Do You Know Why Your Size Curves Are Always Off?
and how to fix them
You think you’ve cracked your size curve.
You’ve looked at sales data. Compared sales to buys. Factored in returns. Maybe even spotted that certain sizes sold out faster. But you can’t ever really measure the missed opportunity; the sales you lost because you didn’t have enough, or the way that shortage skewed sell-through in other sizes.
And still, the same thing happens: you’re overstocked on the fringe sizes and sold out of the core sizes by week two.
If you work in fashion merchandising, you know this pattern well. The industry talks a big game about data-driven planning, but too often, the foundations are built on flawed assumptions. Size curves are one of those foundational areas that look simple on paper and fall apart in practice.
The question is: why? Why are so many teams still struggling to get this right, even with better data and more tools than ever before?
Let’s break it down.
The problem with sales data
Most size curves are based on historical sales ratios. That’s logical. But the flaw is this: sales only tell you what was available to sell, not what was missed.
If you run out of size 10s after ten days, and still have stock in other sizes at week five, your size curve might get rebalanced away from 10s next season. When in reality, that was the size with the highest demand, you just didn’t buy enough.
You might try to adjust, take a few units out of 12s and 14s, add them to 6s and 8s, but it’s still only a guess. You’re never sure how much to shift, and often end up repeating the same problem in a slightly different shape.
That lost demand never makes it into the spreadsheet.
Without tooling that flags early sell-outs by size or estimates missed demand, you’re making rebuys on flawed foundations. You're reinforcing the same mistake.
And this isn’t just a new-season problem. It happens in continuity products too. That core black tee might look like it sells evenly across sizes over the course of a season, but dig deeper and you’ll find uneven peaks, shortfalls, and moments of missed opportunity.
Returns don’t help either
Returns add another layer of noise. Especially when you're dealing with 40%+ return rates.
When returns are that high, and you’re trying to place a replenishment order, you’re not working with the full picture. Some of that stock is coming back. Some of it is coming back in the wrong sizes. Some of it is going to sit in a warehouse for three weeks before it hits your available stock again.
If you re-order too soon, you're at risk of duplicating what you'll get back. If you wait too long, the opportunity window might have passed.
And if you don’t track the reason for returns by size, you're not learning much at all. Was it fit? Was it a colour discrepancy? Did it look like a midi and fit like a maxi?
Most return reports are blunt. Smart tools (and smart teams) break it down by SKU, size, and reason. Because knowing that your 8s get returned more than your 10s matters. Especially if you're trying to correct for future demand.
Size curves don’t stand still
Even for core products, your size curve shouldn’t be static.
Let’s say you always run a certain denim style. Season one, you run out of smaller sizes first. So you shift the curve to favour those. But the next time around, maybe you over-indexed and now the mediums and larges are running low.
Replenishment becomes a balancing act. You're constantly trying to align available stock with fluctuating demand across different leg lengths, fits, and waist sizes. One tweak leads to another, and you’re always chasing equilibrium.
Tools that model these shifts in real-time, based on returns, sell-out velocity, and even regional buying behaviour, are becoming more common. But they still require human judgment. The data can tell you what happened. Only the merch team can decide what to do about it.
Spec changes ruin everything
This is the kind of thing that breaks your perfect plan. A new season rolls around and you place a repeat order on a bestseller. But the fit spec changes. A size 8 this season is not the same size 8 you sold last season. Maybe you adjusted the product spec because returns were high. Maybe a new factory interpreted the pattern slightly differently.
It might be invisible to the customer, or it might cause a wave of confusion and returns. Either way, your clean year-on-year data is now muddy. And that screws with size curve planning, SKU tracking, and customer trust.
If the product changes, treat it like a new one. New SKUs, new size data, new assumptions. Otherwise, your planning data is a mess.
Availability is marketing
Here’s the bit that gets overlooked.
Getting your size curve wrong doesn’t just affect operations. It kills demand.
A shopper clicks through to a product, sees their size is sold out, and bounces. If it happens a couple of times, they stop checking back. The sale is lost before it even begins.
Product availability is a marketing lever. If you’re paying to drive traffic to a product page and the top sizes are out of stock, you’re burning cash.
Worse still, if customers start to associate your brand with inconsistent sizing, unpredictable fits, or low availability, you lose them for good. Not because of poor product, but because of poor planning.
The fix isn’t flashy
Fixing size curves isn’t about some genius algorithm. It’s about learning fast and feeding that learning back in. That means:
Tracking sell-out velocity by size, not just style
Recording return reasons in detail
Tagging spec changes and accounting for them in rebuys
Using tooling to flag demand patterns you might otherwise miss
Most importantly, it means taking a step back each season and asking: did we get this right? Or are we just repeating the same mistakes with new data?
The brands who get this right don’t just sell more. They waste less. They build trust. And they get a lot closer to being in stock in the sizes that actually matter.