Redesigning supply chain planning for permanent tariffs

Tariffs are no longer showing up as a one-off shock in the model. Manufacturing trade-weighted tariffs reportedly doubled year over year, from 1.9% to 4.7%, and 76% of trade professionals now expect tariffs to persist for four or more years [1][2]. That is the point at which tariff modeling stops being a policy watch item and becomes part of the baseline cost structure.

McKinsey's late-2025 survey found only 45% of companies were passing tariff costs through to customers on average, while 43% were planning a U.S. footprint shift over three years, 45% were raising inventory buffers, 39% were dual sourcing, and 33% were nearshoring [3]. The pattern is not uniform, and the sample is large-enterprise heavy, but it does show where planning pressure is landing: sourcing, stock, network, and price all at once. A tariff scenario planning capability only matters if it replaces static assumptions rather than sitting beside them.

Sourcing is where the tariff cost becomes structural

Sourcing decisions are where tariff cost stops being theoretical. Learning Resources moved about 16% of production from China to Vietnam and India [4]. Apple has shifted roughly 10% to 15% of iPhone assembly to India [5]. Those are not headline gestures; they are qualification work, capacity negotiation, and supplier scorecard changes. Once that kind of move starts, the old best-cost-country logic is gone unless tariff exposure, tooling, and lead time all sit in the same decision file.

That is also where static spreadsheets break first. A supplier comparison that leaves duty out of the first screen is already stale, because the organization will later have to patch the model with freight, service, and compliance exceptions. The more useful setup is a sourcing review that forces tariff cost into every candidate scenario before procurement, finance, and operations start debating which supplier is cheapest on paper.

Inventory is now a tariff instrument, not just a service buffer

Inventory is the easiest place to hide tariff pain and the easiest place to waste cash. McKinsey found 45% of firms were increasing inventories as a tariff buffer [3], and Walmart has also used pre-buys, while acknowledging that the buffer is thinning [5]. That pattern is sensible only when the stock is tied to the right SKU families and replenishment windows; otherwise the organization pays for extra days on hand without buying much more resilience. If the inventory model cannot distinguish between a tariff bridge and a permanent service buffer, it is the wrong model.

Here the data quality question matters as much as the policy question, which is why a data readiness assessment for AI inventory optimization is not a side project. If item master data, origin attributes, and lead-time assumptions are messy, the buffer decision will be messy too.

Network design has to absorb tariff cost, not dodge it

Network design is where the planning logic either compounds or corrects the sourcing choice. If tariff cost is real, footprint planning has to compare origin, freight, service, and duty together. That makes static spreadsheets fragile: they can show a landed-cost delta, but they usually cannot refresh the entire scenario set fast enough when policy moves. The useful question is not whether to nearshore in the abstract; it is which node, which SKU family, and which service promise change the network math enough to justify the transfer.

Tools such as an AI network design optimization for multi-DC retail logistics only work when the planning team already knows which constraints are non-negotiable. AI can rank scenarios quickly, but it cannot bless a foot-print assumption that customs, finance, or sales will later have to defend.

Pricing and compliance follow the plan

Pricing cannot be left to the end of the quarter as a separate commercial debate. In McKinsey's survey, only 45% of companies were passing tariff costs through on average [3], which means many are choosing margin absorption, slower repricing, or selective pushback. The planning issue is not whether to raise list price everywhere; it is whether the tariff burden has already been translated into margin floors, channel rules, and customer-specific exceptions before sales commits volume.

Compliance pressure is also moving in the wrong direction. KPMG reported Customs and Border Protection recoveries rising from $118 million in about 400 audits to $235 million in about 500 audits, with $44 million recovered from just 60 audits in the first three months of the current fiscal year [6]. That does not prove every company faces the same exposure, but it does show that customs review is no longer a background process. Any plan built on 2025 legal assumptions should be rechecked against current tariff rates, IEEPA status, and the USMCA review timeline before it is used for a 2026 or 2027 commitment.

Cadence is the redesign

The companies that outperform through 2027 will be the ones that treat tariffs as a standing planning input, not an exception to investigate when finance closes the month. Sourcing, inventory, network, pricing, and compliance need to move on the same scenario cadence so the tariff line stays in the model every time the forecast is rebuilt. That is the real redesign: not a resilience slogan, but a planning rhythm that forces the organization to keep making reconciled decisions while the policy environment keeps changing.

References

1. Tariff Volatility and Global Trade Realignment: Why 2026 Is Different — e2open
2. 2026's Supply Chain Challenge: Confronting Complexity and Disruption in Global Trade — Thomson Reuters
3. Supply Chain Risk Survey — McKinsey & Company, December 2025
4. Learning Resources moves production from China to Vietnam and India — BBC, July 2025
5. Tariff Impact on Supply Chain — Tipalti
6. March 2026 Supply Chain Update — KPMG, March 2026