TL;DR: When CBP flipped on Phase 1 of its Consolidated Administration and Processing of Entries (CAPE) system on April 20, 2026, it opened the door to refunds on duties paid under the International Emergency Economic Powers Act (IEEPA). Phase 1 covers roughly 63% of eligible entries—the ones still unliquidated or liquidated inside an 80-day window—and pushes the other 37% into undefined future phases. The catch: brokers are filing with an average 20% data-error rate, and a single bad entry number can bounce your whole claim into the back of a queue that already holds 53 million entries from 330,000+ importers. The formula is simple: confirm eligibility + scrub your entry data + file clean CSVs early = duty cash back in 60–90 days. Get sloppy and you wait quarters.

The refund window opened. Most brands aren’t ready to walk through it.

For two years, importers paid IEEPA tariffs and booked them as sunk cost. There was no mechanism to claw the money back—just a line item that quietly ate margin. CAPE changes that. It is a refund pipeline bolted onto the Automated Commercial Environment (ACE), and it lets importers of record (or their brokers) submit structured CSV claims and get duties returned by ACH direct deposit.

The problem is that “open” and “easy” are not the same thing. CBP turned on the spigot, then immediately warned that the plumbing isn’t fully spec’d—validation rules, discrepancy handling, and rejection workflows are still being defined in real time.

“The refund is real, but the filing discipline determines whether you see it this quarter or next year,” says Dana Whitfield, a licensed customs broker and trade-compliance advisor at Meridian Customs Group. “We audited a batch of entries filed by other brokers and found errors in one out of every five. Each of those is a rejection, and every rejection is weeks lost.”

Most brands will treat CAPE like a button they press once. The operators who actually recover their money will treat it like a data project—because that’s what it is.

Consider the asymmetry. The duty rate is fixed. CBP’s processing timeline is fixed. The one variable you fully control is the quality of the data you submit. That’s where the entire spread between “recovered” and “stuck in the queue” lives. A brand that wired its trade data into a single source of truth two years ago can file clean today in an afternoon. A brand that’s reconstructing entry numbers from PDFs and broker emails will spend weeks—and still file with errors. The work you did (or didn’t do) on data hygiene is now denominated in dollars.

How CAPE actually works: the five stages

CAPE moves a refund claim through five discrete stages. Understand them, because each one is a place your claim can stall.

The pipeline

1. Submission — You (or your broker) upload a CSV file of entry numbers. The hard cap is 9,999 entries per declaration.
2. Validation — Two checks run. First, the file format is verified. Second, each entry is validated to confirm IEEPA duties were actually paid.
3. Claim assignment — Declarations that pass validation receive a claim number. This is your tracking handle.
4. Review and liquidation — CBP confirms the refund amount and processes the entries.
5. Refund — Funds land via ACH direct deposit to the importer of record or a designated notify party.

The structure looks clean. The risk lives in stage two: validation is binary, and a malformed file or a mismatched entry doesn’t get a warning—it gets rejected.

What Phase 1 actually covers

Phase 1 is deliberately narrow. It handles entries that are unliquidated or were liquidated within the last 80 days. That’s about 63% of eligible entries. Everything else waits.

Status of entry	Phase 1 eligible?	Expected refund timing
Unliquidated	Yes	60–90 days after CAPE acceptance
Liquidated within 80 days	Yes	60–90 days after CAPE acceptance
Liquidated beyond 80 days (“finally liquidated”)	No	Future phase, TBD
Suspended / extended / under review	Partially	Held until entry liquidates
Flagged for reconciliation	No	Future phase, TBD
Designated on a drawback claim	No	Future phase, TBD
Open protest on file	No	Future phase, TBD
Not filed in ACE	No	Future phase, TBD
AD/CVD pending liquidation	No	Manual resolution required

If your entries fall in the “no” rows, filing now does nothing but burn cycles. Sort them first.

The math: what’s actually at stake

Run the numbers before you decide how much energy this deserves. The size of the prize is a function of your import volume and your IEEPA exposure rate.

Recoverable IEEPA duty  =  Annual import value (subject to IEEPA)
                            ×  IEEPA duty rate
                            ×  % of entries in Phase 1 window (~63%)
                            ×  (1 − broker error rate)

Example, a $20M/yr importer:
  $20,000,000  ×  10%  (IEEPA rate)        =  $2,000,000  duties paid
  $2,000,000   ×  63%  (Phase 1 eligible)  =  $1,260,000  addressable now
  $1,260,000   ×  (1 − 0.20 error rate)    =  $1,008,000  if you file dirty
  $1,260,000   ×  (1 − 0.02 error rate)    =  $1,234,800  if you file clean

That gap—roughly $227,000 in this example—is the entire ballgame. It is not the duty rate or CBP’s timeline that decides your recovery. It’s your data hygiene. A 20% broker error rate versus a disciplined 2% rate is the difference between leaving a quarter-million dollars in the queue and collecting it.

Why the queue matters

ACE was already under load before CAPE. Now it’s absorbing roughly 53 million entries from over 330,000 importers at once. There is no priority lane for clean filers—but there is a penalty lane for dirty ones. Every rejected entry goes to the back, gets re-worked, and re-queued behind everything filed since. In a congested system, being early and correct compounds.

The five moves to make this week

You don’t need to wait for CBP to finalize every protocol. You need to get your house in order so that when your entries are eligible, your claim sails through validation on the first pass.

1. Confirm ACE Portal access and stand up your ACH refund account. No ACH destination, no refund—full stop. This is the one piece you cannot fix retroactively once a claim is approved.
2. Segment your entries. Build a list of Phase 1-eligible entries (unliquidated or liquidated inside 80 days) versus everything bound for a future phase. Filing the wrong cohort wastes the window.
3. Audit entry data before you submit. Entry numbers, duty amounts, importer-of-record numbers, HTS classifications. This is where the 20% error rate lives. Catch it now or eat a rejection later.
4. Keep filing protests on entries near their deadline. CAPE does not pause the protest clock. If an entry is approaching its protest window and isn’t Phase 1-eligible, protect your rights the old-fashioned way.
5. Decide who owns this. A trade advisor, your broker, or an internal owner—but someone has to own the CSV discipline end to end. Diffuse ownership is how the 20% error rate happens.

Building a CSV that survives validation

The unit of work in CAPE is the CSV declaration. Get the file right and the rest is waiting; get it wrong and you’re in the rejection-and-resubmit loop. A few hard-won principles:

Batch deliberately, not maximally

The system allows up to 9,999 entries per declaration, but bigger isn’t better. A single bad row doesn’t always fail in isolation—when validation chokes, you want a small enough batch to diagnose and re-file fast. The practical sweet spot most operators land on:

Batch size	Pro	Con
50–250 entries	Easy to audit, fast to re-file on rejection	More declarations to manage
1,000–2,500 entries	Balanced throughput and traceability	Harder to isolate a bad row
9,999 (max)	Fewest declarations	One error can obscure the whole batch; slow to re-validate

Unless your data is pristine and reconciled, file in the smaller bands until you’ve proven your error rate is near zero. Then scale the batch size up.

Reconcile before you export

The fields that get entries rejected are predictable. Validate each of these against your system of record before the CSV ever leaves your machine:

• Entry number — exact format, no transposed digits, no leading-zero loss from a spreadsheet auto-format.
• Importer of record number — must match the IOR on the original entry.
• Duty amount paid — must reconcile to what was actually remitted, not what was estimated.
• HTS classification — the line that determines IEEPA applicability in the first place.
• Entry date and liquidation status — confirms Phase 1 eligibility before you waste the slot.

The leading-zero problem alone accounts for a meaningful share of avoidable rejections. Spreadsheets silently strip them; CAPE does not forgive them. Export as text, not as a number-formatted column.

Broker, advisor, or in-house: who should file

Three models, three risk profiles. Choose based on volume and data maturity, not habit.

Model	Best for	Watch-out
Your existing broker	Brands already running clean entries through one broker	You inherit their error rate—audit a sample before trusting the whole batch
A specialist trade advisor	High IEEPA exposure, AD/CVD complexity, or messy historical data	Costs more; worth it when the recoverable sum is large
In-house filing	Lean teams with strong data discipline and ACE access	You own every rejection; only viable if your entry data is already reconciled

The 20% error rate that’s circulating is a broker statistic. It is not a law of nature—it’s a symptom of filing without auditing. Whoever owns your claims, the non-negotiable is a verification step between “data” and “submit.” If your filing partner can’t describe their pre-submission audit process, that’s your answer.

What this means for your operating rhythm

Duty recovery isn’t a one-time event you delegate and forget. For brands with ongoing IEEPA exposure, it becomes a recurring close-the-loop process:

• Monthly: Reconcile new entries, tag IEEPA-affected lines, and confirm which are entering the Phase 1 window.
• Per CBP update: Re-check validation rules and phase timing. The protocols are still moving—what bounced last month may pass this month.
• Quarterly: Reforecast recoverable duty for finance. As later phases open, your addressable pool grows from the current 63% toward the full amount.
• Annually: Pressure-test your filing partner’s error rate against a sampled audit. If it’s drifting toward the 20% industry average, the partner—or the process—needs to change.

Treat the refund as a line item finance can plan around, not a surprise windfall. The brands that model this into working-capital planning turn a compliance chore into a predictable cash inflow.

The PSC trap: correct before you submit

Here’s the rule that’s tripping people up. Per CBP’s April 22 guidance, once an entry is accepted in CAPE, you can no longer correct it through a Post Summary Correction (PSC). Acceptance locks the entry.

The nuances that matter:

• A PSC you filed before CAPE submission does not block acceptance. Prior corrections are fine.
• If an unliquidated entry gets rejected by CAPE, you can still PSC it before resubmitting.
• The takeaway: find and fix every error before you submit, not after. CAPE is unforgiving about timing. A clean entry that’s wrong is worse than a rejected entry you can still repair.

This is the single most expensive misunderstanding in the system. Operators who submit first and verify later are locking in errors they can no longer fix.

After you submit: tracking, rejections, and recovery

Submission is the start, not the finish. Once a declaration is in, your job shifts to monitoring and triage.

Hold onto your claim numbers

When a declaration passes validation, it gets a claim number. That number is your only reliable handle for following the refund through review and liquidation. Log it against the underlying entries in your own system—do not rely on memory or a buried email. When finance asks “where’s the money,” the claim number is the answer.

Triage rejections fast

A rejection is recoverable, but only if you act before the entry’s other clocks run out. The disciplined response:

1. Read the failure reason. File-format failures and entry-level failures are different problems—one is a CSV issue, the other is a data issue.
2. Fix at the source. If an unliquidated entry was rejected, correct it via PSC before resubmitting. Don’t patch the CSV and re-file the same bad underlying data.
3. Re-batch smaller. If a large declaration bounced and you can’t isolate the culprit, split it and resubmit in smaller bands to surface the bad row.
4. Re-file promptly. Every day a rejected entry sits is a day it falls further behind newly filed clean claims.

Don’t let the refund clock obscure the protest clock

This bears repeating because the failure mode is expensive: CAPE running in the background does not suspend protest deadlines. For any entry approaching its protest window that isn’t cleanly moving through Phase 1, file the protest. Preserve the right; reconcile the bookkeeping later. Losing a protest deadline because you assumed CAPE “had it” is an own goal.

The unfinished edges

CAPE is live, but it’s a Phase 1 product, and it shows. Plan around the gaps:

• AD/CVD limbo. Roughly 166,000 entries subject to antidumping or countervailing duties still need manual resolution. CBP hasn’t published the process. If your goods are AD/CVD-affected, expect a separate, slower track.
• The other 37%. Reconciliation entries, drawback-designated entries, open protests, and finally-liquidated entries are all deferred. There’s no published timeline for when those phases open.
• Validation is a black box. CBP hasn’t fully documented how it reviews entries, handles discrepancies, or manages rejections at scale. Build in buffer; don’t promise your CFO an exact refund date.
• System strain is real. Pointing 53 million entries at an already-loaded ACE means slowdowns are likely. Early, clean filing is your only lever.

CBP signaled its next CAPE progress update would land April 28, 2026—watch for protocol changes and adjust your filing cadence accordingly.

None of this is a reason to wait. The gaps are in CBP’s process, not in your prep work. Your eligibility segmentation, your data audit, and your ACH setup are entirely within your control today. Do them now, so that the moment your entries become processable, your claim is the clean one that sails through while everyone else is still untangling spreadsheets.

FAQ

How long until I actually see the money? For Phase 1 entries—unliquidated or liquidated within 80 days—CBP’s stated target is 60–90 days after CAPE accepts the claim. Entries that are suspended, extended, or under review won’t pay out until they liquidate, which can stretch the timeline by quarters. Treat 60–90 days as a best case contingent on a clean first-pass submission, not a guarantee.

Should I file everything now to get in line early? No. File only your Phase 1-eligible cohort, and only after you’ve audited it. Submitting future-phase entries does nothing—they’re not processed yet—and submitting dirty Phase 1 entries triggers rejections that push you behind everyone who filed clean. Early matters, but early and correct is the whole point. Speed without accuracy just relocates your money to the back of a 53-million-entry queue.

What’s the most common reason claims get rejected? Data mismatches at the entry level—wrong entry numbers, duty amounts that don’t reconcile, importer-of-record errors, or HTS issues. This is the source of the 20% broker error rate. Because you can’t PSC an entry once CAPE accepts it, every error you don’t catch upfront becomes either a rejection (recoverable) or a locked-in mistake (not). Audit before you submit, every time.

---

Implementation Difficulty: 3/5 — The filing mechanics are straightforward; the data hygiene and entry segmentation are where the work and the risk live.

Impact Estimates (for a $20M/yr importer with ~10% IEEPA exposure):

• Conservative: $900K recovered (dirty filing, partial Phase 1 capture)
• Likely: $1.0M–$1.2M recovered (disciplined filing, full Phase 1 window)
• Upside: $1.9M+ recovered as later phases open and the remaining 37% becomes eligible

Time to Value: 60–90 days from a clean CAPE submission to ACH deposit, assuming first-pass validation.

The brands that win this aren’t the ones who file fastest—they’re the ones whose entry data is already clean when the window opens. That’s a systems problem, and it’s the one CommerceOS by Endless Commerce is built to solve: one source of truth for entry data, duty exposure, and landed cost, so your refund claims pass validation the first time instead of bouncing into the queue. Stop treating duty recovery as a fire drill. Book a demo and see what disciplined trade data looks like.