Start with the right to the data
A self-funded plan's claims are the employer's data — and transparency rules from the Consolidated Appropriations Act strengthened the sponsor's right to receive them, detail included. If your TPA or carrier-ASO resists producing usable claims detail, that's a finding in itself (and a fiduciary concern).
You don't need a data warehouse to begin. A quarterly extract and a spreadsheet answer most of the questions that matter; what's usually missing isn't tooling but a short list of questions asked consistently.
The six numbers
- ER visits per 1,000 members per year. The single best signal of avoidable-utilization cost; national benchmarks often land near 150–200 per 1,000. Trend it, and read our avoidable ER visits math for what it's worth in dollars.
- High-cost claimant concentration. What share of total spend comes from the top 1% and 5% of members? Half or more from 5% is common. This drives stop-loss strategy and tells you where case management earns its fee — and where it can't help (stop-loss placement has to carry the rest.)
- Site-of-care spread. For comparable low-acuity diagnoses, what did the plan pay in the ER vs. urgent care vs. telehealth? The multiple you find is the business case for redirection.
- Out-of-network leakage — how much spend escaped the network (or the reference price), and from which facilities.
- Pharmacy's share and slope — total, specialty share, and growth rate; the on-ramp to a PBM review.
- Preventable-condition admissions — hospitalizations for conditions (diabetes, hypertension, asthma) that managed care should catch earlier; the signal that access, not acuity, is the underlying problem.
Turning numbers into decisions
Each metric maps to an intervention: high ER-per-1,000 argues for a genuinely usable after-hours alternative — the design bar we detail in telemedicine for employers; rising preventable admissions argue for access and follow-up, not penalties; heavy site-of-care spread argues for steering and $0-copay design; pharmacy slope argues for contract and specialty scrutiny.
Then close the loop: whatever you deploy, define in advance which of the six numbers it should move and re-pull them on a cadence. That's how a benefit like EZaccessMD prefers to be judged — model the expected claims impact with the ROI calculator, launch, then look for the ER-per-1,000 and site-of-care lines to bend in your own data.