Your landed cost model is lying to you when tariffs change mid-PO
When duty rates shift between PO placement and port arrival, static landed cost models quietly break — distorting pricing, margin forecasts, and drawback baselines. Here's how to fix the math.
Evana Team
Customs & Drawback
The problem
The 90-day gap that breaks your margin math
A procurement team places a PO with a supplier in Vietnam in March. The goods ship in June, clear CBP in July, and hit the DC in August. Somewhere in that window, a new Section 301 exclusion expires, an IEEPA surcharge gets layered on, or a court ruling shifts the effective rate. The landed cost the buyer typed into the ERP in March is no longer the landed cost the company will actually pay.
Most mid-market importers don't catch this until the CFO asks why gross margin came in 200 basis points below plan. By then the PO is closed, the retail price is set, and finance is retro-fitting an explanation instead of a fix.
The root cause isn't bad forecasting. It's that landed cost is usually modeled as a single static number — product cost, freight, duty rate, MPF, HMF — rather than a set of assumptions with different volatility profiles. Duty rate is the most volatile input in the current environment, and it's often the one modeled with the least rigor.
Why static duty assumptions are the weakest link
Freight rates move, but you see them move. FX moves, but treasury hedges it. Duty rates, by contrast, can change by executive order, court order, or an expiring exclusion — and the change applies to goods based on entry date, not order date. That timing mismatch is what silently corrupts the model.
Common ways static landed cost models fail right now
- ⚠️Using the duty rate in effect on the PO date instead of the projected rate on the entry date, ignoring anything scheduled to change in between.
- ⚠️Applying a single blended rate to an HTS code that has been touched by Section 301, Section 232, and IEEPA actions at different times — with no scenario for reversal or refund.
- ⚠️Treating MPF as a rounding error rather than a capped-but-real cost that scales with entered value, not duty rate.
- ⚠️Ignoring the drawback recovery potential on duties paid, which effectively lowers the true landed cost for any SKU that gets re-exported or destroyed.
- ⚠️Locking retail prices off a March landed cost, then eating the delta when July's entry summary comes back higher.
The fix
Model duty as a range, not a point estimate
The upgrade isn't complicated, but it does require finance and trade to sit at the same table. Instead of a single duty rate per HTS, build three: the rate in effect today, the rate you expect on the projected entry date, and a stress-case rate that assumes an adverse policy change during transit. Weight them by probability if you want to get fancy, or just carry all three through the model and let leadership see the spread.
For SKUs with long lead times — anything sourced from Asia with 60+ days from PO to port — the spread between best and worst case can be several points of margin. That's not a rounding error. That's the difference between hitting plan and missing it.
The second move is to treat recoverable duty as a separate line. If a portion of your imports gets re-exported (returns, wholesale to Canada, damaged inventory destroyed under CBP supervision), those duties are recoverable through drawback for up to five years. A true landed cost should net that recovery out — otherwise you're overstating cost on any SKU with meaningful re-export flow.
The drawback line changes the ROI conversation
If your model shows landed cost gross of drawback recovery, a duty drawback program looks like a nice-to-have. If it shows landed cost net of recoverable duty, drawback becomes a structural margin lever — and the ROI calculation shifts from 'how much can we get back?' to 'how much are we overstating cost by ignoring this?' Same dollars, very different internal narrative.
How to operationalize it
A tariff-sensitive landed cost model in five inputs
You don't need a new system to fix this. You need five inputs updated on a defined cadence: HTS classification (audited quarterly), current duty rate stack including Section 301/232 and any IEEPA-related surcharges, scheduled changes or known expirations in the next 180 days, expected re-export percentage by SKU family, and MPF/HMF applied to entered value.
Run those inputs through your existing landed cost formula, but produce three outputs: as-of-today landed cost, projected-at-entry landed cost, and net-of-drawback landed cost. Procurement uses the middle one for PO decisions. Finance uses all three for forecasting variance. Pricing uses the net-of-drawback figure for anything with real re-export flow.
The mechanics of drawback recovery — proving the import-to-export chain, filing within statutory windows, reconciling entry summaries against export documentation — is where most importers lose the recoverable dollars they've already earned. That's the work Evana handles, but even if you never file a claim, modeling recoverable duty in your landed cost is the right analytical move.
The takeaway
Static duty assumptions worked when tariff policy moved slowly. It doesn't anymore. Importers who model duty as a range and net out recoverable duty will price more accurately, forecast with less variance, and make better sourcing calls than importers still working off a single number typed into the ERP on the day the PO went out.