Monday morning. You have a category review in two hours, your inbox has three requests from buyers, and somewhere in your POS export there are SKUs that need a decision today. Not next quarter. Today.
Every SKU in an active category will eventually fall into one of three buckets: cut it, keep it at current depth, or reorder it — and if reorder, at what quantity and to which stores. These are the only three levers a category manager actually controls on a week-to-week basis. Everything else — vendor negotiations, planogram resets, new item introductions — happens on a longer cycle. The cut-keep-reorder decision is the operational pulse.
The problem is that most category teams are making these calls with data that is structurally misleading: chain-level averages, last season's rank files, and gut feel layered on top of both. This article walks through what each of the three decisions actually requires and where the typical data workflow breaks down.
The Cut Decision: When Sell-Through Alone Is Not Enough
The instinct when a SKU underperforms is to look at its sell-through rate and compare it to a threshold — say, 60% at eight weeks. Below that, it's a cut candidate. That logic is not wrong, but it's incomplete in two important ways.
First, sell-through is not shelf-space-neutral. A SKU assigned six facings and selling through at 45% is performing differently than a SKU with two facings at the same 45%. The second item is constrained; the first may genuinely be slow. If you cut both, you've made one correct decision and one expensive mistake. The correct input for a cut decision isn't sell-through rate alone — it's sell-through rate adjusted for shelf allocation. How fast is this SKU selling relative to the space it's occupying?
Second, cut decisions need store-level disaggregation before they become final. A SKU averaging 42% sell-through across 28 stores might be running at 71% in six stores and 18% in the rest. The right call isn't a chain-wide cut — it's a selective reduction that keeps the item in its performing stores and pulls it from the ones where it's occupying space at cost. Category managers who make cut decisions off chain averages are systematically over-cutting in strong-performing locations.
A Practical Cut Threshold
A more defensible threshold combines three signals: sell-through rate at week eight (adjusted for facings), gross margin contribution versus category average, and weeks of supply on hand. A SKU that is below the category sell-through median, below average margin, and carrying more than 10 weeks of supply is a genuine cut candidate. Any one of those conditions alone is not sufficient. Needing all three is a tighter screen, but it produces fewer regret cuts.
The Keep Decision: What You're Actually Deciding
Keeping a SKU sounds passive — you're choosing not to act. In practice, a keep decision means you're committing shelf space and inventory depth to that item for another planning period. That's an active allocation of two scarce resources.
The relevant question for a keep call is whether this SKU is earning its space relative to its alternatives. That requires knowing what a comparable new item or increased depth in an adjacent SKU would produce in the same slot. Most category review workflows don't surface this comparison because the data lives in different exports: current SKU performance in POS, alternative item potential in vendor sell-in sheets, and shelf allocation in a separate planogram file.
We're not saying the keep decision should be over-analyzed — many SKUs are correctly kept precisely because they're stable and predictable, and predictability has value in supply chain planning. A SKU that reliably sells 80 units a week with low variance is a valuable baseline even if it doesn't top the velocity chart. The point is that "keep" deserves a reason, not just an absence of a cut signal.
The Reorder Decision: Quantity and Allocation Are Different Problems
Consider a regional home goods chain running a 22-store footprint. Their bedding category has a queen-sized comforter (SKU #4471) that averages 61% sell-through chain-wide, comfortably above the 55% keep threshold. In week seven of the season, the buyer is deciding whether to reorder and at what depth.
At chain average, the math looks straightforward: 61% sell-through, decent margin, reorder to replenish to four weeks of supply. But store-level data tells a different story. Six stores — primarily suburban locations in the northern part of the region — are running at 88% sell-through and have already stocked out twice this season. Twelve stores are at 55-65%. Four stores in the warmer southern market are at 31%.
A single chain-level reorder quantity allocated proportionally will under-serve the six high-velocity stores and over-stock the four slow ones. The right reorder decision has two components: the quantity (driven by the stores that will actually absorb it) and the store-level allocation (weighted toward velocity, not toward proportional distribution).
This is where most category review processes fail. The reorder quantity is set in one meeting, the allocation is handed to the DC team as an afterthought, and the result is stock-outs in the stores that need it most and excess inventory in the stores that don't.
Reorder Quantity vs. Reorder Allocation
These are two separate decisions that get conflated into one line on a purchase order. The quantity decision is a demand forecast question: how much total inventory is needed to cover the remaining season in the stores where this item is performing? The allocation decision is a shelf capacity and velocity question: which stores get what share of that quantity, and does the planogram support the depth you're planning to ship?
If your planogram for SKU #4471 allocates two facings at a 12-unit shelf capacity, and you're planning to ship 24 units to a single store in a four-week window, you're setting up a back-stock situation where the inventory exists but isn't on the floor driving sell-through. That's not a demand problem; it's a shelf-space problem disguised as a reorder problem.
The Monday Cadence: When to Use Which Signal
Not all three decisions require the same data freshness. Cut decisions based on seasonal sell-through can tolerate a week-old POS pull. Keep decisions on stable SKUs can be reviewed bi-weekly. Reorder decisions during peak season — particularly for weather-sensitive or trend-driven categories — need the most current data available, ideally daily POS refresh on the SKUs in question.
The mistake that inflates Monday morning prep time isn't running too much analysis — it's running undifferentiated analysis. Category managers who pull the same full-category report every week regardless of what decisions are actually pending spend 40% of their prep time on data that won't change any decision. The efficient approach is to have a pre-defined trigger list: which SKUs are within two weeks of a reorder decision window, which are below the cut threshold for the third consecutive week, which have had store-level sell-through divergence of more than 25 percentage points since the last review.
Those are the items that need Monday attention. Everything else is monitoring, not deciding.
Where the Plan Gets in the Way
There's a real cost to category review processes anchored in last season's plan. The plan encoded good information when it was written — vendor terms, expected demand, agreed assortment depth. But the moment the season opens, current sell-through data starts diverging from plan assumptions, and it almost never converges back.
Teams that defer cut-keep-reorder decisions to the quarterly category review because "that's when we update the plan" are making a timing error. The cost of carrying dead inventory for an additional eight weeks waiting for the review cycle is real: markdown exposure increases, shelf opportunity cost accumulates, and the inventory that should have been reordered into the strong stores is sitting in the slow ones.
The plan is a useful starting point. Current sell-through data is what you act on. The gap between them is where the three decisions live, and it needs to be reviewed on a weekly cycle, not a quarterly one.