If you’re running a self-funded plan, somebody has told you the spend problem lives in three places: specialty drugs, complex inpatient stays, and a handful of high-cost claimants nobody can predict. That’s not wrong. It’s just not the whole picture.
The line item nobody focuses on — because it shows up in small increments across hundreds of statements — is the one hemorrhaging the most. It’s primary care visit-driven spend. For a 350-employee plan running ~$14M annually, that line averages $2.6M per year — about $1 of every $5 in your medical spend.
And here’s the part the broker memo doesn’t include: most of that $2.6M isn’t paying for primary care. It’s paying for the infrastructure around the primary care visit.
Where your $14M actually goes
Composite breakdown for a 350-EE self-funded plan, drawn from KFF and MarketScan medians:
- Hospital + outpatient: $4.9M (35%)
- Specialty + Rx: $3.1M (22%)
- Primary care visits + downstream: $2.6M (19%)
- Admin / TPA / network access: $2.0M (14%)
- Wellness / preventive: $1.4M (10%)
Look at row 3. That’s the line we’re going to take apart.
What’s actually inside the primary care line
A single “routine” PCP visit, broken down by what your TPA actually pays:
- Office visit (E&M code): $187 — the 15 minutes with the provider
- In-office lab draw: $43 — drawing blood, not the labs themselves
- Send-out lab panel: $172 — the actual lab work
- Imaging referral: $90 — the order, not the imaging
- Specialist referral: $310 — routing fee + first specialist consult
- Pharmacy hand-off: $58 — PBM coordination
That’s $860 per visit. Multiply by ~2.1 visits per engaged employee per year (CDC + KFF averages), times a ~52% engagement rate on most self-funded plans, and that’s how 350 employees → 380 visits → ~$326K of medical-plan spend on what looks like “routine primary care.”
The catch: roughly 60% of those downstream charges — the referrals, the re-routings, the second-and-third hand-offs — are clinically discretionary. They happen because the visit fee structure rewards routing the patient along, not resolving the issue.
What the structural fix looks like
You don’t fix this by squeezing the TPA, switching networks, or layering another wellness vendor on top. None of those touch the underlying incentive — every visit gets billed individually.
You fix it by moving primary care out of the visit-fee chassis entirely.
The model:
- One on-site (or near-site) clinic, fully staffed by a dedicated nurse practitioner who knows your population
- Flat per-member-per-month fee paid by the employer, ~$95 PMPM in our composite (varies by geography and clinic configuration)
- $0 to the member at the point of care — no copay, no coinsurance, no surprise bills
- Bundled scope: annual physicals, sick visits, chronic disease management, behavioral health, labs, on-site pharmacy fills
Because the provider isn’t billing per encounter, there’s no incentive to route the patient downstream. The clinical resolution rate inside the clinic is dramatically higher because the NP has the chart, the relationship, and the time.
The 12-month math
For our composite 350-EE plan:
- Primary care visit spend: $326K → $0 (bundled). Delta: –$326K
- Downstream referrals + labs: $1.7M → $1.2M (–30%). Delta: –$510K
- Specialist + ER avoidance: $530K → $390K (–26%). Delta: –$140K
- On-site PMPM (350 × $95 × 12): $0 → $399K. Delta: +$399K
- Net 12-month savings: –$577K
A composite 350-employee plan saves roughly $1,650 per employee per year in the steady state — about 4.1% of total plan spend, or ~22% of the visit-driven outpatient segment.
Those aren’t the numbers a vendor promises. They’re the numbers that fall out when you take the visit fee out of the equation and watch what happens to the referrals downstream of it.
The 12-month timeline
What it actually looks like to launch:
- Month 0: Contract signed; site selection + build-out begins
- Month 1-2: Build, hire (or transfer) the NP, integrate with the EHR, set up the on-site pharmacy if applicable
- Month 2: Clinic opens. Initial enrollment burst — typically 35-45% of the roster verifies in the first 30 days
- Month 4: Visit volume stabilizes. Plan crosses break-even on a monthly run-rate basis
- Month 6: Engagement reaches 60%+; downstream referral reduction starts showing up in claims data
- Month 12: Steady state. 65-75% engagement, 18-25% reduction in visit-driven outpatient spend depending on population mix
We use month 4 as the break-even target. By month 12, the program typically returns 1.4-1.8× its annual cost.
Where this works (and where it doesn’t)
We’re explicit about who this fits because we’ve watched too many employers buy on-site care that didn’t survive contact with their population.
This works when you have:
- 100 to 1,500 employees. Smaller and the PMPM math doesn’t pencil; larger and a built-for-you direct-employed model becomes the right call.
- Self-funded or level-funded medical plan. The on-site model recovers spend the plan would otherwise pay; if you’re fully insured, those savings flow to the carrier, not you.
- One primary worksite — or a cluster of sites within ~30 minutes of one another. The model relies on the workforce actually being able to walk into the clinic during a workday.
- At least 40% of your workforce on the company health plan. Engagement scales with plan participation.
This doesn’t work when:
- You’re fully insured (no spend-side savings to capture)
- Your workforce is fully remote and distributed across 12 states
- You’re a Fortune-500 enterprise. We deliberately don’t serve this segment — the model legacy on-site vendors built for them is a different shape, and we’d be the wrong fit.
Want the projection for your specific plan?
Send us three things: roster size (and rough industry breakdown if multi-site), state of primary worksite, and self-funded / level-funded / fully insured status. We’ll send back a side-by-side projection of your current spend vs. the on-site model — within 2 business days. No deck, no sales call required to receive it. Decide whether the math is worth a 15-minute walk-through once you’ve seen the numbers.