The markdown is the receipt, not the decision
End of January markdown rails are not a clearance strategy. They are evidence of a buying mistake made four months earlier.
By the time seasonal stock is sitting in a discount area, the important decision has already passed. The order was placed too early, with too little evidence, or with too much confidence in a single forecast. The team may still improve the outcome through pricing, promotion, and channel shifts, but it is now managing damage rather than managing demand.
I have seen this cycle repeatedly in multichannel fashion and home goods retail. A buyer starts with a reasonable category plan, adds a little extra for safety, and then adds more because a supplier deadline is approaching. The result looks prudent in the moment. At season end, that same prudence becomes aged stock, reduced margin, and working capital tied up in products customers no longer want.
The fix is upstream. It begins with making the initial buy more defensible, while accepting that no forecast can remove uncertainty entirely.
Overstock starts with an imprecise first order
The root cause is usually described as inaccurate forecasting. That is true, but incomplete. Forecast error only becomes expensive when it is converted into an oversized order.
Many retailers still set the first order using a blend of last year's total sales, supplier minimums, a buyer's intuition, and a broad assumption that the new season will behave similarly. Each input may be reasonable on its own. The problem is that the process often does not distinguish between products, locations, channels, or selling periods.
A scarf that sold well in one flagship store is not automatically a good buy for three smaller locations. A product that sold 600 units across a full season may have sold 400 during a short cold spell and then slowed sharply. A total season number hides the velocity pattern that should shape the next order.
There is also a common psychological trap. Retailers fear stockouts because a missed sale is visible and immediate. Overstock arrives more quietly. It is distributed across stores, sits in the back room, and only becomes urgent when the season is nearly over. That delay makes overbuying feel safer than it really is.
The right question is not, “How much did we sell last season?” It is, “How did this product sell, where did it sell, and how quickly did demand change?”
Build the buy from velocity, not a single season total
Previous season data cannot tell you exactly what will happen next. It can show the shape of demand, which is much more useful than a single historical total.
Start at product and location level. Review units sold by week, sell through at key dates, stockout periods, returns, transfers, discounts, and the share sold through each channel. Separate full price sales from promotional sales. If a product sold quickly only after a 30 percent markdown, its historical volume should not be treated as evidence of full price demand.
Then compare products with similar characteristics. The best reference for a mid-priced wool scarf may be last year's comparable wool scarves, not the entire accessories category. Look for patterns such as:
- Initial weekly velocity after launch.
- Time required to reach 25 percent, 50 percent, and 75 percent sell through.
- Differences between stores, ecommerce, and other sales channels.
- Demand lost during stockout weeks.
- Return rates and the effect of promotions on net sales.
- The point in the season when demand began to decline.
This analysis changes the buying conversation. Instead of saying that a category sold 1,000 units last winter, the team can say that comparable products sold 8 units per location per week during the first six weeks, then fell to 3 units, with stronger demand in two locations and a higher online return rate.
That detail supports a staged buy. Place an initial order against the demand that is most defensible, then reserve part of the planned budget for a reorder if early performance confirms the forecast. The exact split depends on supplier lead times and product availability, but the principle is consistent: do not commit all seasonal inventory before the season provides new evidence.
| Buying input | What it tells you | How to use it |
|---|---|---|
| Comparable product velocity | Likely early demand by week and location | Set the initial order and launch allocation |
| Sell through by date | How quickly stock moved at full and reduced price | Set review points and markdown triggers |
| Stockout periods | Where recorded sales understated demand | Adjust the estimate without copying the total blindly |
| Channel performance | Where demand can be fulfilled most efficiently | Plan transfers and reserve inventory |
A forecasting system can help combine these signals across a large assortment, but the output still needs commercial judgment. The purpose is not to produce a magical number. It is to make assumptions visible, compare them with actual performance, and give the buyer a clear reason for the quantity ordered.
A synthetic example: scarves across four locations
Consider a four-location fashion accessories retailer preparing its autumn assortment. In the previous year, scarves sold 1,280 units across the business. The strongest two locations accounted for most of the volume, while the other two had slower but steady demand. The buyer used the total category result and ordered 1,650 units for the next autumn, expecting colder weather and stronger gifting demand.
By the middle of the season, 35 percent of the inventory remained in the slower locations and online. The strongest stores had sold through much of their stock, but the remaining colours and patterns had weaker appeal. The business eventually marked down 35 percent of the original inventory. Sales continued, but the margin on those units was materially lower than planned.
The mistake was not simply ordering 370 units more than the previous year. The bigger mistake was treating the previous total as a uniform demand signal. The retailer could have separated the best performing styles from the rest, modelled location differences, and used an initial order closer to proven early velocity. It could also have held back part of the budget for a reorder into the strongest locations once the season confirmed demand.
That approach would not have guaranteed a perfect result. Weather, colour preference, and gifting behaviour would still have introduced uncertainty. It would, however, have reduced the number of units exposed to the weakest locations and preserved more options after launch.
When the stock is already there, act in sequence
Better buying does not help the units already in the warehouse. Once overstock exists, the tactical objective is to recover as much margin and cash as possible before the product becomes irrelevant.
First, identify the age and remaining demand for each item. Do not apply one blanket markdown across the category if some products are still selling at full price. Set review dates before the season reaches its final weeks. A small, earlier adjustment on a slow item can be better than a deep late discount when customers have already moved on.
Second, use channel sequencing. Move stock to locations or channels where comparable products still have credible demand. A product that is slow in one shop may perform better online, in a larger location, or through a business to business channel. Transfers only make sense when the receiving channel has evidence of demand and the transfer cost does not consume the expected recovery.
Third, markdown with a clear purpose. The first reduction should create a reason to buy while there is still enough season left to sell volume. If the product is highly seasonal, waiting for a larger discount can destroy more value than an earlier, controlled reduction. For the final units, bundle offers, outlet channels, and liquidation may be appropriate, but they should be sequenced after the higher recovery options have been tested.
- Protect full price sales where velocity remains healthy.
- Reallocate stock toward channels with demonstrated demand.
- Review slow items against a dated markdown trigger.
- Use targeted promotions before broad category discounts.
- Clear residual stock decisively before the next season occupies the same space.
The important discipline is to avoid emotional pricing. Keeping an unrealistic price in place does not protect margin if the item does not sell. Margin is recovered through the full decision, including timing, channel, inventory holding cost, and the opportunity to use the cash elsewhere.
Forecasting has limits, and underbuying has a cost
A data based buy does not solve genuinely novel demand. If a retailer launches a product with no historical equivalent, previous velocity may offer little guidance. The same is true for a trend item with an unpredictable life cycle. A product can become desirable in a week, or lose relevance just as quickly.
In those cases, use smaller initial commitments, supplier flexibility, rapid early readouts, and explicit test budgets. The absence of history is not a reason to pretend the forecast is precise. It is a reason to make the risk affordable and create a fast path to learning.
There is an equal danger in reacting too strongly to overstock. A retailer that cuts every initial order may reduce markdown exposure and create stockouts during the weeks that matter most. In many fashion categories, a missed sale during peak demand costs more than a controlled end of season markdown. The customer may buy from another retailer, and the lost full price sale cannot always be recovered through a later reorder.
The goal is therefore not the lowest possible inventory. It is the right balance between availability and exposure. Buyers should identify the products where being in stock is strategically important, then protect those items with a higher service target. For less certain products, use a smaller commitment and more frequent replenishment decisions where the supply chain allows it.
Next season, put the buy review before the supplier deadline: compare comparable product velocity by week, location, and channel, document the assumptions behind the initial order, reserve a reorder budget, and schedule the first forecast check before the first markdown conversation.
