Network redesign under tariffs: where to start

First, price the exposure — flows, not headlines

The common first mistake is treating “tariffs” as one number. Practitioners who handled this well started by building a flow-level exposure map: every lane from origin country to entry point, with the duty rate that actually applies to that product’s classification, the annualized value flowing over it, and who contractually owns the duty — your incoterms decide whether the tariff is your cost, your supplier’s, or a negotiation already priced into your quotes. Two things usually surprise people when this table is done. The exposure is concentrated: a handful of lanes and a handful of classifications carry most of the money, which means the response can be targeted instead of a full network overhaul. And classification matters as much as geography: tariff engineering — how a product is classified, valued, or even where its last substantial transformation happens — is often a bigger lever in the first year than moving a factory.

That map is also what stops the meeting from being a debate about headlines. When the CFO asks what the tariff costs and the answer is a lane-by-lane number with ownership attached, the conversation moves from panic to which exposures are worth paying to remove.

Run scenarios before you commit capital

A network redesign is a bet on where trade policy, labor costs and demand will sit for the next five to ten years — and nobody knows. The practitioners who avoided the expensive regrets all insisted on the same discipline: scenario-test the flows before committing to any asset. Take the two or three tariff regimes that are actually plausible — the current one, a broader escalation, a partial rollback — and run your sourcing-and-distribution flows against each: total landed cost by origin, which customer commitments break, where capacity would bottleneck. If one network configuration survives most plausible regimes at acceptable cost, that robustness is worth more than the optimum on any single forecast. If every option loses money under the escalation scenario, that is not an argument for waiting — it is the number that tells you how much flexibility you can afford to buy.

Sort the levers by how fast they pay back

“Redesign” is not one decision; it is a ladder of commitments with very different speed and reversibility. Practitioners tend to work it in this order. First the moves that take quarters, not years: renegotiate incoterms and duty responsibility with suppliers, re-classify products where the tariff code genuinely allows it, qualify an alternate supplier in a different origin country, shift the flows through an existing distribution center rather than the direct lane. Then the moves that take a year or two: contract manufacturing or co-packing in the destination market — often faster and cheaper than buying a plant — and nearshored sub-assemblies that shorten the dutiable chain without moving final production. Last, the structural moves — owned plants, new distribution nodes, full regionalization — which are multi-year capital decisions that should only be made when the exposure map says the volume justifies them and the scenario work says they survive more than one trade regime.

The failure mode is doing the ladder in reverse: announcing a regionalized footprint while the duty responsibility on next quarter’s purchase orders is still sitting unrenegotiated on your side of the table.

Sequencing a footprint move without breaking the P&L

When a real move does happen — a new plant, a new distribution region — the practitioners who lived through one say the same three things. Run the old network and the new one in parallel longer than feels efficient: dual-running costs real money, but a cutover that fails mid-quarter costs customer commitments. Move product families one at a time, starting with a lane where the stakes are real but survivable, and write down what the first migration actually taught you before the second one starts. And keep the tariff scenario model alive after the move — the network you redesigned for this year’s tariff regime is one election away from being yesterday’s answer, and the companies that re-run the scenarios quarterly see the next change coming while it is still a planning question instead of an emergency.

Three questions worth asking someone who has been through it

1. Which lever moved the most landed cost in your first year — and which one did you have to walk back?
2. How did you decide how much flexibility to buy, versus accepting the exposure and pricing it in?
3. What did the cutover actually cost — the parallel runs, the service misses, the internal time — against the business case you presented?

A consultant answers these with a network-optimization study. A peer at your scale answers them with the lane they moved first, the leverage they got wrong, and what the dual-run really did to their working capital. That difference is the entire product.