Understanding Unwinding: What to Know About Medicaid Call Center Data

Later this summer, the Centers for Medicare & Medicaid Services (CMS) will likely release the first set of data reports related to the unwinding of the Medicaid continuous coverage provision. The first tranche of data is expected to include numbers on call center performance. People’s ability to get help from a state’s Medicaid call center, with minimal wait times and without giving up in confusion or frustration, is an important sign of how well an agency is serving Medicaid enrollees and applicants, particularly during unwinding. CMS is monitoring call center performance and may take enforcement action if it finds that enrollees can’t reach an eligibility worker to complete a renewal or other action.

Although states have been reporting some of this information to CMS for about a decade as part of performance indicator data, this is the first time these metrics will be publicly reported for all states. Federal law requires states to report to CMS, and for CMS to make public, call center data including call center volume, average wait times, and average abandonment rates each month from April 2023 through June 2024.

As health advocates, the public, and the press review these data, it’s important to consider them within the context of the features and setup of call centers in each state, including:

• Call center scope. Some states have dedicated call centers for Medicaid, while others have a single call center that covers multiple programs including Medicaid, the Supplemental Nutrition Assistance Program, and Temporary Assistance for Needy Families. Wait times from a call center that helps clients with multiple programs may be higher than Medicaid-only call centers because they are providing more comprehensive and complex services and may face greater variability in the length of each call.
• Which call center. Some states have multiple call centers, so it will be important to know which call center the data are from. In states where eligibility is determined at the county level, enrollees may have to call their local offices to renew Medicaid or get information about their case, which may not be captured in the state-reported data. Moreover, some states have vendors that handle a portion of Medicaid cases, while state staff handle others.
• Abandonment rate. Abandoned calls can be caused by several call center features that may be good or bad. On the negative side, if the automated system at the beginning of a call (known as interactive voice response or IVR) is long and confusing, people may hang up without getting what they need, or they may need to start over if the selection they chose didn’t lead them to their desired outcome. Or the IVR may ask for a Social Security number or case ID that the caller doesn’t have readily available, leading them to hang up. But call centers may also have positive features that lead to abandoned calls: for example, if the system tells the caller the expected wait time, the caller may choose to hang up and try back at a less busy time. Or a call center may offer a call-back option, resulting in a caller hanging up before they reach a worker and getting a return call when a worker becomes available.
• Misleading data. In some states, call center wait times won’t be representative of callers’ experience. For example, a state could have a worker quickly pick up calls, thereby keeping wait times low. But the worker may only be able to provide basic information and may have to route most calls to another queue for eligibility workers; that queue could have a substantially longer wait time that the data don’t capture.
• Trends over time. Though call centers have varying setups and service options, evaluating trends over time within a state can be informative. Did call volume increase the month after a significant number of Medicaid terminations, suggesting enrollees who lost coverage were calling to figure out what happened or to reapply? As unwinding progresses and more people are impacted, are wait times increasing, potentially indicating a deterioration in customer service?

• Resolution rate. When evaluating the CMS data on wait times, it is important to consider whether callers’ issues are resolved during their call. While resolution rates won’t be included in the CMS data, it’s important context for wait times. That’s because if workers take the time to fully meet callers’ needs — by completing an application or renewal while the caller is on the phone, or even making a collateral contact to verify an eligibility factor — they may handle fewer callers a day, driving up wait times, though the call center is actually providing better customer service.

  On the other hand, some call centers might not offer a Medicaid renewal over the telephone, even though that is a federal requirement, and may instead dispense with calls quickly by directing callers to a website or by telling them to call their local office. Low wait times paired with not resolving most callers’ needs should not be considered good customer service.

To better understand how callers’ issues are resolved and to put other call center metrics in context, stakeholders should gather additional information now on how call centers operate through user research with enrollees, conversations with the state, or test calls. With a better understanding of the call center context, stakeholders can better evaluate the CMS data when they’re released.

Focusing on a single data point without understanding the full context can lead to unintended results, such as lowering wait times without improving customer service. But evaluating these data within their context can help hold states accountable for providing effective customer service during the unwinding and beyond.