Scenario planning replaces the single landed cost estimate with a small set of named futures: current rates, the announced increase, a partial rollback, each costed per product with pre agreed actions attached. The point is not predicting which future arrives. It is deciding in advance what you will do in each, so a policy change triggers execution instead of a fresh round of meetings.
One number is a bet, not a plan
A landed cost built on today's tariff rate contains a hidden wager that the rate survives the product's selling life. When it does not, the organisation improvises: emergency pricing calls, margin surprises in the monthly accounts, sourcing decisions made at conference room speed. The cost of the surprise is usually larger than the cost of the tariff, because improvisation is expensive and late.
For more on this, see this related guide and our earlier explainer on the same theme.
Scenarios convert surprise into selection. Three or four futures, honestly costed, cover the plausible range for most categories. More than that and the model becomes a hobby. Fewer and you are back to betting. For each product that matters, the spread between scenarios is itself the key number: it measures your tariff exposure more truthfully than any single estimate.
Build the scenarios from the levers
Use the levers policy actually moves: the headline rate on your classifications, in transit and exclusion outcomes, and the secondary effects on freight and competitor pricing. A scenario is a coherent combination, not a random dial spin: announced rate with exclusions denied is one world, announced rate with your key exclusion granted is another, and they deserve separate rows because your actions in them genuinely differ.
Cost them fully: duty on the correct value basis, fees that scale with duty, and the knock on to price, margin, and volume at the price you would realistically set. A scenario that keeps margin whole by assuming a price rise the market will not bear is fiction in a spreadsheet costume. Mark assumptions visibly so the argument happens about them, in the open, once.
Attach triggers and owners
Each scenario gets its actions written while heads are cool: the price move, the sourcing conversation, the customer communication, the inventory posture, each with a named owner and the observable trigger that activates it. Triggers should be events anyone can verify, publication of the final text, an exclusion decision, the first entry liquidating at the new rate, not sentiments about where things seem to be heading.
Readers also found this companion piece from our archive useful.
Pre agreement is the entire value. The meeting where a price rise is debated for the fourth time while entries clear at the new rate is the failure this method exists to prevent. When the trigger fires, the owner executes the agreed move and reports. Debate was last quarter, on purpose, when it was cheap.
Keep the set alive
Review scenarios on a fixed cadence and whenever a policy event lands: retire futures that events have killed, add the new plausible, recost the survivors. Archive the old sets rather than overwriting them, because comparing what you expected with what arrived is how the next set gets smarter about ranges, timelines, and your own organisation's reaction speed.
Over a few cycles the practice changes the culture around tariff news. Announcements stop being emergencies and become routing decisions between prepared paths. That calm is worth more than forecast accuracy, which nobody in this field honestly has. You are not predicting the weather. You are packing for the plausible kinds.
Frequently asked questions
How many scenarios should we maintain?
Three or four per major product group is the practical range: status quo, the announced change as published, a softened version such as exclusions granted or partial rollback, and a severe case if one is credibly discussed. More than that dilutes the decision value.
What triggers should activate a scenario plan?
Observable policy events: publication of final legal text, an exclusion decision, an effective date passing, or the first entries liquidating at a new rate. Triggers must be verifiable by anyone, so execution starts without a debate about whether conditions have been met.
How detailed should per product costing be?
Full landed cost per scenario for the products that carry the exposure: duty on the correct basis, duty linked fees, freight effects, and margin at a realistic selling price. Cover the products that drive most duty spend in depth and group the long tail.
Does scenario planning predict tariff policy?
No, and it should not try. Its purpose is pre decision: agreeing actions for each plausible future while there is time to think. When policy moves, the organisation executes a prepared path instead of improvising, which is where most tariff cost surprises actually come from.