The Supreme Court Voided $166 Billion in Tariffs. 330,000 Importers Are Owed Refunds, and 6.1 Million Entries Already Failed.
On February 20, 2026, the Supreme Court struck down every tariff imposed under IEEPA. CBP collected about $166 billion from more than 330,000 importers across 53 million shipments, and none of it comes back automatically. The refund requires filing through a portal where 6.1 million entries have already failed validation. The biggest importers hired counsel in April. The long tail hasn't filed, and some of its deadlines are already gone.
The Problem
Start with the number: $166 billion, which is what U.S. Customs and Border Protection assessed between February 2025 and February 2026 in tariffs imposed under the International Emergency Economic Powers Act, paid by more than 330,000 importers across 53 million shipments. Then, on February 20, 2026, the Supreme Court ruled 6-3 in the Learning Resources case that IEEPA never authorized the president to impose tariffs at all. Every fentanyl-related duty, every reciprocal tariff, every Brazil-action surcharge collected under that statute became legally void in a single opinion. Trade lawyers call it one of the most consequential trade rulings in a generation. They are right.
Here is the part that matters for this business: the money does not come back on its own, because refunds are paid to the importer of record, and only if someone files. CBP built the Consolidated Administration and Processing of Entries (CAPE) system inside its ACE Secure Data Portal, launched April 20, 2026, and the filing process is exactly as friendly as a government acronym suggests. The importer needs an ACE Portal account, which takes three to four weeks to set up, ACH enrollment for electronic refunds, and a CSV listing every entry number being claimed, capped at 9,999 entries per declaration, and there are no paper checks. CBP runs two rounds of validation, strips the IEEPA provision from each entry summary, and pays by ACH in 60 to 90 days after acceptance, with interest accruing at 6 percent for corporations and 7 percent for individuals, compounded daily.
As of September 11, 2026, CBP had accepted $134.7 billion in refund claims and certified roughly $122 billion, interest included, to the Treasury for payment, according to the agency's court declaration. That is 81 percent of the pool by value, which sounds like the story is over until you read the rest of the declaration: importers had submitted about 33 million entries through 286,000 declarations, and 6.1 million entries failed validation. It isn't over. Nearly one in five submitted entries hit a wall, most commonly for falling outside CBP's 90-day reprocessing authority, missing the IEEPA Chapter 99 number, or appearing on an earlier declaration, and another 20,184 refunds totaling $1.3 billion sat on hold for the most mundane reason in the file: the importer never enrolled in ACH.
The risk is the clock: CAPE's first two phases cover only unliquidated entries and entries within 80 days of liquidation, while older liquidated entries need a formal protest under 19 U.S.C. 1514 that must be filed within 180 days of liquidation, or the refund is forfeited permanently. (Protest standing runs wider than CAPE filing: the statute also lets consignees, sureties, and others who paid the charges protest.) Phase 3, which opened October 6, 2026, covers finally liquidated entries, but only for importers who are plaintiffs in the roughly 3,700 pending Court of International Trade cases and who submitted their importer-of-record numbers to CBP by July 30, a pool covering about $11.4 billion, under 7 percent of the IEEPA total. Everyone else with finally liquidated entries past the 180-day mark has no administrative path left. The IEEPA duties started flowing in February 2025, and with liquidation typically landing 314 days after entry, the first wave's entries are liquidating now, on a rolling basis, with each liquidation starting a 180-day fuse that nobody is watching for the importers who never hired counsel.
One structural fact before the business: the finally-liquidated refund path is contested law. On April 17, 2026, a CIT judge ordered CBP to reliquidate finally liquidated IEEPA entries for every importer, not just the plaintiffs before him; the Justice Department appealed, arguing CBP has no statutory authority to reliquidate finally liquidated entries on its own, and a Federal Circuit stay could pause those disbursements while the appeal is pending. Refunds on unliquidated and non-final entries proceed regardless of the appeal. The contested pool is the finally-liquidated one, which is exactly the pool the protest-triage product serves, so the underwriting has to price the litigation risk, not assume it away.
Picture the actual customer, a $40 million distributor of commercial lighting fixtures that paid about $1.8 million in IEEPA duties across 400 entries. The owner has no ACE account, no ACH enrollment, and a customs broker who files entries but never offered to file refund claims for a non-customer upsell. Four hundred entries means a single CAPE declaration, except eleven of them failed validation for reasons the portal explains in codes, and nine liquidated in August, which means the protest window on those nine closes in February. This importer is not confused about whether they are owed money; they are drowning in the procedure for collecting it, which is a different problem, and a billable one.
Why Now
Phase 3 opened October 6, 2026. That day, CBP opened the final administrative window for finally liquidated entries, and it covers only CIT plaintiffs, which leaves everyone else with a binary question, your entries either still have a live path or they don't, answerable only by checking every entry's liquidation status against its protest deadline. That triage question is urgent, unanswered, and worth money to 330,000 importers this month, not next year.
The DIY path is failing at about 18 percent, and 6.1 million failed entries is not a rounding error; it is proof that the self-service portal does not serve the long tail. The failure modes are procedural, not substantive: wrong entry populations, declarations that can't be amended once accepted, entries listed on two declarations, ACH enrollment gaps. Rejection rates ran higher in CAPE's first weeks as filers learned the system, which is what you'd expect from a portal whose users are doing this for the first time. Every one of these is fixable by software that validates before submitting, which is the entire pitch.
The protest clocks are expiring on the earliest entries. IEEPA collection began in February 2025, and since liquidation typically lands 314 days after entry, the first wave's liquidations land in late 2026 with their 180-day protest deadlines in early 2027, so an importer who assumes the portal will handle it and discovers in March that February 2025 entries liquidated in September has already lost the money. Deadlines that pass silently are the best salespeople in regulated industries, which is why the triage scan has to surface every deadline for free, no strings attached: the business earns by being the party that told the importer, not by keeping them in the dark.
New duties are accruing under replacement authorities, and they are drawback-eligible. The tariffs did not end in February, they changed statutes: a 10 percent Section 122 global surcharge ran from February 24 to July 24, 2026, and was replaced by a Section 301 action reaching 60 economies at 10 to 12.5 percent, built on a forced-labor theory. Section 301 duties are fully recoverable through duty drawback when the goods are re-exported, and the five-year drawback window means every new duty paid today is a refund claim maturing tomorrow. Two caveats a trade lawyer would insist on: a Section 301 action built on forced labor is legally novel, forced-labor enforcement has traditionally run through exclusion orders rather than tariff actions, so these particular duties carry more legal risk than the 2018 China 301 duties; and Section 232 duties are a different animal, largely excluded from drawback by presidential proclamation, so the annuity runs on the 301 stack, not the 232 one. The recovery business didn't end with IEEPA. It changed shape, from a one-time refund event into an ongoing duty-management annuity. The wave becomes the business.
The 2021 drawback window is closing now. Duty drawback reaches back five years from importation, which means imports from 2021 are expiring from eligibility in real time through 2026, even though CBP disburses about $1 billion a year in drawback refunds, per the Government Accountability Office, overwhelmingly to large firms with broker relationships, leaving every mid-size exporter sitting on unfiled 2021 drawback to watch money evaporate on a published schedule.
The government pays you to wait, so the pitch is certainty, not speed. This deserves honesty: at 6 percent compounded daily a delayed refund earns the importer interest, which means "file faster" is a weak pitch. The product sells certainty of capture, not velocity of payment.
The $19 Billion Validation Failure: An Original Calculation
Run the arithmetic on CBP's own declaration. Importers submitted about 33 million entries; 6.1 million failed validation, about 18 percent, nearly one in five. At the average duty load of $3,132 per shipment ($166 billion across 53 million shipments), 6.1 million failed entries represents up to $19.1 billion in claimed refunds sitting in validation failures. The assumption is stated plainly: failed entries carry the average duty. If failed entries skew toward smaller importers with messier records, which is likely, the true dollar figure is lower and the count of affected importers is higher; at roughly 160 entries per importer, 6.1 million failures touch tens of thousands of importers, which is worse for the DIY thesis, not better.
Up to $19.1 billion in claimed refunds is stuck in validation failures. Nearly one in five submitted entries failed.
Now the concentration math: $166 billion across 330,000 importers is a mean refund of about $503,000 per importer, a number that misleads in the usual direction. But $134.7 billion, 81 percent of the value, is already accepted, and accepted claims skew large: the importers with counsel filed first and filed big. That leaves roughly $31 billion unclaimed, split between mid-size and large importers still working through counsel and brokers, and the long tail of smaller importers whose refunds are real but too small to justify trade counsel at $500 to $1,500 an hour. A $40,000 refund cannot carry a $25,000 legal bill. It can carry a $499 flat fee. That is the entire wedge, expressed as arithmetic: the recovery industry's unit economics only work below the price point where lawyers operate, and that price point is where software lives.
One more number for the financing leg. CBP pays accepted claims in 60 to 90 days. On a $500,000 accepted refund, the 6 percent statutory interest for 90 days is about $7,400, which the importer keeps. An advance product that funds 80 percent of the accepted claim at 2 percent per month for the waiting period costs the importer roughly $20,000 on that same $500,000 and delivers $400,000 of liquidity 90 days early. For a distributor staring at a tariff-inflated payables cycle, that trade is rational, and the spread is the business.
The Gap in the Market
| Player | What They Do | What's Missing |
|---|---|---|
| Large customs brokers | Several offer flat-rate CAPE filing, CSV prep, and validation handling. | Only for existing brokerage customers. A broker's refund service is a retention tool, not an acquisition channel; the 330,000 importers include hundreds of thousands of companies that are nobody's customer. Brokers do not hunt the long tail at $499 economics. |
| Trade law firms | CIT litigation, protest drafting, eligibility opinions at $500 to $1,500 an hour. | Rational only above roughly $10 million in exposure. A $40,000 refund cannot carry counsel. The firms know this; it is why they publish tier guidance telling small importers not to hire them. |
| GingerControl | The closest thing to the product: drawback service with algorithmic import-export matching, IEEPA refund guides, contingency pricing, filing through a partner broker. | Content-led and mid-market focused. No refund-advance financing, no protest-deadline triage product, no flat-fee long-tail filing tier. The obvious acquirer or the obvious competitor, depending on execution speed. |
| Tariff Recovery Today (Irongate / Master Plan) | Contingency-fee filing team for IEEPA refunds, explicitly not attorneys. | Services, not software. A filing team scales with headcount; it has no triage engine, no ongoing drawback product, and no retention mechanism once the IEEPA wave passes. |
| 1st Capital Financial | "File and Fund": prepares and files the CAPE declaration and advances capital against the refund so the importer doesn't wait on CBP. | A finance company, not a software platform. Financing without the triage engine means underwriting refunds it didn't validate; software without financing leaves the liquidity pitch on the table. The two belong together and nobody has combined them. |
| Evana, Passport | Drawback recovery for ecommerce brands, contingency-based, retroactive five years. | Vertical-specific (DTC/ecommerce) and drawback-only. They don't touch CAPE, protests, or the IEEPA event, and the event is where the urgency lives. |
| CBP's CAPE portal itself | Free, direct, authoritative. The correct path for sophisticated filers. | A roughly 18 percent validation failure rate, three-to-four-week ACE setup, no deadline triage, declarations that can't be amended once accepted. Free is expensive when the filing is wrong. |
The pattern: the top of the market is served three times over, by brokers, lawyers, and contingency filers. The long tail is served zero times, by anyone, as a software product. And hovering over all of it is the cautionary tale of the last great American recovery wave: the Employee Retention Credit, on which the IRS paid out $283 billion across nearly 5 million claims, spawning "ERC mills" that charged 10 to 15 percent contingency, cold-called businesses that didn't qualify, and triggered an IRS moratorium and 2,039 criminal investigations, per Accounting Today's fraud-wave retrospective. The demand the mills proved was real. In tariff recovery, trust is the moat: publish the fee schedule, sign the filings, charge nothing upfront, and tell the prospect with dead entries that they get nothing. The mills never did that last part.
The Solution
To be clear about what this page is: a free idea, not a company. Nobody paid for it, and Live in the Future is not building it. Take it. A tariff-recovery platform with four modules that follow the importer from triage through the durable annuity:
1. Triage (free scan): The importer connects ACE or uploads broker entry data, and the engine classifies every entry into four buckets, CAPE-eligible now, protest before a date, CIT-plaintiff-only, or dead, producing a deadline dashboard that shows each entry's liquidation status, its 180-day protest fuse, and the dollar value at risk in each bucket. Free. This doubles as the lead generation, because the scan is genuinely useful even for importers who never pay, most of whom have no idea which of their entries are already stranded. Scan data is deleted on request, never sold, and never used for lending decisions without explicit consent. The scan is not legal advice; close calls get human review, and the errors-and-omissions policy covers scan errors, not just filings.
2. File ($299 to $1,499 flat, by entry count): CAPE declarations prepared, pre-validated against CBP's known rejection patterns, and filed through a partner broker's power of attorney or prepared for importer self-upload. For liquidated entries inside the protest window, protest packages drafted to partner counsel's template and filed before the fuse burns out. Flat fee, published, no contingency on the straightforward filings: the price has to sit below the lawyer threshold or the wedge collapses.
3. Fund (1.5 to 3 percent per month, roughly 20 to 43 percent APR): Once CBP accepts a declaration, the platform advances 70 to 85 percent of the refund while the 60-to-90-day payment clock runs, which is about double most SMB credit lines, stated plainly so the importer can compare it against a credit line or factoring. Underwriting is the triage engine's output: accepted declarations with clean validation have highly predictable payout. Note the dual role, stated equally plainly: the same company prepares the filing and prices the advance against it, grading its own homework, so advances are non-recourse, if CBP denies or reduces the refund, the platform eats the loss, and the contract says so. This is the fintech spread, and it is the module that makes the platform sticky, because the importer who took your advance is your customer for the next module.
4. Drawback engine (18 to 25 percent contingency): The annuity. Ongoing matching of imports to exports under substitution rules at the 8-digit HTS level, drawback claims filed electronically through ACE, accelerated-payment privileges so refunds land in weeks instead of years. Every importer who re-exports, manufactures for export, or destroys non-compliant inventory is paying new Section 301 duties today that become drawback claims tomorrow. The IEEPA wave brings the customer in; the drawback engine keeps them for a decade.
Revenue Model
| Revenue Stream | Amount | Notes |
|---|---|---|
| CAPE filing (flat fee) | $299–$1,499 | Tiered by entry count. The wedge product; priced to sit under the lawyer threshold. |
| Protest filing + triage | $999 + 10% contingency | For liquidated entries inside the 180-day window. Higher touch, partner counsel review. |
| Refund advance (financing spread) | 1.5–3%/month | On 70–85% of accepted refunds, over the 60–90 day CBP payment window. |
| Drawback contingency | 18–25% of recovered duties | Ongoing annuity. Below the 15–30% traditional brokers charge small programs. |
| Duty monitoring subscription | $149–$599/month | Liquidation watch, new-tariff alerts, exclusion tracking. Retention layer. |
Unit economics at 5,000 importers: 5,000 filings at a $499 average is $2.5M, which is the simple part. Refund advances: 600 importers take advances averaging $60,000 at 2 percent per month over two months, which is $1.44M in spread, earned while CBP does the waiting. Drawback: 800 importers convert at an average $45,000 recovered and a 20 percent contingency, which is $7.2M, and every one of those claims took a machine minutes and a broker an hour. Monitoring: 2,000 subscribers at $249 a month is $6M in ARR by end of year two. Blended year-two revenue near $17M on a team of 25 to 30, with the drawback and monitoring lines compounding while the IEEPA filing line decays, exactly as designed.
Market Size
TAM: Start with the long tail. The Census Bureau counted about 221,000 small and mid-size importers in 2021, roughly 236,000 by 2023; assume about 250,000 today, most of them without dedicated trade counsel, which is a stated assumption, not a measured fact. Assume an average SMB refund of $35,000, also stated: 250,000 × $35,000 = $8.75 billion in long-tail refunds. At an 8 percent blended take rate across filing fees, protest contingency, and financing spreads, the one-time service TAM is about $700M. Add the ongoing drawback annuity: CBP disburses about $1 billion a year today and claim values have more than doubled since 2019, with new Section 301 forced-labor duties accruing; at a 20 percent contingency the run-rate service TAM is $200M+/year. Combined TAM: roughly $700M one-time plus $200M+/year recurring.
SAM: The reachable three-year window: 60,000 SMB importers who paid IEEPA duties, have not filed, and are acquirable through broker partnerships, trade associations, and the triage scan. At a $30,000 average refund and an 8 percent take: 60,000 × $30,000 × 8% = $144M in one-time service revenue, before financing spreads and drawback conversion, which is where the margin lives.
SOM (year 3): 6,000 importers onboarded, roughly 10 percent of SAM's importer count: $3M in filing fees, $1.7M in advances, and $7.2M in drawback contingency as the 2021–2026 drawback vintages mature, for $11.9M in one-time and transactional revenue, plus $6M in monitoring ARR, for about $18M in year-three revenue. Aggressive, but the acquisition funnel is the free triage scan, and every scanned importer with a live deadline is a customer whose deadline we can still save.
Startup Costs
| Category | Cost | Notes |
|---|---|---|
| MVP: ACE data pipeline, triage engine, filing workflow (6 months) | $220K | 2 engineers. Entry ingestion, liquidation tracking, CAPE CSV generation, validation pre-checks. |
| Partner broker integration + POA infrastructure | $40K | The platform files through a licensed broker's power of attorney; the legal and technical plumbing for that relationship. |
| Trade counsel (protest templates, compliance review) | $60K | A customs lawyer reviews every filing template. One bad protest template ends the company. |
| Lending facility setup + state licensing review | $50K | The Fund module is lending. Usury caps and state licensing vary; get the opinion letters before advancing a dollar. |
| E&O and cyber insurance | $35K/yr | Table stakes when you touch client filings and advance capital. |
| Customer acquisition (broker partnerships, trade associations) | $80K | The triage scan is the funnel; partnerships are the distribution. |
| Founder salaries (year 1) | $240K | 2 founders, modest. The CEO should be able to sell to importers; the CTO builds the pipeline. |
| Operating buffer | $50K | Hosting, CBP interface costs, support line. |
| Total | $775K |
Limitations
Eighty-one percent of the pool by value is already claimed. Say it again, because the whole pitch depends on not misunderstanding it: the remaining value skews small, scattered across importers whose individual refunds are real but modest. This is a volume business or it is nothing. Anyone modeling whale recoveries is modeling a market that filed in April.
The IEEPA wave does not replenish. It is a one-time event with a finite pool and expiring deadlines, which means the company's durability is entirely a function of converting filers into drawback and monitoring customers. A platform that files 10,000 CAPE declarations and retains nobody is a profitable corpse. Don't build that.
CBP could change the game mid-play. The agency could extend protest deadlines, simplify CAPE validation, or open Phase 3 to non-plaintiffs, and any of those would shrink the service window. Separately, the finally-liquidated refund path is under Federal Circuit appeal: if the court grants the government's requested stay, disbursements on that pool pause while the appeal runs, which turns accepted-but-unpaid claims into a litigation asset, exactly the exposure the Fund module's non-recourse terms are priced for. Regulatory tailwinds cut both ways; today's complexity is tomorrow's simplification project.
Standing is strict and unforgiving. Only the importer of record can claim; the platform cannot manufacture eligibility for freight forwarders, customs brokers' clients who weren't the IOR, or consumers, no matter what the marketing implies. Every triage scan will surface prospects who are owed nothing, and the honest product tells them so, which caps conversion and is the entire point.
The Fund module is a lending business wearing a software costume. State licensing, usury caps, and capital requirements are real, and advancing against government receivables means underwriting CBP's payment behavior, which has a known failure mode: 6.1 million entries failed validation, and an advance against a failed declaration is an unsecured loan to an importer who may not repay it. One more underwriting wrinkle: CBP nets an importer's outstanding federal debts from refunds before disbursement, so the advance has to be sized against the net refund, not the gross.
The ERC precedent is a warning, not just an analogy. The mills proved the demand and then poisoned the well: aggressive marketers, ineligible claims, an IRS moratorium, criminal investigations. A tariff-recovery platform that cold-calls importers with "free money" messaging will inherit that taint. The defense is structural: no upfront fees, published pricing, signed filings, and eligibility opinions that sometimes say no.
Drawback has its own graveyard. CBP's drawback backlog has run around 50,000 claims, filers complain the ACE system caps declarations at 9,999 lines and chokes on file sizes, and one industry veteran's summary, "drawback is simply not in the 21st century," is both the opportunity and the warning: the plumbing is old, the edge cases are endless, and the 5-year substitution rules punish sloppy recordkeeping with denied claims.
Not every failed entry is a fixable one. Entries subject to antidumping or countervailing duty orders, drawback-associated entries, and ineligible entry types are correctly rejected by validation; the 6.1 million includes real rejections alongside procedural ones, and the platform's pre-checks have to distinguish them or they will file losing claims at volume. That is the correlated-failure risk the DIY world doesn't have: one bad validation rule strands thousands of importers simultaneously, which is the platform's largest unpriced risk. And a final sourcing note: the load-bearing figures in this piece, $166 billion, $134.7 billion accepted, 6.1 million failed, come from CBP's court declarations as reported by trade press, not from a public dataset the reader can download. Treat them as the best available numbers, not as audited ones.
Strongest Counterargument
This is a twelve-month arbitrage, not a company. The pool is 81 percent claimed by value, the remaining importers are either already working with brokers or have refunds too small to matter, and any competent customs broker will file CAPE declarations for its customers at a flat rate. The IEEPA event will be fully processed by mid-2027, the protest windows will close, and the platform will be left holding a drawback product that GingerControl, Passport, Evana, and half a dozen AI matching startups already sell. The honest end state is a feature inside someone else's trade compliance suite, acquired for the customer list.
There is real force in this, and the honest version of the business concedes the first half. As a pure IEEPA play, anyone starting today is late to the top of the market, the twelve-month window is real, and the acquisition-by-incumbent end state is not a failure mode, it is a plausible exit. The rebuttal is not that the IEEPA wave lasts forever but that the wave was never the business, which is the 250,000 long-tail importers that brokers structurally do not serve, because brokers file for customers and do not acquire non-customers at $499 economics. It is the protest-triage layer that nobody productized, the financing spread that no software company combined with filing, and the drawback annuity on the replacement tariff stack, Section 301 forced-labor duties accruing since July 2026, fully drawback-eligible, five-year window, compounding. The ERC mills proved that a large, confusing, time-boxed government payout reliably creates a recovery industry; they also proved that the winners are the ones who convert the event into retained relationships instead of strip-mining it. And the "brokers will handle it" thesis is already falsified at 18 percent: 6.1 million failed entries are the market's answer.
What You Can Do
If you imported under IEEPA: Pull your ACE entry report this week and check liquidation dates. Anything unliquidated or within 80 days of liquidation goes through CAPE. Anything liquidated longer ago needs a protest assessment against the 180-day clock, now, not next quarter. Enroll in ACH refunds before you file anything; CBP pays electronically or not at all, and $1.3 billion is currently on hold for importers who skipped that step. Verify every entry number before you file: a rushed wrong filing can strand the refund it was meant to save. And if your entries are finally liquidated and you are not one of the roughly 3,700 CIT plaintiffs, get trade counsel's honest answer about your options, because the administrative path is closed.
If you re-export, manufacture for export, or destroy non-compliant imports: Run a drawback assessment going back to 2021, because the five-year window is expiring on your oldest entries first. The 99 percent refund on Section 301 duties, including the new forced-labor tariffs, is the durable version of the IEEPA story. Most eligible mid-size firms have never filed; the GAO's $1 billion a year flows overwhelmingly to companies with broker relationships you probably don't have.
If you're a customs broker: Your customers are covered; your non-customers are the market, so white-label a triage engine and let the free scan do the acquisition your sales team can't do at $499 economics, because the brokers who productize the long tail will own it while the rest watch a software company intermediate their industry's largest refund event.
If you're building this: Sign the partner broker before writing the triage engine, because without filing authority you have a dashboard, not a product. Get the lending opinion letters before advancing a dollar, because the Fund module's unit economics die under a usury violation. Publish your fee schedule on day one and build the "you get nothing" path into the triage scan, because the ERC mills are why every importer you pitch will assume you're one of them until you prove otherwise. And start with the entries whose protest clocks expire first: fuses are the funnel.
The Bottom Line
$166 billion is the largest tariff refund in American history. Claiming it requires an ACE account, a CSV capped at 9,999 rows, and a calendar that tracks 180-day protest fuses across every entry you ever filed, which explains why the biggest importers handled this in April, with counsel, at $1,500 an hour, while everyone else stares at a portal with an 18 percent failure rate and a Phase 3 that opened October 6 for everyone except them. A platform that triages every entry for free, files the winnable ones at a flat fee, advances the refund while CBP takes its 90 days, and then keeps the customer for drawback on the tariffs that replaced IEEPA is not betting that trade chaos continues. It is betting that 250,000 small importers would rather pay $499 than read the Code of Federal Regulations. It is betting that the honest version of the ERC playbook wins the recovery industry's second act: published pricing, signed filings, eligibility opinions that sometimes say no. The screen costs less than the broker's hourly minimum. The money is real. The paperwork is the business.