Price Is the
Last Lever

Ingredient and input prices move on a lag of roughly 12 to 18 months, which places the real weight of 2025 trade policy squarely between April and October 2026. Businesses absorbed much of that first wave through margin rather than price. But that absorption is what's breaking now.

The rate itself keeps reconstituting. It ran from about 2.4% in early 2025 to a peak near 28% that spring, and stands at 12.1% in July 2026. After the Supreme Court struck down one set of tariffs, the rate was reworked through other legal authorities; under current law it is now set to ease to roughly 9.8% by year-end, though announced measures could push it higher. The number moves, but the direction of travel for a manufacturer's cost base does not: it has reset several times above the 2.4% baseline.

General food inflation adds to the squeeze from a different direction. The USDA projects all food up 3.2% in 2026 and food-at-home up 2.8%, with beef up 7.5% on a cattle herd at a 75-year low. Eggs are the one category headed down, off 30.4% as flocks recover. The point is not the cause of any single line. It's that the cost base a manufacturer buys is climbing across almost every input at once.

Watch how the disciplined operators handle it. In fiscal 2026 Conagra carried a tariff headwind of roughly 3% of cost of goods, much of it on the tin plate steel and aluminum its cans and lines depend on, and worked to offset much of it through accelerated cost savings, alternative sourcing, and targeted pricing. Its response isn't a blanket price increase. The company is using targeted pricing, sorting its portfolio and pushing price only where a category can hold it, while managing its canned staples for cash.

The reason is elasticity. Today's shopper trades down faster than yesterday's, and the big brands are pricing for it. Looking ahead to fiscal 2027, Conagra is guiding to mid-single-digit volume declines, especially in frozen, assuming shoppers react more sharply to price than they historically have. Private label has even less room to move. Its whole advantage is the distance between its price and the national brand's, so passing a tariff straight through erases the reason a shopper reached for it. Raise price across the board and you protect the margin on each unit while losing the units.

When McCormick first modeled its tariff hit, the gross annualized number reached about $140 million in 2025. For fiscal 2026 the company now expects roughly $50 million of incremental impact. It didn't get there by raising prices to cover the rest. It got there upstream.

McCormick shifted procurement toward lower-tariff countries, pulled productivity savings across the P&L, and used pricing selectively rather than across the board. The pricing it did take weighed on near-term margin on purpose, because it chose to hold the demand base going into 2026.

The work behind a move like that is analytical before it's operational: modeling landed cost country by country, so a buyer can see which origins carry the tariff and which don't before a single contract shifts. This is the first place AI earns its keep against a tariff, mapping exposure across a supplier network faster than a team can by hand, well before it ever touches a forecast.

Notably, however, part of 2026's relief was a one-time legal refund, not operational mitigation. After the Supreme Court struck down the IEEPA tariffs, McCormick recovered about $28 million and Conagra about $6 million. Those refunds do not repeat. The durable gains, the ones that hold whatever the courts decide next, come from sourcing and productivity.

Sourcing has limits, and McCormick named one. Some raw materials are not available from US suppliers at all, so procurement can only move so far. The principle still holds. A dollar of tariff removed at the border beats a dollar pushed onto a price-sensitive shopper. One protects margin and volume together. The other protects margin and risks the volume.

The next lever is quieter, and it's where the capital is going. A tariff raises the price of an input. So does buying that input at the wrong moment, in the wrong quantity, on a guess. When commodity, metal, and freight markets move this fast, forecasting error becomes its own cost line, and it is one you control.

This is the job AI was built for, and it helps to be exact about it. It reads supplier history, commodity and freight signals, weather, and shipment risk, then tells procurement what to buy, how much, and when. It is the difference between locking a price ahead of a tariff and paying the spot market after it.

Procurement leaders are already moving. In the Hackett Group's 2026 Procurement Key Issues Study, 80% named AI-enabled technology the most transformational force in their function over the next five years.

The spending follows. Conagra is raising capital investment to about $550 million in fiscal 2027, up from $423 million, aimed at supply chain modernization and in-sourcing rather than at discounting through the cycle. Its Project Catalyst program leans on AI to strip out cost and complexity. The firms with the most exposure are investing to buy and run smarter, not to cut price faster.

Conagra can commit $550 million and McCormick can re-map a global supplier network. Most manufacturers reading this can't, and the instinct is to conclude the playbook is for someone else. It isn't. The logic scales down cleanly; only the scope changes. You don't have to model every input. You have to model the few that move the most.

Start with your top three volatile inputs, the ones where a tariffed origin or a badly timed buy does real damage, and work only those. Map their exposure by supplier and country. Qualify one or two alternate origins on each. Put a disciplined forecasting cadence on their purchase, even a lightweight AI-assisted one, so you are buying on signal instead of on habit. A mid-market operator running that on three inputs captures most of the margin a Conagra captures across three hundred.

What makes the discipline hold is measurement. Track your own input cost base against a benchmark like the FIE Input Cost Index, so you see a cost move forming before it lands on the P&L, while sourcing and forecasting still have time to work.

The operators who protect margin aren't the ones who react fastest at the shelf. They are the ones who saw the input cost coming.