What an Assortment Matrix Actually Is
An assortment matrix is a structured view of everything a category sells — organized by subcategory, brand, price tier and role — that shows not just what's currently ranged, but what should be, at what depth. It's the document a category manager uses to decide what to add, keep, or remove, and to defend that decision with data rather than buyer memory or vendor relationships.
Most assortment automation stalls for a simple reason: a system can't determine the right number of SKUs for a category that was never given a role, a KPI, or a limit in the first place. The matrix is what supplies those inputs.
Category Roles: The Rule Every SKU Decision Should Trace Back To
Category management treats each category as a small business in its own right, and every category typically plays one of a small number of commercial roles that determines how it's ranged, priced, and promoted.
| Role | Commercial function | Typical assortment approach |
|---|---|---|
| Traffic driver | Brings shoppers in; often price-sensitive, high-visibility items | Wide range, competitive pricing, high in-stock priority |
| Routinely purchased | Habitual, repeat-purchase staples | Deep range on core SKUs; consistency matters more than novelty |
| Margin builder | Higher-margin items that fund category profitability | Curated range; merchandising and placement matter |
| Convenience / seasonal | Occasional, impulse or time-bound purchases | Narrow, tightly rotated range tied to a season or occasion |
Assigning a role first, before touching any SKU-level data, is what makes the rest of this process defensible — a "keep, cut, add" decision justified by a category's role and KPI holds up in a line review in a way that "it's always been ranged" doesn't.
The 80/20 Rule and the Case for Rationalization
Retail assortments consistently show the same skew already covered in ABC analysis: typically 20-30% of SKUs generate around 80% of category revenue, while a long tail of slow movers consumes shelf space, working capital, and operational complexity for comparatively little return.
The goal of rationalization isn't the smallest possible range — it's the most productive one. Done with data rather than opinion, cutting the unproductive tail typically frees space and cash for the winners and simplifies day-to-day operations, without necessarily reducing total sales.
Tail-Cut Analysis: Deciding What to Remove
Not every low-ranked SKU is a removal candidate — some are new products still building a sales history, others are required for a complete range even at low volume. A defensible tail-cut process checks each candidate against a short list of questions:
- Is this SKU new? Give recently launched products a minimum evaluation window before judging them by the same standard as established lines.
- Does anything get lost if it's removed? Substitutability testing — whether a near-identical alternative already exists in the range — determines whether a cut is safe or leaves a real gap.
- What role does it serve beyond direct revenue? A low-volume item that completes a set, satisfies a small but vocal customer segment, or anchors a category's credibility may earn its place despite thin numbers.
- What's the removal actually worth? Freed shelf space and working capital only matter if they're redirected to something more productive — cutting without a plan for what replaces the SKU captures none of the benefit.
Demand Transference: What You Actually Lose When You Delist
Removing a SKU rarely removes 100% of its demand from the category — some of it transfers to a similar product you still carry, and some of it is lost entirely to a competitor or simply doesn't get spent. Demand transference modeling estimates that split, so every removal decision carries an actual value attached instead of an assumption that the sales simply vanish or simply move.
| Outcome when a SKU is delisted | What it means for the category |
|---|---|
| High transference to a similar in-range SKU | Safe cut — most of the revenue stays in the category |
| Low transference, high leakage to competitors | Risky cut — real revenue loss even if the individual SKU looked unproductive |
| Demand simply disappears (impulse or occasion-specific item) | Cut is close to neutral for the category as a whole |
Putting It Together: A Practical Assortment Review Process
- Confirm or assign the category role and the KPI that role implies (traffic, margin, basket completion) before reviewing any SKU.
- Run ABC-XYZ classification across the category to separate value and predictability, as covered in the first article of this series.
- Flag tail-cut candidates using the questions above rather than a blanket revenue threshold.
- Model demand transference for each candidate before finalizing a cut list.
- Decide what fills the freed space — new SKUs, deeper stock on existing winners, or a deliberate narrowing — so the review produces a plan, not just a shorter list.
- Re-run on a fixed cadence — seasonal categories need this more often than staples, but every category drifts from its matrix over time as new products launch unreviewed.
This closes the loop across the series: ABC-XYZ tells you where a product sits in value and predictability, price monitoring tells you what it should cost, and the assortment matrix tells you whether it belongs in the range at all — three views of the same category, built from the same underlying data.