Employer Health Plan Denial & Appeal Outcome Vault

A crowdsourced, PHI-minimized database where US employees and benefits brokers log real denial reasons and appeal outcomes from employer-sponsored health plans, the two-thirds of privately insured Americans a new March 2026 federal transparency rule explicitly leaves uncovered, turning anonymized submissions into a searchable insurer/plan scorecard plus a paid broker report and a referral funnel into existing AI appeal-letter tools.

研究阶段进度

① 需求扫描

② 市场调研

③ 可行性分析

分诊打分总分: 27/35

市场调研评估

6.4/10

评估阐述

Scale 0-10, 0 = no real opportunity, 10 = large and undercontested. Demand side (documented, high confidence): 154M Americans on employer-sponsored coverage sit outside CMS-0057-F's March 2026 disclosure mandate, per direct-fetched ProPublica and KFF sources; KFF's fielded survey shows 21% of employer-insured people had a claim denied in the past year and 69% of those did not know they could appeal. Competition side (confirmed gap, real capital nearby): six named competitors checked directly (AuthDenied, Counterforce Health, Claimable, Aegis, Sheer Health, Fight Health Insurance); none combine employer-plan coverage, an insurer-level scorecard, and crowdsourced data, but Claimable ($10M raised including Mark Cuban) and Aegis (Y Combinator) show real venture capital is already circling the adjacent appeal-drafting layer and could extend into this gap. Score held below 7 by two factors sized transparently rather than assumed away: (1) bottom-up TAM/SAM/SOM funnels from KFF/DOL population data put the realistic 3-year combined SOM at roughly $115K-$330K/year, a real but modest paid market; (2) the core crowdsourcing-willingness assumption (will patients submit real denial letters into a shared pool) has no first-person evidence behind it in this or the prior research pass, and the closest comparable, Amino Health, needed $80M+ and a decade before finding sustainable distribution through brokers/employers rather than direct-to-consumer, which validates this idea's broker-channel plan but confirms this is a multi-year build, not a quick win.

可行性评估

淘汰

可行性评分4.5/10

评估阐述

Scale 0-10, 0 = no realistic path to a working business, 10 = clearly buildable, well-financed, low-risk path. Technical feasibility is strong (denial-letter extraction and search are a standard fit for current AI tooling) and the regulatory gap is confirmed with primary sources. The score is held below the pass bar because both revenue lines (a $9-15/month individual tier and a $500-2,000 broker report tier) depend entirely on one unproven assumption: that patients will submit real denial letters into a shared pool. No first-person evidence of that willingness exists in this research or the prior research pass, and neither the new federal disclosures nor the two states with partial employer-adjacent data actually cover self-funded employer plans, so there is no fallback data source if crowdsourcing fails. Financial modeling shows the individual tier's ~$18 average lifetime value cannot support any paid acquisition channel (LTV/CAC 0.08-0.24 against a $74-235 freemium-app CAC benchmark), forcing full reliance on organic search against incumbents that already rank; the broker tier is the only plausible path to break-even (LTV/CAC 0.83-6.25 depending on channel mix, roughly 19-52 reports/year needed to cover $19K-51K/year fixed opex) but a partner-led acquisition motion is required to clear a healthy ratio. Initial capital is estimated at $51,000-$134,000 (build, seed content, and specialized health-data-privacy legal review), all of which has to be spent before the core crowdsourcing assumption can be tested at real scale. Two funded competitors (Claimable, $10M raised; Aegis, YC-backed) sit close enough to add an insurer-level scorecard before this product proves out its data advantage, and the realistic three-year combined SOM ($115K-$330K/year) is modest relative to the capital and multi-year timeline required, echoing the roughly decade-long, $80M+ path a much larger comparable consumer health-data company needed before finding sustainable distribution. Ruled INFEASIBLE rather than FEASIBLE-with-caveats because the single biggest risk is existential to the entire model and cannot be cheaply falsified before most of the build budget is committed; a staged pilot (free seeded search tool first, measuring real submission intent before building the full crowdsourcing pipeline) could reopen this verdict if it produces positive evidence.

Employer Health Plan Denial & Appeal Outcome Vault

Track: Data Product (crossover with B2B2C / Supply to Audience Holders) | Market: overseas | Status: PENDING_RESEARCH | Created: 2026-07-20T00:00:00Z | Updated: 2026-07-20T00:00:00Z

One-liner

Discovery Method

Method: Trend Sniffer + Pain-point Extractor + Idea Generator
Signal (Trend): CMS's Interoperability and Prior Authorization Final Rule (CMS-0057-F) forced Medicare Advantage, Medicaid managed care, CHIP, and ACA marketplace plans to publish calendar-year-2025 approval/denial/appeal-overturn/turnaround metrics on their own websites, with first reports due March 31, 2026, about four months before this scan. The disclosure produced immediate follow-on activity: a free aggregator (AuthDenied) launched compiling over 1,200 plans' worth of the new data, and AI appeal-assistant startups pulled in real venture money in the same window (Claimable raised $10M including Mark Cuban, covered by Bloomberg in April 2026; YC funded a provider-side appeals startup, Aegis). Denial-rate transparency is a live, dated story right now, not a stale trend.
Signal (Pain): ProPublica's 2026 investigation found the disclosure mandate stops at ACA marketplace, Medicare Advantage, and Medicaid, covering under 10% of privately insured Americans, while the Department of Labor, which oversees roughly 2 million employer plans covering about two-thirds of insured workers, has never gotten a public denial-rate disclosure rule through (a 2016 attempt died in 2019 under insurer, employer, and union opposition). KFF's fielded consumer survey found 21% of people with employer-sponsored insurance had a claim denied in the prior year, and 69% of everyone who got a denial did not know they had a right to appeal. A KFF senior fellow put it directly in the ProPublica piece: "It's all knowable. It's known to the insurers, but it is not known to us."
Evidence: assets/evidence.md, direct-fetched quotes and figures from propublica.org, kff.org, authdenied.com, counterforcehealth.org, and getclaimable.com.

Demand Details

Who: roughly 155 million Americans get coverage through an employer plan, the segment CMS-0057-F does not touch. Two sub-audiences feel this directly. Employees who already have a denial in hand and no way to gauge whether their insurer's denial and appeal-overturn behavior is normal or an outlier before they decide whether to fight it. And benefits brokers or HR consultants advising employers during open enrollment, who currently have zero comparative denial-behavior data to put in front of a client choosing between TPAs.

What they want: the same thing ACA marketplace and Medicare Advantage enrollees got handed for free in March 2026, a way to check "how does this insurer actually behave on denials and appeals" before or after a bad decision lands. What AuthDenied built for the regulated plan types, this segment wants for the unregulated ones.

How they express it: nobody has built the submission side of this dataset yet. AuthDenied, the strongest existing aggregator, is explicit in its own FAQ that commercial fully-insured and self-funded employer plans are out of scope. Counterforce Health and Claimable both help an individual draft one appeal letter well, but neither shows an insurer-level track record anywhere in their product. The paper trail runs through investigative journalism (ProPublica) and a national health-policy survey (KFF) documenting the exact shape of the gap, rather than through first-person Reddit complaints, which stayed blocked to direct search this session as usual; that is treated as a medium-confidence gap on the crowdsourcing-willingness question specifically, not on the existence of the gap itself.

Monetization model:

Free: search aggregated denial and appeal-overturn stats by insurer/TPA, procedure category, and state, seeded at launch with the newly published CMS-0057-F figures and any state-published employer-adjacent data (Vermont, Connecticut) to avoid a fully cold start, then thickened by crowdsourced submissions over time.
Individual premium ($9-15/month): pull comparable overturned denials for your own exact procedure and denial-reason code once you upload your own EOB, the differentiated layer a generic AI appeal-letter tool cannot offer without the underlying dataset.
Broker/HR-consultant tier ($500-2,000/report): a denial and appeal-outcome scorecard across the TPAs an employer is evaluating, sold into open-enrollment RFP season, a comparison point brokers cannot get anywhere else today.
Referral/affiliate revenue: point users who want help acting on a denial toward Counterforce Health or Claimable rather than competing with them head-on; the vault is top-of-funnel, they own the drafting step.

7-Dim Triage Scores

Demand Pull 4 / Acquisition Feasibility 4 / Agent Advantage 4 / Low-Volume Economics 3 / Operator Hand Lightness 4 / Market Trend 5 / Policy Redline 3 -> Total 27/35

Score rationale:

Demand Pull 4: the data gap and its scale are documented by an investigative newsroom and a national fielded survey, both directly fetched, not inferred. Docked one point because the willingness of patients and brokers to actually submit real denial letters into a shared pool has no first-person confirmation obtained this session.
Acquisition Feasibility 4: long-tail SEO on "does [insurer] cover / deny [procedure]" queries has proven organic demand (ValuePenguin, MoneyGeek, and others already rank content here), patient advocacy communities and benefits-broker associations (NAHU and similar) offer warm channels, and referral partnerships with Claimable/Counterforce turn potential competitors into distribution partners.
Agent Advantage 4: extracting structured fields (insurer, plan type, procedure, denial reason, outcome) from unstructured denial letters at scale, and running natural-language comparable-case search over the resulting corpus, is a strong, direct fit for an AI pipeline, a notch above the manual-survey-style data entry underlying the two closest analogues in this library (freelance rate and streamer sponsorship vaults).
Low-Volume Economics 3: the free consumer search tier is close to worthless in any given insurer-plus-procedure cell until real submission volume accumulates, the standard data-product cold-start problem. Partially offset versus a pure-crowdsource model because the seeded public CMS-0057-F/KFF data and the broker-report tier both generate value at low volume.
Operator Hand Lightness 4: mostly a submission pipeline, extraction, aggregation, and search interface, with periodic moderation for spam or fabricated entries and legal review of the disclaimer language; no licensed-professional bottleneck in the core product.
Market Trend 5: rising on two independent fronts at once right now, a live regulatory disclosure event four months old and real venture funding flowing into adjacent AI appeal-assistant startups in the same window.
Policy Redline 3: not a banned category and existing competitors already operate appeal-assistance products without apparent legal trouble, but this sits next to health-data privacy law (state consumer-health-data statutes such as Washington's My Health My Data Act can apply to any entity that collects consumer health information, not only HIPAA-covered entities) and next to medical, legal, and insurance advice. Requires a PHI-minimizing architecture (discard the raw uploaded document after extraction, keep only de-identified structured fields) and a strict "informational only, not medical, legal, or insurance advice, verify with a licensed professional" disclaimer as first-class design requirements from day one, which is real, ongoing compliance overhead rather than a one-line disclaimer.

Downstream Hints

Key assumption to falsify first: will patients and brokers actually submit real denial letters and outcomes given health-data sensitivity, or does that sensitivity run higher than the freelance-rate and sponsorship-deal analogues already in this library? Check response rates on any existing patient-advocacy denial-sharing communities or Facebook groups as a proxy before committing to a full crowdsource-first build.
Closest competitors, all directly fetched and confirmed to stop short of this idea: AuthDenied (free, comprehensive, but explicitly limited to CMS-0057-F-covered plan types per its own FAQ, no employer-plan coverage, no crowdsourced data), Counterforce Health (free nonprofit appeal-letter generator, no insurer-level data), Claimable ($39.95 flat fee, $10M raised, no insurer-level data, individual appeal drafting only), Aegis (YC-backed, provider-side appeals, not a patient-facing denial-rate tool).
Compliance must-check for research: whether state consumer-health-data statutes (Washington My Health My Data Act and similar CA/CT/NV laws) apply to a non-HIPAA-covered entity collecting consumer-submitted health information, this is unverified and should be resolved before any build commitment, not assumed either way.
Compliance discipline carried forward regardless of the above: never retain raw uploaded EOB/denial documents beyond the extraction step, display only de-identified aggregates by default, explicit informational-only disclaimer at every point a user might read the output as advice.
Geographic scope: this is a US-specific opportunity built entirely around CMS-0057-F, ERISA, and DOL regulatory structure; it does not port to other overseas markets without a fresh regulatory scan.
Cross-track note: primary structure is Data Product (crowdsourced-plus-seeded dataset, same shape as lanes 6 and 68), with a secondary B2B2C angle through the broker/HR-consultant report tier, and a light SEO/content-asset dimension since individual denial-reason lookup pages should compound organic search traffic over time the way lanes 43 and 50 do. Track registered as Data Product; flag in research if the broker channel proves strong enough to warrant listing this as a dual-track lane going forward.

assets/ Evidence List

assets/evidence.md, direct-fetched quotes and figures from propublica.org (regulatory gap and DOL history), kff.org (2023 consumer survey denial/appeal-awareness stats), authdenied.com and its FAQ (CMS-0057-F scope and explicit employer-plan exclusion), counterforcehealth.org and getclaimable.com (closest competitors, both confirmed to lack insurer-level denial data).