ShortPay — underpayment auditor for independent specialty practices

1. One-liner

ShortPay reads every 835 against the payer contract and claws back the claims insurers quietly short-paid.

2. Trend signal — why now?

Underpayment is the revenue leak nobody is watching. Denials get screamed about; underpayments slip through because the claim got paid — just not in full. The numbers are not small:

• MD Clarity / industry data: 1.8–3.4% of paid claims carry an unrecovered payer underpayment; a 3% rate on a $10M practice ≈ $300K/year evaporating silently.
• Multi-specialty groups lose $150K–$400K/year to revenue leakage hidden in denial write-offs, underpayments, and missed charge capture (Medical Billers & Coders, 2026).
• Tebra / Aroris: “most independent practices have at least some services priced below their contracted allowable rates without realizing it” — when payers update fee schedules but never align claims logic, providers get short-paid with no explanation.

The enabling shift: payer contracts and 835/ERA remittances are now machine-readable end-to-end. LLMs can extract a structured fee schedule (DRGs, carve-outs, lesser-of provisions, escalators) out of a messy contract PDF in minutes — work that used to require a human contract analyst billed hourly. CMS’s 2026 prior-auth transparency rules (every denial must carry a specific reason) are pushing the whole industry toward line-level payment scrutiny. The tooling to detect underpayments at the line level was, until ~12 months ago, only economical for $1B health systems running R1/Cloudmed.

Provenance:

• Signal 1 (demand): “Most independent practices have at least some services priced below their contracted allowable rates without realizing it” + $150K–400K/yr leakage — https://www.tebra.com/theintake/getting-paid/rcm-and-claims/how-to-reduce-medical-billing-underpayments , https://www.medicalbillersandcoders.com/blog/is-your-practice-losing-revenue/ — 2026-06-16
• Signal 2 (feasibility): 835/ERA is a fully structured EDI transaction (CO-45, allowed amount, CARC) and LLMs now extract structured reimbursement terms from contract PDFs cheaply — https://arrowhq.com/blog/835-era-file-guide , https://www.mdclarity.com/revfind — 2026-06-16
• Signal 3 (economic): Underpayment recovery is a real outsourced industry — R1/Cloudmed is “the largest and longest-running outsourced underpayment recovery program in the country,” MD Clarity sells RevFind on contingency — but built for health systems, not 1–10 provider practices — https://www.r1rcm.com/solutions/underpayment-recovery , https://www.mdclarity.com/comparison/best-healthcare-underpayment-recovery-services — 2026-06-16 Category: Geographic arbitrage (enterprise tool → SMB-affordable tier)

3. The opportunity

The incumbents — MD Clarity’s RevFind, Waystar, FinThrive, Experian, R1/Cloudmed — are real and good. They are also built, priced, and sold for hospitals and large groups. Their reference cases are “$1.8B health system,” “$8M in missed revenue,” “$25M total recoveries.” A solo gastroenterologist or a 4-doc orthopedic group doing $4M does not get a call back from R1, and could not stomach the implementation if they did. So the small practice does one of two things: nothing (eats the leak), or pays an outsourced biller who “doesn’t have the time and bandwidth to chase” — the incumbent’s own marketing copy admits this is why practices give up.

What they all do badly for the small end of the market:

• Onboarding assumes a contract-management team. ShortPay needs one thing: upload your payer contracts + connect your 835 feed. The LLM does the contract-to-fee-schedule extraction the enterprise tools make you (or a consultant) do by hand.
• Pricing assumes a six-figure book. A $99–$399/mo or modest contingency tier is invisible to enterprise vendors and perfect for a practice clawing back $20K/year.
• They flag; they don’t fight. ShortPay drafts the appeal letter with the exact contract clause and CARC cited, ready to send. That last mile is where small practices stall.

10× better isn’t a smarter algorithm. It’s “a $4M practice gets enterprise-grade underpayment detection in an afternoon for the price of a streaming bundle.”

4. Target market

• Primary customer: Owner or practice manager of an independent single-specialty practice, 1–10 providers, in a procedure-heavy specialty where contracted rates and underpayments bite hardest — orthopedics, gastroenterology, dermatology, ophthalmology, cardiology, pain management. US-only at launch.
• Why they buy: “Payers happily pay us less than they owe and we never catch it.” Owners feel the squeeze directly — independent practices already get reimbursed less than half of hospital-owned rates for the same service, so every short-paid claim on top of that is intolerable. They have no contract analyst and their biller is buried in denials.
• Rough TAM reasoning: ~395K active physician group practices in the US; ~42% private/independent; ~37% single-specialty. Conservatively 40,000–60,000 small independent specialty practices in procedure-heavy specialties. Add the secondary channel: thousands of third-party RCM/billing companies (Kareo/Tebra-adjacent, DrChrono billers, regional shops) who could white-label ShortPay across their whole book. At even $200/mo, 50K practices is a $120M ceiling — far past the $5M target.
• Why now for them: Margin compression is acute, hospital-buyout pressure is at record highs (private-practice share fell from 60% in 2012 to 42% in 2024), and CMS 2026 reforms force payers to itemize reasons — making line-level underpayment detection both more urgent and more tractable.

5. Product sketch (MVP)

• Contract ingest: upload payer contract PDFs; AI extracts the fee schedule — allowed amounts by CPT, lesser-of provisions, carve-outs, escalators, timely-filing limits — into a structured rate table the owner can eyeball and correct.
• 835/ERA feed: connect remittance files (clearinghouse export or direct 835 drop). Each paid line gets matched to its contracted rate.
• Underpayment ledger: every claim where paid < contracted (net of legitimate CO-45 contractual allowance) surfaces in a worklist, ranked by dollar value and recoverability.
• Pattern alerts: “Aetna has been paying CPT 45378 12% under contract since March” — systemic underpayments, not just one-offs.
• Appeal drafter: one-click letter citing the specific CPT, contracted rate, CARC, and contract clause — formatted to the payer’s appeal channel.
• Fee-schedule guard: flags your own charge master line items priced below contracted allowable (the silent self-inflicted leak).
• Recovery dashboard: dollars flagged, dollars recovered, yield by payer — the number the owner screenshots for the next contract negotiation.

6. AI angle — what’s load-bearing

Remove the AI and this is a spreadsheet nobody fills in — which is exactly why the leak exists today. The AI does the two jobs that previously required paid human specialists: (1) turning a non-standard, payer-specific contract PDF into a clean structured fee schedule (the bottleneck that makes enterprise tools need a consultant), and (2) reading every 835 line, reconciling it against that schedule net of legitimate adjustments, and writing the appeal in payer-ready language with the right clause cited. Contract extraction across hundreds of payer formats is genuinely an LLM problem — rules engines have failed at it for a decade, which is why this stayed an enterprise-services business. AI is the reason a 2–3 person team can now offer at $200/mo what used to cost a contingency fee on a hospital’s book.

7. Localization angle (if any)

N/A — this is a US-only play by design. The wedge is the US payer-contract system (commercial fee schedules, 835/ERA EDI standard, CARC/CO-45 codes, the appeal-rights regime). That specificity is the moat, not a limitation. No India/LatAm version makes sense; single-payer and other systems don’t have the contracted-rate-vs-paid-rate gap that defines the product.

8. Business model — path to $1M–$5M ARR

• Pricing: Two tiers. Self-serve SaaS at $149–$399/mo per practice (by provider count / claim volume). Performance tier for the skittish: free detection + 15–20% of recovered dollars for the first 6 months, then convert to flat SaaS. RCM-biller channel: volume license, e.g. $49–$99/practice/mo across their book.
• ACV: ~$3,000/yr blended (mix of $200/mo SaaS and contingency converts).
• Rough math to $1M ARR: ~340 practices × $250/mo × 12 = $1.02M. Or land 6–8 RCM billing companies each running 60+ practices.
• Rough math to $5M ARR: ~1,400 practices direct, or a dozen mid-size RCM partners white-labeling across 100+ practices each, plus an enterprise tier for multi-site groups. Entirely inside the 40K–60K TAM.
• Expansion path: start with underpayment detection → add denial-prevention, contract-negotiation benchmarking (“you’re 9% under market on CPT X across all payers”), and charge-capture leak detection. ACV grows as ShortPay becomes the practice’s revenue-integrity layer.

9. Go-to-market wedge — first 100 customers

• Contingency pilot, named list: pull 500 independent orthopedic + GI practices from NPI/CMS provider directories and state medical board lists. Offer a free 60-day underpayment audit: “We’ll find what Aetna short-paid you last quarter; you pay 20% of what you recover, nothing if we find nothing.” Risk-free framing closes a money-losing owner fast. Expect 8–12% to take a free audit, half to convert.
• RCM-biller channel: the highest-leverage move. Sign 5–10 small third-party billing companies (the DrChrono/Kareo-adjacent shops) who already hold the 835 feeds for dozens of practices. ShortPay makes them look good and adds a revenue line — one signed biller = 40–80 practices.
• Specialty MSO / association angle: procedure-heavy specialties have buying groups and state societies (AIMPA advocates for 14,000+ independent physicians). One webinar + a member discount puts ShortPay in front of exactly the right owners.
• Founder-led proof: run the audit by hand on the first 10 design-partner practices, publish anonymized “we found $X in 90 days” case studies — the only marketing a skeptical practice owner trusts.

10. Build complexity — justification

Medium. Off-the-shelf: LLM contract extraction, 835/EDI parsing libraries (the format is standardized), standard web stack, a BAA-covered cloud. Custom work: the reconciliation engine (paid vs contracted net of legitimate adjustments — must be precise or it cries wolf), the contract-extraction QA loop (owners must trust the rate table), and HIPAA operational hygiene (a config and discipline burden, not a pre-launch approval). A 2–3 person team ships a credible v1 in ~3–4 months, faster if the first version is concierge (founder runs the audit, software catches up).

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Reviewing your own claims data; appeals are a contractual right. BAA required, no license.
Ethical — no harm / dark patterns	✅	Recovers money legitimately owed under contract. Pro-provider, anti-leak.
Market exists (evidence above)	✅	Quantified leakage, existing enterprise vendors, outsourced recovery industry.
1–5 person team can build this	✅	2–3 people, ~3–4 months to v1.
Launchable with <$50K / ₹40L	✅	Off-the-shelf APIs + cloud; concierge-first keeps build cost low.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	17/20	Direct, recurring, quantified money loss owners can’t see. Hair-on-fire once shown the number.
Demand evidence	15	12/15	Multiple independent sources, funded enterprise incumbents, an outsourced recovery industry. Direct SMB-tier demand still to be proven.
Build feasibility	15	10/15	Standard stack + EDI, but reconciliation precision and contract-extraction QA are non-trivial; HIPAA hygiene adds discipline.
Distribution clarity	15	11/15	Named lists + contingency offer + RCM-biller channel. Healthcare sales is slower than the offer implies.
Revenue mechanics	15	12/15	Pricing benchmarked to incumbents; contingency lowers buying friction; clear customer count to $1M.
Time to first revenue	10	7/10	Contingency pilot can pay inside 8 weeks; full SaaS conversion is slower.
Defensibility	10	5/10	Execution + accumulating contract-extraction library + biller-channel lock-in. Copyable, but the small-practice niche is unglamorous and incumbents won’t chase it down-market fast.
Total	100	74/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required — needs someone who can build precise reconciliation logic AND someone who genuinely understands payer contracts, CARC/CO codes, and the appeals process. A technical founder + an RCM/billing-veteran advisor is the ideal pair.

Key assumptions to validate (3–5)

1. Assumption: Small specialty practices have enough recoverable underpayment ($10K+/yr) to justify paying. How to test: Run free manual audits on 10 design-partner practices; measure recoverable dollars found.
2. Assumption: AI can extract a usable fee schedule from real-world payer contract PDFs at >90% line accuracy. How to test: Feed 25 actual contracts across 5 payers; have an RCM expert grade the extracted rate tables.
3. Assumption: Practices will connect their 835 feed / hand over remittances to a young vendor. How to test: Track how many of the first 30 prospects actually complete data connection vs. stall on trust/security.
4. Assumption: RCM billing companies will adopt rather than view it as a threat to their own value. How to test: Pitch 10 small billers; measure interest in white-label vs. defensiveness.

Risk flags

1. Sales-cycle risk: Healthcare buyers move slowly and security-review even small vendors. The contingency offer is the antidote but doesn’t fully erase it.
2. Incumbent down-market move: MD Clarity or Waystar could ship an SMB tier. Mitigant: they’re structurally focused on health systems and contingency on big books; the small-practice niche is unattractive to them — for now.
3. Accuracy / trust risk: A reconciliation engine that flags legitimate contractual allowances as underpayments destroys credibility instantly. Precision is existential, not a nice-to-have.
4. Platform/data-access risk: Reliance on clearinghouse/835 export pathways; if a major clearinghouse closes access, onboarding friction rises.

14. Structured verdict

Score:                  74/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder + RCM/payer-contract domain advisor
Time to revenue:        8–12 weeks (contingency pilot); 4–6 months (SaaS conversion)
Capital to launch:      $15–30K (₹12–25L)
Top 3 assumptions to validate first:
  1. Recoverable underpayment per small practice ≥$10K/yr — manual audits on 10 design partners
  2. AI fee-schedule extraction ≥90% accuracy on real contracts — expert-graded test set of 25 contracts
  3. Practices will connect 835/remittance data to a new vendor — track data-connection completion rate on first 30 prospects
Kill criteria:
  - Abandon if median recoverable underpayment across 10 audited practices is <$5K/yr (no ROI to sell against)
  - Abandon if contract-extraction accuracy can't clear ~90% after tuning (false positives kill trust)
  - Abandon if <2 of first 10 RCM billers show channel interest AND direct data-connection completion stays <30%

15. Next step — 1-week validation sprint

• Day 1–2: Recruit 3–5 independent specialty practices (orthopedics/GI) via warm intros or an RCM-biller contact. Get them to share one quarter of de-identified 835 remittances + their top-2 payer contracts.
• Day 3–4: Manually (with LLM assist) extract the fee schedules and reconcile that quarter’s paid lines against contracted rates. Tally recoverable underpayment dollars per practice.
• Day 5: Decide go / no-go on a falsifiable bar: median recoverable underpayment ≥ $2,500 per practice per quarter (≈$10K/yr) across the cohort, AND at least 2 of the 5 owners say “yes, send the appeals and take a cut.” If the dollars aren’t there or owners shrug, kill it — the whole thesis rests on visible, recoverable money.