Price pack architecture: what it is, how to build it and how to audit it

Three numbers describe every rung. Compute them for each pack, in order of size.

The 27 oz pack costs 10.0% more per ounce than the 18 oz. At the 18 oz unit price, 27 oz would cost $8.10. The shelf asks $8.91, so a shopper who trades up to save pays $0.81 extra. Figure 1 shows the same ladder.

Unit price should fall as pack size rises. That expectation is common enough that the UK Competition and Markets Authority (CMA) tested it on a basket of 26 grocery items from 11 retailers on one day in May 2023. Bigger packs often had lower unit prices, but the savings were not consistent from one step to the next, between products or between retailers, and some larger packs cost more per unit. EY makes the shopper’s side of the same point: a larger pack that does not promise a lower price per kilogram than the smaller one can leave a negative brand impression.

Beyond the direction, no primary source we found gives a correct step size for consumer goods. The figures vendor pages quote (5% to 10% between sizes, 10% to 15%, 40% between tiers) come without data, and the 40% is a rule for software tiers, not pack sizes. What can be said:

• One rule is firm. A larger pack should not cost more per unit than a smaller one in the same container and format. That is the inversion test.
• Everything else is relative to the category. Measure the distribution of step discounts across the brands in the category and treat the tails as the questions.
• Normalize steps of different widths. A step from 12 oz to 18 oz (1.5 times) should save less than a step from 12 oz to 24 oz (2 times). Expressing every step per doubling of size makes them comparable. Shelfprice’s default screen expects 10% to 15% off per doubling and lets a category override it. Treat that as a screening convention, not a published standard.

Five ways a ladder breaks

A break is a question, not a verdict. An entry pack priced high per ounce is working as designed. A large pack priced to match a club-store comparison can be deliberate. What turns a break into a defect is that nobody chose it.

A new architecture, or a redesign, goes in this order. Each step narrows the next.

1. Start with occasions and budgets. Who buys the pack, for what use, and what shelf price will they pay? The entry price point comes from the shopper’s budget, not from a unit price.
2. Choose the core pack and fix its price. Every other pack is indexed to it.
3. Add the other roles at sizes that serve a real occasion. Keep each a visible step from its neighbors, so the ladder has no crowds.
4. Set unit prices by role. Entry highest, stock-up lowest. Choose each step as a discount per doubling of size and compare it with the steps in your category.
5. Check the shelf price points, then rivals and the store brand at the same size, against any price thresholds you have evidence for.
6. Ask who sets the shelf price. A recommended price the retailer does not follow is a different ladder. Recompute with the shelf prices you expect.
7. Test before a wide launch, as described below.

Step 4 is arithmetic. A discount of 12% per doubling sits inside Shelfprice’s default 10% to 15% band. Anchored on an 18 oz core pack at 30.0¢ per ounce, it gives the ladder below before prices are rounded to shelf price points.

Inversions and quantity surcharges

Economists call a larger pack priced above the smaller one per unit a quantity surcharge. Sprott, Manning and Miyazaki studied how grocery prices are set, and where surcharges appear, in the Journal of Marketing in 2003. A Purdue release from 1999 reports that, in a study of canned tuna, larger sizes were often 20 percent more per ounce. A 2026 study of 38 New Zealand supermarkets and 40 products found about one product in four, 26 percent, with a surcharge. The CMA found larger packs dearer per unit in toilet roll, cheese and laundry pods across more than one retailer, and called such cases relatively unusual.

The rates differ because the categories, countries and counting rules differ, so we do not quote one. What the studies agree on is that the pattern exists in ordinary grocery pricing. Here is one case from the CMA data.

Promotions are not architecture

Audit regular prices. A deal on one pack changes the ladder for a week, and EY notes that promotions can make two small packs cheaper than a larger one. Take the ladder from Example 1 and put the 12 oz pack on a deal at $3.30. Its unit price falls to 27.5¢, which is 8.3% below the 18 oz pack at 30.0¢. For that week the ladder runs the wrong way, and the 12 oz has become the best value on the shelf. Nothing in the structure changed. Count how often it happens, how deep it goes and which packs it hits as a separate promotion analysis, and do not let it contaminate the structural one. The CMA found the same in its data: promoted items were often, but not always, the cheapest per unit.

Price thresholds

Demand does not respond to price along a smooth curve. EY describes price thresholds, points where shoppers’ reaction is sharp, and says that this is why manufacturers who cannot push a price across a threshold change the pack instead. That is the logic behind shrinkflation, and it makes pack size the second price lever.

A break has a limited set of repairs. Each belongs to a party, and each has a side effect.

A repair, worked through

Return to Example 1. The 27 oz pack is 10.0% dearer per ounce than the 18 oz. Priced at the 18 oz unit price it would cost $8.10, which removes the inversion but leaves a step of zero. Priced at $7.49, it costs 27.7¢ per ounce, a step discount of 7.5% against the 18 oz, and an index of 92.5.

Price move or pack move: a cash-margin model

Take a pack that sells for $4.00 and costs $2.00 to make, so margin is $2.00 a pack. Two moves keep the shelf price at $4.00. Shrinking content by 10% raises margin per pack to $2.20 and raises the price of each unit of content by 11.1%. Adding 10% content cuts margin per pack to $1.80 and lowers the price of each unit of content by 9.1%.

Volume then responds. The model assumes volume changes by elasticity times the change in the price of each unit of content, and cash margin is packs sold times margin per pack.

This matches the direction EY reports for a 50% margin: shrinkflation helps only when elasticity is low, and once demand is elastic only adding content at the same price grows cash margin, because the volume gain outweighs the lower margin per pack. The numbers are our own model, not EY’s. Real elasticity is not constant, especially across a price threshold, so use the model to see which way a pack move pushes margin and what elasticity it needs, not to forecast volume. Measuring the elasticity of a specific pack is a separate piece of work.

Test before you roll out

L.E.K. describes the usual sequence for designing a new pack: diagnose the category, test options with a consumer conjoint survey, then simulate the shelf to estimate volume lift before committing. A pilot in a few stores is the simple version. The same note reports a soup pack of just over 15 oz priced 10% below the standard 22 oz can and carrying about 30% more per ounce. Those are the consultancy’s own figures, but the point holds without them: a small pack priced higher per ounce can be deliberate, and a ladder audit should ask before it corrects.

1. Define the set. One brand, one subsegment, one container and format, one retailer or channel at a time.
2. Build pack points. Roll flavors up to brand, subsegment and total size. Put total quantity in one unit. Mark any pack whose size you had to guess.
3. Take regular price. Leave out a pack with no regular price and name it. Do not substitute its deal price.
4. Compute unit price, step discount and the index to the core pack for every rung.
5. Flag. Every negative step first. Then steps far from the category’s usual, gaps and crowds. Use the distribution across the category as the yardstick, not a number from a web page.
6. Check each flag against a role and a lever. Drop the deliberate ones. For the rest, name who can move the price, the size or the count.
7. Repair the ladder as a whole, recomputing after each change, then run the promotion pass separately.

1. Comparing pack prices instead of unit prices, or comparing weight and volume units without a conversion.
2. Reading a deal as architecture. One promoted pack inverts a sound ladder for a week.
3. Treating the recommended price as the price shoppers see.
4. Counting flavors as separate packs, which makes the ladder look crowded.
5. Judging an entry pack on unit price alone. Its job is a low absolute price.
6. Applying a grocery ladder to club, convenience or dollar channels, where pack roles differ.
7. Averaging unit prices of different pack counts. An average price per ounce is total dollars over total ounces.
8. Fixing one rung without recomputing its neighbors. A repair can create a new inversion or a crowd.
9. Choosing shrinkflation by default, without elasticity and margin.
10. Assuming every surcharge is an error. Find out whether someone chose it before correcting it.

• Shelf prices do not show volume, cost or elasticity. They find the candidates. They cannot say which repair pays. That takes sales data, a margin file and a test.
• One retailer is not the market. A ladder that works in grocery can fail in club or convenience, and a single chain’s shelf reflects its own pricing choices.
• Packs that are not substitutes. A travel size and a family size serve different occasions. A unit-price comparison between them is arithmetic, not a finding.
• Concentrated products. Price per ounce favors dilute products. Use a per-use measure as a second, labeled metric.
• Short ladders and close sizes. Two rungs give one step and no curve. Stretching a step between nearly equal sizes to a full doubling blows a small price difference up into a huge percentage, so Shelfprice does not call a step flat or steep unless the larger pack is at least 1.3 times the smaller.
• Misread sizes. A pack size parsed wrongly from a product title creates an inversion that does not exist. Validate before you flag.
• The per-doubling convention. A constant percentage off per doubling fits some categories better than others. Fit the category before you adopt a band.

Shelfprice runs steps 2 to 5 of the audit on Kroger group shelf prices for a brand’s competitive set: pack points with flavors rolled up, regular price for architecture, and deals held apart as their own finding. It plans a ladder’s repairs as one consistent set of moves instead of one suggestion per step. It sees shelf prices and sizes for the stores it samples. It does not see your sales, costs or elasticity. The PPA analysis page describes the product.