Retail Assortment Management Software: What It Does After the Plan Is Set

Assortment planning and assortment management are not the same job

These two phrases get used interchangeably by vendors and they should not be. They describe different work, happening at different times, answered by different data.

Assortment planning is a pre-season decision. Which SKUs are we carrying, in which stores, at what depth, for the coming season. It is a forecasting and financial exercise, run once or a few times a year, and it produces a plan.

Assortment management is what happens for the rest of the year. The plan meets reality: a SKU never lands, a store never sets the new planogram, a slow mover eats four facings for eleven weeks, a regional winner never gets rolled out. Management is the ongoing job of noticing those things and correcting them while the season is still running.

Most chains have bought software for the first job and do the second one in spreadsheets. That is the gap this article is about.

Why the plan drifts, always

A well-built assortment plan starts degrading from the day it ships. The reasons are boring and they are the same everywhere:

• Execution gaps. The store did not set the planogram, or set it wrong, or set it three weeks late. The plan says the SKU is there; the shelf disagrees.
• Supply gaps. A SKU is late, short-shipped, or allocated to a different region. The space stays empty or gets filled with whatever is nearby.
• Local reality. The cluster model said these forty stores behave alike. Six of them do not, and they have been under-assorted since week one.
• Cannibalization. The new item sells, but it takes its volume from the item next to it rather than adding any. The plan counted it as incremental.
• Quiet failure. A SKU stops selling in a store and nobody notices, because zero sales looks identical to no demand until you check whether it is actually on the shelf.

None of these show up in a plan-versus-actual report at the category level. They net out. Category sales look close to plan while individual stores and SKUs are well off it in both directions.

What assortment management software should actually do

Strip the feature lists away and there are five questions worth paying for. If a tool cannot answer these, it is a planning tool with a dashboard.

1. Is the assortment actually present? Not the plan, the shelf. Which SKUs that should be carried in this store are not selling, not stocked, or not set. This is the single most common failure and the easiest to miss, because a SKU that was never set generates no sales, no returns, and no complaints.

2. Which SKUs are earning their space? Sales per facing, per store, per week. A SKU that ranks acceptably chain-wide is often carried in a third of stores where it has never justified a facing.

3. Where is the tail costing more than it returns? Every assortment accumulates SKUs nobody has revisited. The question is not whether to cut them chain-wide, it is which stores should stop carrying them.

4. What is not carried here that works next door? The inverse, and the one that gets ignored. A SKU performing well in demographically similar stores and absent from this one is straightforward upside that no plan-versus-actual report surfaces.

5. Did the change work? When you add, cut, or re-space a SKU, did total category sales in those stores move, or did the volume just relocate? Most tools report the SKU. Fewer report the category effect, which is the number that matters.

The measurement problem underneath all of it

Assortment decisions are causal questions dressed as reporting questions. "Did adding this SKU grow the category" is not answered by looking at the SKU's sales. It is answered by comparing what happened in stores that got the change against comparable stores that did not.

Without that comparison you will systematically overrate additions, because a new SKU always sells something, and underrate cuts, because the volume that moves to a neighbouring SKU is invisible if you only look at the one you removed.

This is why assortment work so often produces confident decisions that do not compound. The reporting confirms whatever was done.

Assortment management runs on shelf truth, not plan data

Every question above depends on knowing what is really on the shelf, store by store, more often than a field rep visits. That is the constraint. A plan-versus-actual view built only on POS and inventory records inherits every error in those records.

The practical version of shelf truth for most chains is not cameras in every store. It is inference from data you already have: a SKU that normally moves twelve units a day selling zero while the system shows stock on hand is telling you something specific about that shelf. We wrote about how that works in automated shelf audits without field reps, and about the related detection problem in real-time out-of-stock detection.

Where shrink fits

Assortment and loss are usually managed by different teams, which is a mistake at the SKU level. The National Retail Federation's National Retail Security Survey put the average shrink rate at 1.6% of sales in fiscal 2022, up from 1.4% the year before, which the NRF calculated as $112.1 billion across the industry.

That is an industry average, not your number, and the reason it belongs in an assortment conversation is that shrink is not evenly distributed across a range. A handful of SKUs in a handful of stores carry a disproportionate share of it, and those SKUs are frequently still being ranged on gross margin that never accounted for the loss. An assortment view that ignores shrink will keep recommending items that lose money at the store level. Our guide to grocery shrink software goes further into that side.

Questions worth asking a vendor

• Does this tell me what is on the shelf, or only what the plan and the inventory system believe?
• At what frequency? A weekly refresh answers a daily question with stale data.
• When it recommends a cut or an add, can it show the category-level effect, or only the SKU?
• Does it work store by store, or does it stop at cluster level? Most of the money is inside the clusters.
• What does it do on the first day, before any historical data has accumulated in it?
• Does it reconcile against shrink, or is margin taken at face value?

A reasonable place to start

Do not start with a full re-range. Start with the smallest question that has a clear answer: for one category, which SKU and store combinations are ranged and showing no movement at all over the last four weeks. That list is nearly always longer than the category manager expects, it costs nothing to produce from data you already hold, and it separates distribution failures from genuine demand failures before anyone argues about the plan.

Fix those first. Re-ranging around a shelf that was never set correctly just moves the error.