Key takeaways:
- Sizing is often siloed within retail organizations, leaving no clear owner and creating a persistent source of margin erosion through stockouts, overstock and markdowns.
- Retailers should be cautious about attributing broad sizing shifts to GLP-1 drug adoption; the data doesn’t yet support population-wide change, and over-indexing on that narrative can obscure real planning problems.
- Advances in AI and cloud-native data infrastructure now allow planners to build far more granular size curves, though human expertise remains essential to steering those models.
Sizing doesn’t get a seat at the table the way assortment planning or allocation does. It’s granular, data-intensive and organizationally murky. Depending on the retailer, it might live in merchandising, planning, allocation or its own team. The result: core sizes sell out, fringe sizes pile up on clearance racks and nobody treats it like the money problem it actually is.
James Theuerkauf, VP of Product at Anaplan and co-founder of Syrup Tech, which Anaplan acquired in 2025, has spent much of his career examining exactly this problem. In a recent special Deep Dive episode of the Retail Remix podcast, he spoke with host Kate Robertson about the real cost of getting sizing wrong, how the technology has evolved and why better size planning might be retail’s most underrated sustainability lever.
Why Sizing Gets Overlooked
Theuerkauf pointed to several reasons sizing tends to fall to the bottom of retailers’ priority lists. The volume and granularity of size-level data is one barrier. Getting clean, accurate data down to the size-store level has historically been difficult, which pushes sizing into the “too hard” pile.
There’s also what he called the “good enough fallacy.”
“You as a retailer just rely on historical size curves, and that feels like it’s good enough,” Theuerkauf said. “The reality is they’re often inaccurate and they lead to margin erosion, but you feel like it’s good enough.”
Organizational fragmentation compounds the issue. “Sometimes it falls within merchandising, sometimes in planning, sometimes in allocation, sometimes its own team,” he said. “There’s this lack of clear ownership and accountability.”
Has Sizing Gotten Harder?
On balance, Theuerkauf said it probably has, though not for the reasons many retailers might assume.
SKU proliferation is a major factor. Sizing now spans more dimensions, including length variations, extended size runs and multi-dimensional pan sizes. Trend cycles have also compressed, particularly with the rise of short-form social media platforms such as TikTok and Instagram Reels.
“What used to be fine to think of as size curves that are fairly constant — that’s not the case necessarily anymore,” he said.
At the same time, he said, the problem is becoming more tractable. Cloud-native data infrastructure has made granular, multi-source data far more accessible than it was 15 years ago, and advances in modeling, including deep learning and neural networks, have given retailers tools to make sense of that complexity in ways that basic statistical models couldn’t.
The GLP-1 Question
A significant portion of recent retail coverage has focused on GLP-1 weight loss drugs as a driver of shifting size demand. Theuerkauf pushed back on the scope of that narrative.
“In the data and the retailers that we work with, it’s hard to be able to, in an intellectually honest way, say that we’re seeing permanent population-wide shifts,” he said.
He cited research indicating that only approximately a third of GLP-1 participants remain on therapy after a year, and that usage is concentrated in specific demographic groups. In the U.S., he noted, approximately 11% to 12% of adults use GLP-1 medications, skewing toward older and higher-income consumers.
Some of the retailers Theuerkauf works with are, in fact, seeing stronger performance in larger sizes, which runs counter to the prevailing narrative. He also noted that factors such as free return policies for online apparel have an independent effect on return rates that has nothing to do with body size changes.
“Seeing or taking on the narrative that GLP-1s are the primary driver, I think that just overlooks consumer buying behavior that we know,” he said.
His recommendation: focus on the fundamentals. “Understanding root causes, knowing your customers, prioritizing brand-level consistency across components and then modernizing the technology used for sizing to drive clear financial returns,” Theuerkauf said, rather than reacting to headline-driven trends.
What Getting Sizing Wrong Actually Costs
Theuerkauf put the broader financial stakes in context. Globally, stockouts and overstocks carry an estimated cost of $1.7 trillion in retail.
Sizing errors drive that number from multiple directions. A stockout in a core size means a lost sale, often to a competitor. Excess inventory in fringe sizes ties up working capital, accumulates financing costs and eventually gets discounted or sent to outlets. There’s also a customer loyalty dimension, such as when a customer really likes a product but can’t find it in their size.
“That’s really frustrating and creates a long-term value problem,” he said.
The upside of getting it right is correspondingly significant. Theuerkauf described a set of compounding benefits: higher full-price sell-through, lower markdown liability, reduced return costs, freed-up working capital and stronger brand loyalty. Accurate size distributions also give merchants more confidence to commit to new style introductions and higher-margin capsule collections upfront.
There’s a price integrity dimension, too. “Retailers avoid conditioning their shoppers on waiting for clearance sales on off sizes,” he said. “You can protect the full-price brand perception of your brand by getting sizing right.”
How the Technology Has Changed
When Theuerkauf co-founded Syrup Tech in 2020, most of the industry was still relying on autoregressive models or historical sales curve fitting. “It hadn’t changed much in the preceding 30 or 40 years,” he said.
Today, he said, the underlying technology looks fundamentally different. Cloud-native data architectures have made it possible to pull and process orders of magnitude more data. On the modeling side, attribute embeddings can now incorporate product imagery, fit characteristics and style data alongside sales history. Sophisticated forecasting engines can simultaneously process signals including localized weather patterns, regional demographic shifts, marketing campaign velocities and channel-specific return rates.
“The output is just a much, much smarter size curve,” he said.
Keeping Planners in the Loop
Better models don’t mean less human involvement. Theuerkauf said planners should be deeply involved during implementation, not just consulted after the fact.
“Advanced algorithms, they can provide a really strong, mathematically sound, optimized baseline at whichever level of granularity you want to,” he said. “But at the end of the day, the planner continues to be the pilot of this.”
He described a shift in how that role should function: away from manual, line-by-line spreadsheet adjustments and toward managing by exception. Modern AI-native systems should surface anomalies, he said, rather than requiring planners to search for them.
“You’re not looking for that needle in the haystack, but that system is presenting that needle to you on a silver platter and you then inspect that needle,” he said.
The Sustainability Angle
Theuerkauf also connected better size planning to environmental outcomes. In apparel, he said, the industry produces an estimated 2.5 billion to 5 billion excess items annually. For roughly every 10 items produced, only three are sold at full price; three to four are sold at a discount; and approximately three end up being destroyed.
Sizing compounds this problem. More accurate size forecasting reduces overproduction, which cuts waste not just in finished goods but also in raw materials, water and energy. Fewer unsold items reach the end of a season, which means fewer end up in landfills. A leaner, more responsive supply chain also lowers the overall carbon footprint across manufacturing, transportation and disposal.
“There’s the initial effect of, well, you just produce less of things you don’t need,” Theuerkauf said. “But there’s all kinds of these second- or third-tier effects of making this operation more sustainable.”
Where to Start
For retail leaders looking to improve their sizing approach, Theuerkauf offered a tiered recommendation. Those who don’t yet have the data infrastructure in place to support AI-driven sizing should start there. For those who do, he suggested picking a single category or brand where size issues are already visible and running a focused test.
“You don’t need a broad overhaul of your entire planning suite,” he said. “Sizing is a really nice one because you can just plug in and really start seeing value immediately in something that, as we discussed at the beginning, is heavily overlooked.”
