Everyone’s talking about replacing people with AI. In healthcare, that conversation is loud—ambient scribes, denial automation, robotics in the OR. Wall Street is pricing in automation upside like it’s inevitable.
So when you model an RCM consolidation deal, the spreadsheet looks inviting:
“Consolidate vendors, centralize collections, automate workflows—margins expand 300-500 bps.”
By Year 3, you’re staring at a 400 bps margin collapse that wasn’t on the model.
Here’s the uncomfortable truth: You bet that technology would do the work that people actually do. In RCM, people are the margin—the ones who fight denials, manage payer relationships, understand local billing rules. You can’t algorithm your way around that.
And when you try? You discover the worst case: you still have the people, but now you also have expensive automation. The margins compress anyway.
The Data Point: One Platform’s Automation Bet
A recent diligence look at an RCM platform tells the story. Here’s what happens when a company bets big on automation without understanding what people contribute.
The margin trajectory:
| Year | Gross Margin |
|---|---|
| Year 1 | 76.7% |
| Year 2 | 73.7% |
| Year 3 | 72.6% |
| Projection | 66.0% |
That’s a 1,000 bps cliff over three years. And here’s what matters: This isn’t operational friction. This is the company’s choice to automate.
The R&D bet:
- 2023: $2.18M
- 2025: $7.57M
- Increase: 247%
They were building product tools to automate RCM, reduce billing specialists, improve denial management algorithmically.
The result:
- RCM margins compressed 600 bps
- Billing specialists? Still there (someone has to handle payer calls, appeals, edge cases)
- Engineer payroll? Now on the books too
It’s the textbook worst case: high opex, no margin improvement.
Three PE Diligence Mistakes
1. Denial Rate Assumptions
The model says: “Consolidate, improve denials from 8% to 5%, capture $50-100K savings per $1M ARR.”
Reality: Denials are payer-driven, not scale-driven. Medicaid, VA, Medicare Advantage each have different rules. Better documentation helps 2–3 points, not 5.
Here’s the critical insight: Reducing denials by 3–4% requires someone who knows the payer’s specific rules, can call the reviewer, provides context, and understands when to fight vs. accept. That person costs $40–80K annually.
The math: A 1% improvement = $50–100K in annual savings. You just spent that hiring the person who makes it happen. You need 3–4% improvement just to break even on headcount. It doesn’t happen.
The data confirms it: Denial rates are stable at 8–10% across the board, regardless of scale. Companies with lower denial rates aren’t using better technology—they’re using experienced billing specialists.
2. Customer Concentration Risk
This company’s customer base tells the story:
- Buying groups: 53.5% of RCM ARR
- One customer: The single largest buyer
- Top customer NRR: 128.7% (propping up blended metrics)
- Everyone else NRR: 91.3% (shrinking)
When the top customer “recontracted,” something revealing happened: PF adjustments reduced RCM ARR. They repriced downward to keep the customer.
Why? Because that customer finally understood: The company’s value isn’t the platform. It’s the people who manage RCM well. Once customers understand that, they have leverage.
The rest of the customer book is on a 91% NRR path. That compounds to 68% retained revenue by Year 3.
3. OpEx Eats the Gains
| Metric | 2023 | 2025 | Change |
|---|---|---|---|
| OpEx as % of Revenue | 57.4% | 65.8% | ↑ 820 bps |
| EBITDA Margin | 19.3% | 6.9% | ↓ 1,240 bps |
Revenue grows. OpEx grows faster. Why?
The company hired product teams to automate RCM margins while keeping the RCM operations teams. Automation didn’t work as expected, so you end up running two organizations:
- Engineers building half-working tools
- Billing specialists handling what the tools can’t
High opex. No margin improvement.
How to Stress-Test an RCM Deal
In RCM diligence, flip the standard script. Don’t ask “What’s the automation upside?” Instead:
1. Customer concentration risk
- Top 3 customers = % of ARR?
- If >50%, stress the repricing scenario. Assume -12% in Year 2.
2. Denial rate drivers
- Who manages appeals? If “the software,” expect margin compression.
- If “experienced specialists,” margins might hold.
3. Customer NRR reality
- Look at NRR by tier, not blended averages.
- If top customer is 128% and everyone else is 91%, you’re watching a two-tier business collapse.
4. RCM margins by payer type
- Medicaid: 68–72% (compliance overhead, people-dependent)
- Private Pay: 78–82% (lower-touch)
- VA: 70–75% (requires specialists who know VA systems)
- Medicare Advantage: 72–76% (each plan different; people navigate them)
The pattern: Margin depends on quality of people managing payer requirements, not scale or technology.
The Contrarian Insight
We’re in an era when enterprise software assumes automation replaces labor. Wall Street loves that narrative. But the RCM data tells a completely different story:
The platforms with stable or expanding margins are the ones that kept their people and used technology to support them—not replace them.
Three archetypes:
| Type | Margin | Trend | Outlook |
|---|---|---|---|
| Pure-play specialists | 78–85% | Stable | Scales within payer universe |
| Multi-payer, disciplined | 70–75% | Declining | Can stabilize at ~72% ceiling |
| Consolidators chasing scale | 65–72% | ↓ Accelerating | No path without model reset |
The pattern is clear: The winners aren’t the ones with the fanciest automation. They’re the ones who understood that in RCM, people create the margin.
What This Means for Deals
RCM is not a margin-expansion play. RCM is a people-dependent cash-flow extension play.
The value of RCM consolidation comes from:
-
Shortened DSO (45 → 35 days = $1–2M working capital freed)
- Created by specialists who know how to collect
-
Reduced volatility (predictable monthly collections)
- From relationships with payer contacts
-
Operational risk reduction (no vendor crises)
- From institutional knowledge of edge cases
All of it comes from people.
The Bottom Line
In an era when everyone is betting on replacing people with AI, here’s the truth:
RCM deals that work are the ones where you keep the people and use technology to support them. Deals that blow up are the ones where you assume technology is the upside and people are overhead.
This resets deal economics:
- Good RCM deals: 15–18% IRR (stable margins, real DSO improvement, people-centric model)
- Bad RCM deals: 8–12% IRR (margins collapse, top customer renegotiates, opex stays elevated)
One platform spent $7.57M on R&D to replace people. The margins compressed anyway. They still have the people.
That’s the margin cliff. And it’s the lesson.
Scott Decker is Managing Director of BluCascadia Advisors, where he leads due diligence and board strategy for healthcare IT investments. This analysis draws from diligence work across multiple RCM platforms and board governance at operating companies in post-acute and healthcare IT.