80.7% of appealed denials overturned · Only 11.5% of denials are ever appealed · Enforcement drift costs $50K–$575K per event · BH denial rates 2–3× higher than general medical · 46% of orgs use AI for RCM — denial rates still rising · Complex appeals take 3–6 hours without precedent · 80.7% of appealed denials overturned · Only 11.5% of denials are ever appealed · Enforcement drift costs $50K–$575K per event · BH denial rates 2–3× higher than general medical · 46% of orgs use AI for RCM — denial rates still rising · Complex appeals take 3–6 hours without precedent ·

Healthcare reimbursement appears rules‑based. In practice, it operates behavior‑based.

Same criteria. Same clinical scenario. Different outcomes — by insurer. We capture what actually works — and make it governable, by you.

Get Your Free Assessment See How It Works →
Overturned When Fought
80.7%
of appealed denials are ultimately overturned
KFF, ACA Marketplace Denial Analysis, 2024
Ever Appealed
11.5%
of denied claims are ever contested
KFF, ACA Marketplace Denial Analysis, 2024
The Gap
Recoverable
Revenue
abandoned, not lost — because the appeal process is too costly and inconsistent to pursue at scale
§ 01

Why appeals are won in the gray area

Clinical criteria are published, but enforcement varies by reviewer, plan, and context. Denials get overturned when your billers know how each insurer really behaves: "I just know how that insurer works."

Your team applies expertise and judgment case by case. We capture those Precedents as structured, reusable intelligence the whole team can use.

§ 02

Capture. Normalize. Govern.

Three phases that turn tribal knowledge into institutional intelligence.

01

Capture

We observe real reimbursement outcomes and structure them into Precedent Objects — which evidence combinations worked, in what sequence, for which payer.

02

Normalize

We map outcomes across payers so your team knows exactly how each insurer enforces criteria — not just what the guidelines say.

03

Govern

Precedents are human-readable, editable, and version-controlled. When enforcement shifts, your team flags it. The intelligence stays current.

See the full workflow →
§ 03

When payers shift enforcement, speed is the moat

AI retrains on statistical significance. We test hypotheses in real time.

Traditional AI Approach
2–6 mo.
To detect & retrain on enforcement drift
50–200+
Outcome-labeled examples needed
$50K–$575K
Cost per drift event
Stratum Approach
1–2 wk.
To detect, test, & update precedent
~10
Directed tests to validate hypothesis
$1K–$5K
Cost per drift event
§ 04

Built for behavioral health first

BH exposes the largest gaps between published policy and real enforcement — making it the ideal starting point for precedent-based intelligence.

Behavioral health denial rates run 2–3× higher than general medical, with enforcement patterns that vary dramatically by payer. The criteria are interpretive — "medical necessity" for PHP vs. IOP is a judgment call, not a binary threshold. And the billers who know how to navigate this carry it in their heads.

That makes BH the highest-leverage domain for precedent capture. The gap between "what guidelines say" and "what actually wins" is widest here — and the cost of not closing it is steepest.

Explore Behavioral Health →
2–3×
Higher denial rate vs. general medical
37%
High-volume payers: med necessity denial rate
$8,500
Avg. complex BH denial claim value
§ 05

Start with a free assessment

We'll review 25–50 of your recent denials and show where Precedent already exists — or where it needs to be captured. Diagnostic only — no precedent mining until you're ready.

✓ Free ✓ 5–7 business days ✓ No obligation

See where your revenue is hiding

Most providers discover 30–45% of their denials follow patterns their team has already solved — they just can't reuse the intelligence at scale.

Get Your Free Assessment
§ 06

Run the numbers yourself

Free calculators built on the same intelligence framework we use with clients. No email required.

§

KPI & ROI Calculators

See what denials cost your facility — and what precedent intelligence recovers.
∑

Unit Economics

Complex denials cost 10× more per recovered dollar. See the AI automation paradox.
?

Vendor Scorecard

6 questions that reveal whether your system builds institutional memory.
§ 07

Common questions

What exactly is a Precedent Object? +
Think of it as a structured record of what actually worked — which evidence combinations, in what sequence, moved outcomes for a specific payer and denial type. It's institutional memory, captured and reusable.
How long does implementation take? +
Fast. Sprint kickoff is 1–2 weeks, and you'll see first precedent captures within 30–60 days as we work cases together.
Is it HIPAA compliant? +
Yes — fully. All data handling follows HIPAA requirements with BAAs in place. Precedent Objects are de-identified by design: payer ID + denial code + clinical scenario + evidence elements + outcome. No PHI.
How are you different from AI RCM tools? +
AI RCM optimizes throughput on high-volume, low-complexity claims. We operate on the complex 30% that AI can't template — the interpretation-dependent, payer-behavior-driven denials where institutional knowledge determines outcomes. Explainability shows you what the model decided. Governance lets your team fix it when it's wrong.
See all FAQs →
Sources
[1] KFF, ACA Marketplace Denial Analysis — "High-volume payers reach 37% denial rates for medical necessity" (KFF, 2024)
[2] Experian, State of Claims 2025
[3] AMA Prior Authorization Survey, 2024
Stratum Collective — p. 01