The Great SKU Diet: Apparel Is Cutting the Rack, and Betting the Margin on Fit

On September 4, Lululemon told the market it had taken roughly 15% of the products out of its stores, describing the goal as a less dense shopping environment, and it is rolling that change across the fleet. In the same breath it slowed the build: about 35 net new company-operated stores this year rather than the 40 it had planned, roughly 10 of them in North America, and around 40 pop-ups at year end against 65 a year ago.

None of which is a retreat from volume. Management said in the same update that it is chasing about 20% more product volume than last year, which in practice means more open-to-buy held back for follow-on orders on the styles that actually sell. Fewer things, more of each. That trade has become the apparel operating model of 2026, and the trade press has already given it a name: the great SKU diet.

The edit is the visible half of that strategy. The half that decides whether it works is the size curve underneath it, the split of units across sizes behind every style that survives the cut, and that curve happens to be moving right now for reasons with almost nothing to do with merchandising taste. What follows is an attempt to read the diet the way a finance team would: where the margin sits inside a narrower assortment, and the one place the whole exercise does not reach.

1. The Diet Has a Name Brand Attached

SKU rationalization has been a consultant's slide for years, and it became a headline the moment a brand of Lululemon's size did it in public. This was not a category trim. Product count came down across the fleet, store openings slowed, the pop-up estate shrank by more than a third, and the money that freed up went into chase capacity on proven sellers.

The pressure behind that decision is legible in the same quarter's numbers: Canadian revenue down 11%, United States revenue down 8%, and women's leggings, the franchise the brand was built on, off roughly 20%. When the core slows like that, width stops reading as optionality and starts reading as carrying cost.

The arithmetic is unforgiving in apparel because the margins are thin before anything goes wrong. Eightx puts public apparel operating margins for the 2025 fiscal year between 4.08% at American Eagle and 19.9% at Lululemon, with the direct-to-consumer midpoint sitting near 6%, while inventory days across public comparables run anywhere from 65 to 147 against a target of 30 to 60 for a private brand doing $5 million to $50 million. At a 6% operating margin, a season of dead width is not a rounding error. It is the year.

So the playbook looks much the same wherever you turn this autumn: narrower ranges, sharper in-season reads, and replenishment that moves before the markdown does. The theory is sound. The execution risk sits one level further down.

2. The Curve Beneath the Cut

Cutting styles is the easy half of the decision. Working out how many smalls, mediums and extra larges to buy behind the styles that remain is the half that pays or does not, and it is precisely where the industry is currently out of position.

Impact Analytics estimates that roughly 400 million apparel units are at risk of misalignment with actual demand, carrying about $5 billion in inventory value and margin exposure between them. Set against that, it puts the upside from better size precision at 130 to 210 basis points of gross margin growth and a 3% to 5% cut in markdown sales. For a brand running a 6% operating margin, 130 basis points of gross margin is not an optimization. It is a different year.

What makes the problem hard to eyeball is the sensitivity. A single percentage point of share lost at the large end shifts roughly 120 million units a year across the market, and a three-quarter point decline at medium moves about 90 million. Medium carries the heaviest volume weight in most assortments, so by the time the pressure reaches it, the error is no longer sitting at the fringe of the curve. It is sitting in the middle of the buy.

There is a channel split worth planning around as well, since Impact Analytics finds physical stores migrating toward smaller size bands materially faster than digital. A single national curve applied to both will therefore be wrong in two directions at once, short on smalls in the stores and long on larger sizes online, with markdown waiting at either end.

Eightx runs the math on one buy, and it clarifies the stakes. Take 1,000 units at $80 retail and $32 landed, a 60% gross margin, with the fringe sizes eventually marked half off: buying that style on a flat curve rather than an optimized one gives up about $4,600, or roughly 11% of the gross profit on the buy. Extend the same error across a season and the analysis estimates that poor size-level visibility costs apparel retailers up to about 20% of profit. The rack looks fine either way. The curve is where the money went.

3. Why the Curve Is Moving

Size curves are built on history, and history has rarely been a poorer guide, because the bodies underneath the curve are changing faster than the planning cycle that describes them.

GLP-1 medication is the reason. eMarketer puts current United States use at about 16 million people, nearly triple the 2022 figure, with projections running as high as 35 million by 2030, and Retail Dive reported in April that nearly a quarter of United States households had a user as of last autumn, more than half of whom had already shopped for new clothing in anticipation of a size change. The national obesity rate has come off its 2022 peak of 39.9% to 37.0%, which is a modest move in public health terms and a very large one inside a size curve.

The individual effect is more dramatic than the aggregate suggests. Bold Metrics describes typical use as 20 to 50 pounds of loss, which is one to five full sizes, and finds that 80% of users expect to need new clothing. The same analysis notes that 46.5% stop the medication within twelve months, and that weight tends to return within roughly 1.7 years of stopping.

Read those two findings together and the planning implication is the whole story. This is not a one-time step change to a new and smaller curve. It is a population moving through sizes in both directions on a multi-year cycle, which means a brand that re-plans its curve once and treats the answer as settled will be wrong again within two seasons, in the opposite direction.

4. The Overcorrection Already Under Way

Some brands are not waiting for the curve to settle. H&M has removed 3XL and 4XL from its United States women's online assortment, and Aritzia, Mango and Zara have each reduced the number of larger-size items they carry. The retailers most exposed to the top of the curve felt it first: Retail Dive reports Destination XL sales down 7% against 2024, with executives describing the medication's effect as a larger problem than they had anticipated, and Torrid net sales down more than 9% over the same year, while Bold Metrics puts Destination XL's fourth-quarter decline at 6% and notes that as many as a quarter of its customers use the medication and are postponing purchases until their size settles.

That word, postponing, is the one to sit with. A customer who delays a purchase has not left the category. She is waiting, and cutting her size out of the assortment is what turns a delay into a defection.

Liza Amlani of Retail Strategy Group told Retail Dive that brands are overcorrecting, with some planning to eliminate extended sizing outright, and the counter-evidence sits in the same research. The obesity rate is still 37%. Active users of these medications plan to increase spending on casual clothing by 19%, footwear by 15% and activewear by 10%, 48% report buying more premium clothing than before, and Bernstein sizes the incremental opportunity at up to $13 billion a year in apparel spending. The customer changing size is buying more clothes, not fewer. Whether she buys them from you depends on whether you kept her size while she moved through it.

Amlani also named the structural reason this keeps going wrong. In her description, "fit lives in design and size lives in planning", and the two functions rarely talk to each other. The curve is a planning artifact, the fit that sends the garment back is a design one, and nobody owns the seam between them.

5. The Bill the Diet Does Not Pay

Which brings us to the limit of the whole exercise. A narrower assortment with a well-planned curve does two genuinely valuable things: it reduces the chance that the wrong size is the only size in stock, and it cuts the markdown at the end of the season. Both are real money. Neither of them touches an order that has already been placed.

Size and fit remain the largest single cause of returns in apparel, sitting behind roughly half to two thirds of them depending on whose compilation you read, with category return rates running 20% to 30% and higher still in fast fashion. Bracketing, the practice of ordering several sizes with the settled intention of sending most of them back, has become ordinary behavior rather than an edge case.

Notice that bracketing is immune to everything described above. A leaner rack does not stop it, a perfect size curve does not stop it, and complete fit data does not stop it either, because the shopper ordering a small and a medium is not confused about her size. She is outsourcing the fitting room to your reverse logistics. Virtual try-on helps at the margin, and it helps most with the shopper who is genuinely unsure.

The season only makes the bill larger. As we covered in the 2026 holiday forecast, the National Retail Federation expects 17% of holiday sales to be returned, with the online rate at 19.3%, and every one of those units will have been picked, packed and shipped under a peak carrier surcharge before coming back to be restocked in January against a curve that has moved again in the meantime.

So the diet is an inventory decision, made once a season, about what sits on the shelf, while the return is an order decision, made thousands of times, about what sits in a cart. They are different problems, and only one of them is solved by buying better.

6. A Checklist for Apparel Merchants

Five moves, in the order they pay.

First, plan the curve by channel. Stores and digital are drifting apart, so a single national curve now guarantees an error in both. Split them before the spring buy.

Second, re-plan the curve more than once a season. The demand signal underneath it is a moving cycle rather than a step change, which makes last year's curve a worse guide than it has ever been, and an annual planning rhythm simply cannot see the move in time to act on it.

Third, buy narrow and hold chase capacity rather than width. This is the part Lululemon got right, and it is the part most brands will skip, because the cut only pays if the money it frees goes back into follow-on orders on proven sellers. A diet that merely buys less is a sales forecast reduction wearing a margin costume.

Fourth, keep the sizes at the ends while the customer moves through them. The evidence says the shopper changing size spends more, not less. Trim depth at the ends if the math demands it, but understand the difference: deleting the size deletes the customer.

Fifth, put a decision at the order. Assortment planning works on the shelf, weeks or months before anyone buys, whereas the signals that separate a good order from an expensive one are present at checkout, in the cart itself, before anything ships. That is a different instrument from the buy plan, and most brands do not have one.