Returns are demand too, and planning gross units is why you over-buy

Most DTC plans have a blind spot the size of the returns dock. The forecast predicts gross units, the buy is placed against gross units, and returns get handled somewhere downstream as a logistics problem: a reverse label, a restock, a refund. Nobody plans for them, because returns feel like operations, not demand.

That framing is the mistake. A return is not an operational footnote. It is demand information. It tells you that a customer wanted the product enough to order it and not enough to keep it, and the rate at which that happens varies enormously by category. Plan gross and ignore that, and you will over-buy the exact styles that come back the most, every single season.

The math is unforgiving and completely predictable. If a category returns at 40 percent and you plan gross, you are buying to a demand number that overstates real, kept demand by two-thirds. The units come in, a huge share bounces back, and you are left holding stock the plan never accounted for, in the styles that were always going to do this.

Return rates are not uniform, and that is the whole problem. A basics category might return at 8 percent. Fitted apparel, dresses, tailored pieces, anything where fit is a gamble, can run 30 to 40 percent. Footwear sits high too. Fashion and going-out styles that get worn once and sent back push higher still. When you plan every category on gross units, you apply the same implicit assumption, zero returns, to categories that behave nothing alike.

Follow it through. Two styles both forecast to sell 1,000 gross units. One is a basic that returns at 8 percent, so 920 stay sold. The other is a fitted style that returns at 38 percent, so only 620 stay sold. Plan both at 1,000 and you have bought the fitted style to a kept-demand number that is 60 percent too high. You will discover this on the returns dock, not in the plan, which is the most expensive place to discover anything.

This is why the high-return styles are so often the ones sitting in the clearance bay at season end. It was never a taste failure. It was a planning failure that baked in an assumption of zero returns for a category that returns at four in ten. The buy was over-scaled from the first purchase order, and no in-season heroics could fix a number that was wrong before the goods shipped.

The returns also arrive on a delay that makes the over-buy hard to catch until it is too late. A customer buys today and returns three weeks from now, so for the first few weeks of a season the high-return style looks like it is selling beautifully. Gross sales are strong, the dashboard is green, and someone suggests chasing it. Then the returns wave lands, and the units you sold and the units you chased both come back at once, all at the moment the season is winding down and there is nowhere for them to go but markdown. Gross planning does not just over-buy the initial order. It tempts you into a reorder on demand that was always going to reverse itself.

Forty percent is not an edge case in fitted apparel and footwear. It is Tuesday. And it means the difference between a plan that holds and a clearance problem is whether you planned the units you would keep or the units you would ship.

The fix has two parts, and both are forecasts. First, forecast the return rate itself, at the category level and finer where the data supports it, because return behavior is a demand signal you can predict from history just like sell-through. Fitted dresses return at a rate; wide-leg trousers return at a different rate; the same silhouette returns differently online than in a store that offers fitting rooms. This is learnable, and it is stable enough to plan against.

Second, plan the buy on net units, kept demand, not gross. Tightly's platform forecasts gross demand at SKU by size by channel and forecasts the return rate for that category and channel, then plans the buy against the net: the units that will actually stay sold. The high-return styles get bought to their real kept demand, which is lower, and the low-return styles are not penalized for behavior they do not have. The over-buy comes out of exactly the styles it was hiding in.

There is a size dimension too, and it is where returns and size curves collide. Returns are not spread evenly across the curve; the ends often come back more than the middle as customers order two sizes to compare and send one back. Modeling the return rate by size, not just by category, sharpens the curve as well as the depth, so you are not over-buying the extra-large twice: once for weak demand and again for high returns.

Forecasting the return rate also has to account for the fact that not every returned unit comes back sellable, and not every one comes back on time. Some returns arrive damaged, worn or past the point where they can go back on the shelf at full price. Others land so late in the season that they can only be cleared. So the number that matters is not just the gross return rate but the recoverable, back-in-stock-in-time rate, which is lower. A plan that assumes every returned unit reappears as clean, resellable inventory is making the opposite mistake to gross planning: it is over-counting supply instead of over-counting demand. The right model forecasts both the return rate and how much of it actually comes back in a state and a timeframe you can resell.

This is squarely a DTC problem in a way it is not for wholesale or store-only retail. When a customer buys online and returns by mail, the brand owns the entire reverse flow and sees every unit come back, which is both the burden and the opportunity. The burden is that the returns hit your inventory and your capital directly. The opportunity is that you have the data to predict them precisely: every order, every return, every reason code, tied to a style, a size and a channel. Wholesale buyers rarely see this. A DTC brand that ignores it is throwing away the one advantage the channel hands it for free.

The shift in mindset is small and the effect on the buy is large. Stop treating the gross forecast as the demand number and start treating net, kept demand as the number the buy is planned against. Forecast the return rate as a first-class input, by category, channel and size. Then the plan reflects the units customers will actually keep, and the reverse-logistics dock stops being where you learn what the plan got wrong.

A sale that comes back was never a sale. Plan the units your customers keep, not the units you ship and pray about.

Do this and the high-return categories stop being your perennial clearance problem. They get bought to their real, net demand from the first purchase order, the over-buy comes out where it always lived, and the working capital you were quietly burning on units destined for the returns dock goes back into styles that actually stay sold. Returns are demand too. Plan for them like it.