Nextgov Tuesday. “Historically, staff would provide an expected date that was based on a ‘gut instinct.’ And in reality, this could be the separate gut instinct of hundreds of staff.”
With sights already set on modernization, the procurement team wanted to see if it would be possible to use data to create more consistency and accuracy within their estimates and to ultimately improve customer planning.
The Current Model
Work to develop this crystal ball for contract signing traces back to an earlier initiative that involved a use case to forecast award time.
Those participating in that effort leveraged data from the IRS’ internal contract writing system and broader federal procurement systems, the spokesperson noted. Their analysis revealed multiple factors that influenced the time it takes to reach an award.
“For instance, higher dollar values can require greater procedural reviews and more approvals that can lengthen processing time. But when it was submitted relative to fiscal year deadlines and employee workload also could impact time to award,” the spokesperson said. “The statistical analysis helped determine to what extent these various factors had on time to award.”
Machine learning algorithms were used to develop predictive models that could estimate award dates based on those relevant elements. After that, those involved were focused on how they might operationalize some of the insights gained and make them accessible to others as a part of the agency's broader data-driven decision-making aims.
“From that, the idea for a user-accessible web interface was born,” the official explained.
They confirmed it took about 11 months to conduct their initial analysis and develop the prototype application that exists for use today. A press release published by the IRS said the tool draws from historical data to gauge the number of days to contract award for requisitions in the IRS's Integrated Financial System—Procurement for the Public Sector.
“Our current model is showing we are able to predict the award date within +/- 30 days nearly 90% of the time,” the spokesperson said. “And we hope with greater data input over time that we can further increase the precision of these estimates.”
The expectation is that primary users of this dashboard would be those within the agency. An internal customer, for instance, could look up their in-process procurement request and get a projected contract award timeframe.
“We currently have this in a development environment and we’re working on transitioning this to an accessible user interface for IRS staff who could click on an internal website link to access it,” the spokesperson noted.
A Bird’s-Eye View
At this point, those involved are honing in on migrating the app out of its testing environment and into a more handy web version for insiders to use.
“We will be focusing on change management principles to increase adoption and gather feedback from users,” the spokesperson said. “Our goals are to continue to build trust in new applications and help make data more accessible and understandable to our partners.”
Those who helped make it have also had the opportunity to present the dashboard and relevant research to other federal audiences—most recently, at the Defense Department’s Acquisition Research Symposium.
“Overall, there’s a lot of excitement about the prospects of this tool,” the spokesperson said. They added that the best-case scenario outcome of this platform would be “for the federal acquisition workforce to see beyond just the current status of procurements and gain a bird’s eye view of what lies ahead.”
The predictive resource also results from an ongoing partnership between the IRS and Data and Analytic Solutions.
“DAS Inc. is conducting exploratory research on how data science techniques can support procurement success,” the spokesperson said.