How to reduce overstock in retail

Overstock is a lagging indicator of a decision you already made

By the time inventory is recognizably overstock, every decision that created it is weeks old. The center-store departments at our 62-store operator were carrying 6.3 weeks of supply against a 3-week ceiling, and the order behind every one of those weeks was placed, confirmed, received, and shelved. The money is spent, the space is committed, and the only remaining choices are about recovery rate. This is why any attempt to reduce overstock that starts at the markdown report is expensive by construction: it starts at the last possible moment, when every cheap option has already expired.

Nobody decided on 6.3 weeks. It accumulated, one reasonable-looking order at a time, in a department slow enough that no single line looked alarming and no report aggregated them until the seasonal markdown review.

Reducing overstock is therefore two jobs that get confused with each other: clearing what exists, and closing the tap that fills it. The first is a merchandising exercise with a known playbook. The second is where the durable saving is, and it is mostly about order review rather than about inventory at all.

Find it before the markdown report does

Overstock hides from averages. A department at a healthy 2.4 average weeks of supply can contain a hundred lines at eight weeks, because the fast movers pull the mean down. Every detection method that works looks at the tail.

Three signals, in order of how early they fire:

The first is the only one that fires before you own the inventory, which is why the over-order flag is worth more than any overstock report. The second is the best available detector for stock you already have, and it costs nothing to compute: read the 90th percentile of weeks-of-supply per department rather than the mean, and the tail stops hiding.

The third is the one chains most often lack. An item with 5 weeks of supply is fine in March and a disaster in the second week of a six-week seasonal window, and a flat weeks-of-supply threshold cannot tell the difference. Compare cover against the remaining selling window, not against a constant.

Clear it in the right order

Once it exists, the sequence matters, because each step burns margin and the cheap steps are finite. Work down the ladder and only descend when the step above is exhausted.

Step 1 sounds too obvious to list and is the most commonly missed, because the order guide keeps recommending against a target that was set before the excess existed. If the replenishment model still targets 3 weeks and the store holds 6, the model will happily order again the moment on-hand dips under target. Suppress the item, do not just intend to.

Step 2 is where the biggest free recovery usually sits, and it is the one that requires chain-level visibility to even attempt. Overstock is rarely uniform: the same item is over at 14 stores and under at 9. Rebalancing turns a markdown into a transfer, at transfer cost rather than margin cost. But you cannot rebalance what you cannot see across stores, which is the whole argument for store-level inventory visibility.

Where overstock actually concentrates

Chains tend to look for overstock in the wrong places, because attention follows dollars and overstock follows unpredictability. Sorting the center-store excess at our 62-store operator by cause produced a distribution that surprised the buying team.

Only 8% was what the team had assumed the whole problem was: getting the forecast wrong on important items. Sixty-one percent sat in the first two rows, which are not forecasting failures at all. They are policy artifacts, a minimum order and a uniform target, both producing predictable excess on items nobody was watching because individually they are small.

That is the argument for treating overstock as a policy problem rather than an execution problem. A better forecast on core items addresses 8% of the units. A minimum-order renegotiation and a segmented target address 61%.

The promotional row is worth a separate note. Promotional over-buy becomes overstock when the lift assumption was wrong, and it is systematically under-detected because the inventory arrives justified: there was a promotion, it was planned, the buy was approved. Comparing post-promotion on-hand against the pre-promotion baseline within two weeks of the event catches it while transfer is still an option.

The markdown timing question

When a markdown becomes necessary, the recurring error is going too shallow too late. A 15% markdown that fails, followed three weeks later by 30%, then 50%, recovers less than a single well-timed 30% would have, because the item spent six extra weeks occupying space and depreciating.

The discipline: decide the terminal date first, then work backwards to the depth that clears the units by that date at the observed elasticity. If 40 units must clear in four weeks and the item sells 3 a week at full price, no realistic markdown gets there and the honest answer is liquidation or a transfer, not a sequence of hopeful discounts.

Two supporting rules worth writing down. Mark down the whole chain position at once rather than store by store, because staggered markdowns move units between stores rather than out of the system. And measure recovery against the original cost, not against the last price, or every subsequent markdown looks successful relative to a number you already conceded.

Reduce overstock for good: close the tap

Everything above is recovery. The saving that persists comes from four changes upstream, none of which are inventory decisions.

Fix the order review so over-ordered lines are visible. At store 1046's dairy department, 8 of 9 over-ordered lines never appeared in the order screen because the net-ask filter dropped them. Every one of those is future overstock created in full view of a competent manager.

Set targets by demand shape rather than one chain-wide number. A flat 3-week target guarantees excess on slow items, as the replenishment guidance sets out. The slow, erratic tail is where overstock concentrates, and it is exactly where a uniform target is furthest from right.

Negotiate the minimums. Minimum-bound items cannot be ordered correctly. If a supplier's 40-case minimum forces six weeks of supply into a small-format store every cycle, that is a purchasing term producing predictable overstock, and it should be priced into the landed cost when the supplier is chosen.

Cut the tail. Some overstock is not an ordering failure, it is an assortment failure: items that should not be in the set at all. The bottom three deciles of the center-store set contribute 3.6% of dollars while occupying 30% of the SKU count, and those items generate overstock structurally because their demand is too low and too erratic to order well at any minimum. That is an assortment decision, not a replenishment one.

The space argument nobody makes

Overstock is discussed as a working-capital problem, which undersells it. The carrying cost of the money is real but modest. The expensive part is that excess inventory occupies backroom space and shelf facings that a productive item could be using.

Center store at our operator returns $288 per selling square foot annually, the lowest of any department, while holding the most weeks of supply. Those two facts are connected. Space allocated to slow, over-covered items is space not allocated to the categories returning $704 and $812 per foot. That excess is occupying the most contested resource in the building, which is a larger cost than the tied-up cash and almost never counted as one.

That reframing changes which excess you prioritize. Ranked by dollars, the overstock list is dominated by expensive items carrying a few weeks of extra cover. Ranked by space-weeks consumed, it is dominated by cheap bulky slow movers that would never appear on a working-capital report. The second ranking is the one that frees capacity, and it points at assortment and space decisions rather than at ordering.

Doing this in Scout

The detection signals above are all computable from data chains already have, and they are almost never computed, because each one requires joining on-hand, on-order, rate of sale, and a seasonal window across every store.

Scout runs those as standing views. The 90th-percentile weeks-of-supply read is per department per store rather than a chain average, so the tail is the default view rather than something you go looking for. Cover is compared against the remaining selling window where a seasonal curve exists, which is what separates "5 weeks of supply in March" from the same number in week two of a six-week window.

The rebalancing candidates fall out of the same view: items over target at one store group and under at another, ranked by the units a transfer would move. That list is the difference between a markdown and a transfer, and it is not assemblable from a single store's order screen.

Scout finds and ranks the overstock and tells you where the tap is open. It does not execute transfers, set retail prices, or run your markdown process: those stay in your merchandising and pricing systems.

Setting a target you can actually hit

Chains that successfully reduce overstock tend to set the goal in weeks of supply at the tail rather than in total inventory dollars. Dollars are the number finance wants and the wrong number to manage against, because the fastest way to cut inventory dollars is to stop ordering the expensive fast movers, which is exactly backwards.

A target like "90th-percentile weeks of supply under 5 in center store by Q3" is harder to game, points at the specific lines causing the problem, and does not reward starving the categories that turn. Report the dollar figure to finance as the outcome; manage the tail.

Summary

• Overstock is the residue of ordering decisions weeks old, so the highest-value detector is the over-order flag at the order, not the markdown report.
• Read the 90th percentile of weeks-of-supply, not the mean, and compare cover against the remaining seasonal window rather than a flat threshold.
• Work the clearing ladder in order and do not skip "stop ordering." The biggest free recovery is usually rebalancing between stores, which requires chain-wide visibility to even see.

Further reading: preventing over-ordering closes the tap, and store-level inventory visibility is what makes rebalancing possible.