Commerce bots speed HR services delivery

The Commerce Department is using bots to cut the time it takes employees to correct inaccurate forms and process records from hours to minutes – a game-changer that’s giving employees in the Enterprise Services Division time to focus on more complicated customer service tasks.

That’s how Bill Daniels, federal lead for ES, described the robotic process automation (RPA) projects his division has undertaken. He spoke July 15 during a UiPath webinar.

“Our organization really looks for return on time, not necessarily return on costs,” Daniels said. As a  shared-services contact center, “time is of the essence in responding to our customers and having folks available to respond to customers,” he said. ES is also providing RPA as a service to the whole department, he added.

RPA is the next level of the basic automation that has been around for years. Think of the automated math that Microsoft Excel performs, for example, said General Dynamics IT’s Civilian CTO Mike Cole.

“Now I’m actually able to take a business process and automate that from start to finish,” he said. “I can have a bot that can do multiple functions. It can be looking for things. It can be attended or unattended.”

Commerce contracted with GDIT for help with building the backends of the automations it’s using, while Golden Key Group, Deloitte and IBM provide operational and other support.

One HR bot ingests 55,000 personnel records daily and processes them into ES’ service management system, which starts and ends employee services, updates information on employees, such as where they work, and drives workflows. The bot also downloads data via an application programming interface from the Treasury Department’s HRConnect and looks for those records with differences. This “delta run” takes about 45 minutes to execute, while a full load takes about two hours. The bot stitches the information together and loads it into Commerce’s staging environment inside a ServiceNow platform.

“It is the core foundation that makes our employee service center run on a daily basis,” Daniels said. “By 8:30 every morning, by the time the call center and the contact center is getting spun up, they have all the refreshes from any personnel data up and available to them before the first customer actually has called in.”

The ability to focus on exceptions is crucial, Daniels said. “We are investing in automations that allow us to put the routine stuff up on the shelf and let it run so that the team is really providing the customer service toward those cases that are exception-based -- the 20%, 30% where you [wonder,] ‘What is going on here?’” he said. Whereas many call centers have a goal of getting people on and off phones quickly, ES emphasizes engaging in meaningful conversations “because the routine stuff is taking care of itself,” he added. “The folks that we have in the call centers are now providing a much more personalized experience, which is what someone wants in an HR call center.”

The first bot ES built used screen-scraping technology to reduce the burden on employees who were reading service-request forms to make sure they were filled out correctly. One request can have five associated forms that must all have consistent data in fields such as name and contact information. The bot can ensure the quality and accuracy of these requests in about 45 seconds, highlighting any forms with mismatched fields so that they can be copied and pasted from the log file, put into ServiceNow, and returned to requesters to fix within minutes or hours – a process that used to take two to three days, Daniels said.

Currently, ES is working on a bot that loads information coalesced across multiple systems and platforms into its service management platform. “This bot verifies every ounce of that data, so it’s coming from one system, being downloaded, loaded into a staging environment, transformed and hitting target destinations in our service management system. This bot really quickly looks at that entire life cycle and validates that beginning to end,” he said.

The bot has looked through thousands of rows from a subsystem, verified the quality and will point to any failures in that process, allowing ES to quickly correct those exceptions.

“If we had a person doing this, it would take them three to four days to validate that data for one day,” Daniels said. “The bot does it for us in just over an hour and a half on a typical production run.”

Now that bots are handling a heavy processing load and ES has amassed “a ton of metadata,” he said, the division is exploring virtual agents, which can provide personalized service to contact center customers while freeing human call takers to address more complex inquiries.

“We have seen a tangible uptick in our customer satisfaction rates -- not quite a full percentage point, but when you’re looking at 55,000 customers, even a three-tenths of an uptick in customer satisfaction is a pretty big uptick in the sense of a six-month period,” Daniels said.

ES’ bot journey is not uncommon in the federal government, which is coming around to bots overall, GDIT’s Cole said. “I think we’re starting to see the government accept a lot more automation so they can use their resources more wisely,” he said.