Oregon's Coordinated Care Organizations got a 3.4% average rate increase in 2024. Their per-member costs grew by more than 10%. The result: an average net operating margin of essentially zero — 0.001% statewide — with seven of the sixteen CCOs posting outright losses. This isn't an Oregon story. It's a national pattern playing out in nearly every Medicaid market. The Milliman data is stark: across 184 Medicaid MCOs nationally, aggregate underwriting margins went from +2.4% in 2023 to -1.0% in 2024 — a $9.9 billion swing in a single year. More than half of all plans lost money on their Medicaid book. If you're running a small or mid-sized Medicaid plan right now, you already know this in your bones. The question isn't whether the pressure is real. The question is what you can actually do about it.
You cannot renegotiate your capitation rates on a Tuesday afternoon. You cannot change your member population's acuity. You cannot restructure your provider network fast enough to move the needle this fiscal year. What you can control — often with more impact than plans realize — is how efficiently your back office operates.
Where Administrative Cost Actually Lives
The first problem with most administrative efficiency discussions is that they stay at the level of the org chart. "Reduce headcount." "Consolidate functions." That's not a strategy. It's a budget cut dressed up in management language.
Real administrative cost lives in specific workflows. Here's where to look.
Claims Touches
Every time a claim requires human intervention — a human opens it, looks at it, routes it, corrects it, or makes a decision about it — that's a "touch." Each touch costs money. For a plan processing 500,000 claims annually, reducing average touches from 1.8 to 1.2 per claim is not a rounding error. At even $8–12 of labor cost per touch, the math gets significant fast.
The plans with high touch rates share common root causes: incomplete or stale provider data causing routing failures, authorization mismatches that create pend queues, member eligibility that isn't reconciled in real time, and claims editing logic that hasn't been updated to reflect current contract terms. None of these are glamorous problems. All of them are fixable.
Enrollment Churn
Medicaid populations churn faster than any other market segment. Members gain and lose eligibility constantly — and every eligibility change creates downstream administrative work if your enrollment reconciliation isn't running continuously.
The cost of enrollment churn isn't just the enrollment transaction itself. It's the claims that adjudicate against the wrong eligibility status. It's the member services calls from people who think they're covered but aren't — or vice versa. It's the encounter data submissions that reflect a member who shouldn't have been on your roster. Phantom member problems — where a member has technically disenrolled but hasn't been cleanly removed from your operational systems — are more common than plans admit, and they propagate errors across claims, analytics, and quality reporting.
Plans that treat enrollment as a solved problem and batch-reconcile weekly or monthly are carrying this cost invisibly. The work still happens — it just shows up as downstream exception handling, rework, and corrected submissions.
Call Volume as a Cost Symptom
Your member services call volume isn't just a staffing problem. It's a diagnostic signal. High call volume on eligibility questions usually traces back to enrollment reconciliation gaps. High call volume on claims status usually means you have a claims pend backlog or a slow adjudication cycle. High call volume on referrals or authorizations often means your member-facing materials are out of date or your network directory has errors.
Most plans look at call volume and ask: how do we answer these calls more efficiently? The better question is: why are members calling in the first place? Resolving the upstream issue — the enrollment error, the claims delay, the stale provider data — eliminates the call entirely. That's a different math than hiring faster call center staff.
Manual Workarounds
Every manual workaround in your operation has a cost that's rarely measured directly. A staff member who exports data from your claims system into Excel to do a calculation that should happen automatically. A monthly report that requires someone to pull from three systems and reconcile by hand. A state submission that requires manual reformatting because your platform doesn't natively support the required format.
These workarounds accumulate over years. They're rarely documented. When the person who built them leaves, institutional knowledge walks out the door. And they consume staff time that could be spent on work that actually requires human judgment.
Benchmarking: There's More Variation Than You Think
Here's something that often surprises plan leaders: the administrative cost variation across similarly-sized Medicaid plans is substantial. The Medicaid MLR framework requires plans to spend at least 85% of capitation revenue on clinical services — leaving 15% for administration and margin. But what plans actually spend on administration within that envelope varies widely.
Composite data across Medicaid MCOs for 2023 showed an average Administrative Loss Ratio of 7.9%. But averages obscure a wide range. Plans running lean, well-integrated administrative operations can land materially below that figure. Plans with fragmented systems, heavy manual workflows, and high staff-to-member ratios can run substantially above it — sometimes while also delivering worse service quality, because the administrative burden is distributed inefficiently rather than effectively.
If your administrative costs are above 10% of premium equivalents, that's worth examining seriously. If you don't know where you land, that's the first problem to solve.
Automation vs. Staffing: The Size-Dependent Tradeoff
Small plans — say, under 40,000 members — face a specific dilemma. The automation investments that make large plans efficient (a mature claims editing engine, a real-time eligibility integration, an automated encounter data validation pipeline) require upfront capital and technical capacity to implement and maintain. You may not have the IT team to build and sustain them internally.
This creates a real temptation to staff around the gaps instead. Hire someone to manually check eligibility before each claims run. Hire someone to manage the encounter data exception queue. Hire a coordinator to handle the state reporting format conversions. These are real jobs, and for a time, they work.
The problem is that staffing around system gaps doesn't scale, and it doesn't improve. As your membership grows, the manual workload grows with it. The automation investment required to eliminate those roles doesn't get smaller — it gets more complicated because you've built operational habits around the workarounds.
The question for plans under 100,000 members isn't whether to automate versus staff. It's whether to build the automation internally, buy a point solution, or operate on a platform that handles it as part of an integrated back-office model. The honest answer depends on your growth trajectory, your internal technical capacity, and how many of these problems you're trying to solve simultaneously.
What to Stop Doing
Sometimes the most valuable administrative efficiency move is subtraction. A few common candidates:
Stop running eligibility reconciliation on a weekly batch cycle. Real-time or daily reconciliation costs more to implement once but eliminates a class of downstream rework that's expensive and ongoing. The break-even point arrives faster than most plans expect.
Stop manually building state submission files. If your team is spending hours each month reformatting encounter data, prior authorization reports, or quality metric submissions to match state specifications, you're paying for a translation layer that should be automated. This is solved infrastructure, not custom work.
Stop treating your claims pend queue as a normal operating condition. A large pend queue is a symptom of upstream data problems — provider file gaps, authorization mismatches, coordination of benefits issues. Clearing pends manually is treating the symptom. Finding why claims are pending in the first place is treating the disease.
Stop accepting that your member services team will "just figure it out." When member services staff are navigating multiple systems, interpreting inconsistent data, and improvising because they don't have clean eligibility and claims information at their fingertips, you're paying for their time while also delivering worse member experience. That's the worst of both.
Identifying Your Highest-Cost Manual Workflows
The practical starting point for any administrative efficiency effort is a workflow audit — not a high-level process map, but a ground-level inventory of where staff time actually goes.
Ask your operations leads to track, for two weeks, every task that involves opening more than one system, copying data between systems, reformatting data for a submission, or manually reviewing exceptions that should have been auto-adjudicated. You'll find patterns quickly. The highest-cost manual workflows almost always cluster around the same root causes: data that doesn't flow cleanly between systems, eligibility information that isn't current at the point of claims adjudication, and submission requirements that your platform doesn't natively support.
These are solvable problems. Not overnight, and not without investment — but with a clear return, because the manual cost is recurring and the fix is one-time (or at least amortized over years).
The plans that are navigating this margin environment without cutting clinical programs or member services quality are largely the ones that did this audit two or three years ago and made the infrastructure investments. The plans that are cutting corners now are the ones that didn't.
If you're trying to understand where your administrative cost is concentrated and what it would take to address it, Ayin works with small and mid-sized Medicaid plans on exactly these problems — enrollment, claims, encounter data, and the integrations that tie them together.