Missouri Department of Social Services: Improving benefits access for residents

Challenge: Long wait times limited access to assistance

Like many traditional contact center operations, the Missouri Department of Social Security (DSS) was already dealing with a high call volume before the pandemic hit. The Family Support Division (FSD) saw a sharp rise in applicants for its Supplemental Nutrition Assistance Program (SNAP) with so many residents out of work. But applicants had to be certified through a mandatory interview process, and the program couldn't keep up.

"We saw a lot of families who had never been on public assistance before," says Nichole Conway, Program Manager at the Missouri Department of Social Services. "They don't need a difficult government process to frustrate them at their lowest economic point. The stresses they're going through are very, very real. FSD really wants to make sure that applying for benefits they need to survive isn't one of those stresses."

Under its legacy system, staff members had to follow up manually on requests for appointments. This required a lot of human effort to check schedules across approximately 125 local resource centers.

Solution: Virtual agents automate interview scheduling

After investigating a range of options, the state of Missouri partnered with Google Public Sector and Genesys to deliver a new intake and scheduling experience with an automated contact center solution that would help manage inquiries to FSD and schedule SNAP interviews.

New virtual agents were developed on Google Cloud using Dialogflow, Cloud Run, Cloud Firestore, Cloud Armor, and Cloud Load Balancing to automate the process of checking availability across all DSS resource centers. These virtual agents can find an appropriate time to schedule the required interviews, either in person or over the phone. Users can also return to the virtual agent to cancel or reschedule appointments if necessary. These appointments are then fed into the Genesys system to automate calls for the allocated time.

The virtual agents also offer answers to common queries about the application process, how to submit required information, and personalized insights on which policies may apply best to the caller's needs. At any time, residents can ask to have their call transferred to a live agent.

DSS also implemented Looker Studio and BigQuery for complete operation visibility, analytics, and reporting.

Results: Dramatically reduced wait times for citizens in need

Since the virtual agents went live, average wait times have dropped by 70%. "Once implemented, the positive surveys shot up immediately," says Conway. "Residents were so glad to have the opportunity and to actually receive the call at their appointment time. The amount of survey responses also increased. We've gotten a lot of good feedback from the community about this."

The virtual agents respond to 32,000 requests per month, with only around half choosing to continue through to a live agent.

"The rest are satisfied with the answers they get from the bot and don't need a live agent to intervene," says Conway. "Not only has that helped efficiency, it's also helped with staff retention. We have many staff on the phone all day, every day, and we can now rotate them between chat and the phones to help with the burnout of taking back-to-back calls."

Missouri DSS has achieved impressive initial results as it continues working to improve access to information and support for state residents in need of SNAP. The success to date offers an optimistic outlook as they explore further opportunities for virtual agents across government service contact centers.

Google Meet to enable virtual interviews

The state of Missouri aims to expand the use of Google Cloud technologies within its contact center systems. The state plans to use Google Meet to conduct virtual in-person appointments, and the team is exploring the use of AI-powered translation to further improve the accessibility of its virtual agents.

"If you had asked me, even five years ago, 'When would our agency be utilizing AI?,' I would have said, 'Maybe in 20 years,'" says Conway. "But we're doing it, and it's way easier than you think."