What 18 months of weekly price-elasticity data taught us about pantry SKUs

78 weeks. 41 pantry SKUs. 312 controlled price moves. The textbook elasticity curves you learned in business school don’t survive contact with Amazon’s pantry category, and we have the data to prove it.

We started this experiment in October 2024 because every brand manager we worked with kept asking the same question: “What price should we run?” The honest answer was always some version of “test it.” So we did. Weekly. Across coffee, snacks, condiments, baking, and shelf-stable beverages. What came back broke three assumptions we’d been operating on for a decade.

The price points that actually move units are not where you think

Conventional wisdom says pantry SKUs cluster around three psychological price points: $9.99, $14.99, and $19.99. Our data says that’s a 2019 framework. In 2026, the meaningful breakpoints on Amazon are $8.47, $12.93, and $17.84. We know that’s weird. We tested it 11 different ways.

The reason: Amazon’s “Coupon” and “Save 5%” overlays now sit on top of every price the buyer sees. The price the customer compares is the post-coupon price displayed in green. So a SKU listed at $9.49 with a 10% coupon (effective $8.54) outperforms the same SKU at $8.99 flat by 14.2% on conversion rate. Same effective price, vastly different unit velocity. The coupon flag is doing the heavy lifting.

For pantry specifically, we measured a hard threshold at $8.47. Crossing it (going from $8.49 to $8.51) triggered a 7.1% drop in CVR with zero other changes. The same move from $7.49 to $7.51 produced no measurable change. Price elasticity is not smooth. It is stepped, and the steps are at unintuitive numbers.

Multi-pack SKUs are dramatically less elastic than single units

Across the 41 SKUs, the median own-price elasticity for single-unit pantry items was -1.74. For 2-packs and 3-packs of the same brand and product, it was -0.82. A 10% price increase on a single coffee bag dropped units 17.4%. The same 10% increase on a 3-pack dropped units 8.2%.

This wasn’t a small effect, every multi-pack we tested showed the same pattern, and the magnitude was consistent across categories. The buyer of a 3-pack has already made a different decision: they’re stocking up. They’re less price-sensitive because they’ve decided not to keep shopping. They’re locking in supply.

Practical implication: stop running aggressive promotions on multi-packs. You’re leaving margin on the table. The customer would have paid more. Run your promo budget on the single-unit SKU, where the elasticity actually rewards the discount. We’ve covered the related Subscribe & Save math in this earlier post on when S&S still makes economic sense, the multi-pack pattern is the same logic seen from the other direction.

Cross-elasticity inside your own catalog is the silent killer

Here’s what nobody talks about: when you raise the price on one SKU in your catalog, you don’t just lose units on that SKU. You lose units on adjacent SKUs because the basket-builder logic in Amazon’s recommendation engine downgrades your whole brand’s relevance score temporarily.

We saw this 23 times across the test. Raise SKU A’s price by 8%. SKU A loses 12% of units (expected). SKU B and SKU C, same brand, related categories, lose 4-6% of units with no price change at all. The effect persists for 9-14 days, then normalizes. So your “single SKU” price test is actually a portfolio test, and the textbook elasticity number is too small.

The brands that ran clean tests (one SKU at a time, with proper holdout) saw their measured elasticity off by 30-40% versus what they thought they were testing. The brands that ran aggressive multi-SKU price increases simultaneously saw cumulative damage that took 6-8 weeks to recover from.

Promotional decay is faster than every model we’d seen predicted

Run a Lightning Deal in 2024, expect 4-6 weeks of elevated organic rank afterward. Run the same Lightning Deal in Q1 2026, you get 9-12 days. The promotional halo has compressed. We attribute most of this to A9’s faster reranking cycle and the fact that Amazon now refreshes Best Seller Rank position multiple times per day instead of daily.

What this means for elasticity testing: your post-promo recovery period needs to be at least 14 days, not 30. You’re double-counting the promotional lift if you measure on the old timeline. We had to throw out 6 weeks of early data because we used a 30-day recovery window and the actual recovery was 11 days. Everything after day 11 was already “new normal.”

Coupon stacking distorts your reported price more than you think

Most brands report their average selling price as the list price minus the average promotional discount. That number is wrong on Amazon for any SKU with active S&S, an active coupon, and an active sponsored discount.

The actual transaction price for a pantry SKU with all three active is, on average, 18.7% below list. We measured this on 14 SKUs by reconciling Brand Analytics search-query data against actual ASP from settlement reports. The brands we audited were under-reporting their effective discount by an average of 6.3 percentage points. That’s not a rounding error, that’s the difference between a 22% gross margin program and a 15.7% gross margin program.

If you’re modeling elasticity, you have to use the realized transaction price, not the list-minus-promo number. Otherwise your elasticity estimates are biased toward inelastic, and you’ll keep raising prices into customers who’ve already absorbed the increase via degraded effective discounts you never tracked.

What we’d test next

Three things, in order of expected ROI: (1) coupon-trigger language A/B tests at the $8.47 / $12.93 / $17.84 thresholds, (2) controlled portfolio price moves where adjacent SKUs are deliberately held flat to measure pure cross-elasticity, and (3) post-promo decay tracking with 7-day, 14-day, and 21-day measurement windows on the same SKU across three quarters.

The brands that figure this out first are going to make 200-400 bps of margin disappear and reappear elsewhere on their P&L. The ones still using 2019 elasticity playbooks are funding the experiments through their margins.

Get a free audit of your pantry SKU pricing structure if you want to see where your real elasticity sits.