Situation

This plan wasn’t caught flat-footed by the Medicaid unwinding. They had care managers, outreach processes, and a team that understood the stakes. What they didn’t have was a targeting system: a way to know which members were actually at risk before coverage lapsed.

Without one, the only option was to work on the entire member base. When state files arrived on time, the team could attempt broad outreach. When files didn’t arrive — which happened — there was no list at all. Members slipped through not because the team failed to act, but because the data needed to act precisely simply wasn’t there.

Compounding the problem: a portion of auto-assignments from the state were systematically under-delivered to this plan. The operations team sensed it. They had no way to prove it, and no leverage to act without evidence. The data required to solve this problem doesn’t live inside the plan. It lives upstream, in state systems no internal team is positioned to access.

Solution

Agilian’s data infrastructure was already running. State portal relationships, eligibility data patterns, and 834 cross-reference logic were in place. With no state data discovery required, the plan could get insights from Agilian fast.

From receipt of the first 834 files, Agilian delivered a working Smart List within 72 hours. Full engagement was live within two weeks. The care management team went from all-or-nothing outreach to a prioritized, deduplicated list of exactly which members needed attention, segmented by 0/30/60/90-day urgency tier, before the reinstatement window closed.

In parallel, Agilian analyzed state reference files and confirmed what the operations team had suspected: the plan was receiving fewer auto-assignments than its member composition warranted. Armed with that evidence, the plan went directly to the state and had the assignment imbalance corrected. That recovery wouldn’t have been possible without independent data no internal team was positioned to surface.

Results

What the pilot really delivered was more than 673 members. It delivered the pattern recognition, the confirmed root causes, and the decoded eligibility data relationships that became the foundation for everything that followed.

The pilot was designed to prove the model, not maximize volume. It did — and then some. The insights that emerged in six weeks became the foundation for everything that followed and allowed the plan to scale results when the MVP launched.

See what happened when the program scaled. →