Price moves that move nothing: a retail price optimization memo

Two numbers landed on my desk this autumn, and together they describe the pricing problem most mid-market retailers will carry into Q4. The first: a major snacks and beverages manufacturer raised prices 1.3% and got 0.6% value growth for it. The second: only 22% of US shoppers now believe an advertised discount reflects a real saving. A price increase that does not stick and a discount that nobody believes are the same failure seen from two sides. This memo is about why that is happening and what retail price optimization has to look like if it is going to work at all.

Two numbers from this autumn

The first number comes from NIQ's Pricing under pressure analysis, published in September and updated in mid-October. Its Global Inflation Tracker (225 categories, 58 countries) puts total CPG inflation at 2.7%. Global FMCG value grew 1.8% in the 52 weeks to March 2026, but unit volume grew only 0.7%. The gap is price. And the price is not landing cleanly: in one worked example, a 1.3% increase produced 0.6% value growth. In another, a 0.3% increase was followed by a 0.1% decline in value.

Inside the same portfolio, NIQ found regular price elasticity ranging from about 1.1 to nearly 2.0, and it varied again by channel. Read that twice. A 2% increase on the item at 1.1 costs you a little volume. The same 2% on the item at 2.0 costs you roughly twice as much, and in some channels you end up with less revenue than you started with. One blanket number across both is not a strategy. It is an average that is wrong for every item it touches.

The second number comes from the other end of the shelf. Simon-Kucher's 2026 Holiday Shopping Study of more than 1,500 US adults found that only 22% fully trust advertised savings. Retail Dive reports that the same figure was 46% a year ago. Trust in the discount halved in twelve months. The same study has US households planning to spend 5.5% less this season, with 62% shopping earlier to spread the cost.

Other forecasters are more optimistic on the headline. NRF expects holiday sales to grow 4.2% and pass $1 trillion. But Bain, quoted in the same Retail Dive piece, says inflation accounts for more than half of that nominal rise, and Deloitte finds nearly 80% of shoppers expect higher prices on holiday purchases. Whichever forecast you prefer, the shopper arrives at your store already assuming the price has gone up and already doubting the sticker that says it has come down.

Why blanket price moves stop working

For most of the last three years, pricing in mid-market retail has been managed as a portfolio-level decision. Cost inflation arrived, leadership agreed a percentage, the percentage went into the system, and the team moved on. That worked while every shopper was absorbing increases everywhere and had no way to compare. It stops working for three reasons that are all visible in the NIQ data.

Elasticity is an item property, not a category property. A 1.1 to 2.0 spread inside one portfolio means the "right" increase for the category is wrong for most of its items. The item at 1.1 could have carried more. The item at 2.0 should not have moved at all, or should have moved with a pack change instead. NIQ's diaper case study makes the point: facing low-cost rivals, the brand kept value sales at 100 and raised profit to 107 by making small pack-count changes rather than cutting price. That is a decision you can only make one item at a time.

Shoppers now mix tiers inside one basket. NIQ's 2025 consumer work found 54% willing to choose premium and 68% seeing store brands as a good alternative to established brands. These are the same people. They buy the premium coffee and the private-label beans in one trip. So a uniform increase does not move them out of your store. It moves them across your shelf, from the item whose price you raised to the one whose price you did not, and your mix does the rest. The value growth you expected is eaten by the trade-down you created.

The increase is now checkable. Croatia makes retailers show a dated anchor price next to the shelf price. Price-history features are now built into the shopping tools consumers use every day. The pressure in the headline rate is still there, but the shopper's ability to see which specific items carried it has gone from nothing to near-total in about two years. The blanket increase was always wrong on the elasticity maths. Now it is also visible.

The discount side of the same problem

The 22% number is the mirror image. A discount only works if the shopper believes the reference price was real. Three years of reference-price inflation, "was" prices that were never charged for long, and promotional calendars that never end have taught shoppers to discount the discount. They have been proved right often enough that half of them now assume the saving is fiction.

The instinct going into a soft Q4 will be to go deeper. If 20% off is no longer believed, try 30%. That is the blanket-increase mistake run in reverse. Deeper discounts on items shoppers already distrust do not restore trust. They confirm the suspicion that the original price had room in it, which makes the next full-price week harder, which makes the next discount deeper. Simon-Kucher's headline finding is that over half of consumers are cutting non-essential purchases or trading down to lower-priced brands. A deeper percentage does not change that. A credible one might.

Credibility has a mechanical definition. The reference price must be a price the item sold at, recently, for a meaningful period. The discount should be on items where the elasticity justifies it, so the volume response pays for the margin given up. And the retailer should be able to explain, if asked, why this item is on promotion this week and that one is not. The Dollar General settlement in New York last week is a reminder that regulators now check whether the price on the shelf, the price in the ad and the price at the register agree. Promotional credibility is the same discipline applied to the "was" price.

What retail price optimization means in practice

The phrase has been stretched to cover everything from a spreadsheet macro to a black-box model that reprices hourly. For a retailer with 10,000 to 200,000 SKUs and a pricing team of two to five people, it means something more specific. Retail price optimization is the discipline of setting price one item at a time, inside rules the business has agreed, with a reason attached to every change.

That has four parts, and all four are needed for Q4. (If you are evaluating tools, our guide to price optimization software covers what to ask a vendor; this memo is about what to ask your own team.)

Item-level elasticity, not category averages. You need a working estimate of how each item responds to price, by channel where the channels differ. It does not have to be perfect. NIQ's chocolate case study tested four scenarios (do nothing, downsize, raise price, innovate), chose a combined path, and the model's forecast landed within plus or minus 9% of the actual result. A 9% error band on a decision you would otherwise make blind is a large improvement. The point is to stop treating 1.1 and 2.0 as the same number.

Roles before rules. Not every item needs an elasticity debate. The 150 or so lines that set your price image get one treatment. The long tail that nobody compares gets another. Items in between get a third. We wrote about picking those lines when the discounter cuts 1,200 lines. The same logic applies when cost goes up: the known-value items carry less of the increase, the tail carries more, and the rule that says so is written down.

Rules that produce a price, not a suggestion. "Stay within 3% of the market leader on KVIs, hold margin floor at 22% on the tail, cap any single increase at 5% per quarter" is a pricing policy a CEO can read and a system can execute. The output of the rule is the price. The exceptions are the items where the rule could not be satisfied, and those are the ones a human looks at. That is how a team of three manages 80,000 SKUs without either rubber-stamping everything or reading every row.

A reason on every change. This is the part most tools skip and most retailers will need most this winter. Every price that moves should carry its rule, its inputs and its approver. Partly that is for the regulator and the anchor-price law. Mostly it is for you: when the 1.3% increase returns 0.6%, the question is not "did it work" but "which items did it fail on and why", and that question is unanswerable if the increase was one number applied to everything.

NIQ's recurring recommendation across its case studies is to simulate before acting: price, pack and promotion scenarios tested against the elasticity estimate before the change goes live. For a mid-market retailer that used to mean a consultant and six weeks. It should now mean a grid, a rule and an afternoon.

Four asks for the CEO this quarter

None of these require a new system to be live before Black Friday. All of them change what the pricing team does in the next four weeks.

1. Ask for the elasticity spread, not the elasticity. Have the team show you the range across your top 500 items, by channel if you have more than one. If the answer is a single number, that is the problem. If the spread looks like NIQ's 1.1 to 2.0, the blanket increase you approved in spring is where the margin went.
2. Ask which items carried the last increase and what happened to them. Volume, mix and value, item by item, against the items that did not move. This is the post-mortem on the 1.3% that returned 0.6%. Most teams have never run it because the increase was applied as one number and nobody kept the list.
3. Ask for the promotional reference-price audit before the holiday calendar locks. For every planned promotion: what was the regular price, for how many of the last 90 days was it charged, and does the "was" price survive the reference-price test a regulator or a price-history tool would apply. Pull the promotions that fail. A shorter, credible promo calendar will outperform a longer one that 78% of shoppers assume is fake.
4. Ask for the rule in writing. One page: roles, floors, caps, competitor bands, approval thresholds. If it cannot be written in a page, it cannot be executed by a system or explained to a shopper, an auditor or your board.

What this does not change

Item-level pricing does not make cost inflation disappear. If your input costs are up 4%, the money has to come from somewhere, and some of it will come from price. What changes is where. The increase lands on the items that can carry it and is kept off the ones that cannot, and the decision is made with the elasticity estimate in front of you rather than behind you.

It also does not make promotions obsolete. Shoppers still respond to a real saving on an item they wanted. The 22% figure is not a verdict on discounts. It is a verdict on reference prices that were never real, and that is fixable.

And it does not require perfect data. The retailers we talk to usually have two years of transaction history, a cost file and a competitor feed of varying quality. That is enough to estimate elasticity for the items that matter, assign roles, and write the rule. The perfect model can come later. The blanket number cannot wait until it does.

If you want to see where your own price moves have been landing, our free pricing audit takes a product list and a competitor set and shows the spread. Or if you would rather talk it through with someone who has run the post-mortem on a few hundred price increases, get in touch.