Assortment and SKU rationalisation
Many retailers carry more product lines than they need. Find out which ones earn their place, and which cost more than they earn.
- Free 30-minute call
- No commitment
- You own the code
In short: An assortment optimisation chooses which products a retailer keeps in its range, adds or drops. It weighs what each item earns after its costs, the shelf space it takes and how many shoppers would switch to a substitute if it disappeared.
The problem
The range review is on Thursday. You bring a list sorted by sales, with a mark near the bottom. Everyone agrees to cut everything below the mark. Six months later the range is as long as before, and you earn no more than before. Nobody can say which cuts lost sales and which ones cost nothing, because shoppers simply bought another of your products.
Signs your range has grown too long
Ranges grow for years on “let’s just add this line, a customer asked for it”. Every product in every size or flavour is its own stock-keeping unit (SKU), and the count only goes up. The warehouse fills with lines that sell a few units a month. You spend hours on items that barely sell. The rest of the range pays for them in storage, planning effort and capital tied up in stock.
Why “cut the tail” fails
The usual review ranks SKUs by sales and cuts the tail. It scores each line on margin alone: sale price minus purchase price. What the line costs in the warehouse and in planning is left out. So lines that cost more to carry than they earn stay in the range, because their margin looks fine on paper. Each SKU is also judged on its own. So the review misses cases where a cut costs almost nothing, because shoppers simply buy another of your products instead.
It also misses the opposite case. A new line may take most of its sales from your own similar products. Then adding it earns little. A flat rule like “cut the bottom 10%” can’t tell which lost sales go to competitors. And new lines get added without a test, so the range refills within a season.
What changes
Today, the range review sorts by sales, cuts the bottom and hopes. With our model, every line gets a score: what it earns once you subtract what it costs to stock and plan. The model shows where its sales would go if you cut it. The review discusses a short list with reasons attached. New lines face the same test before they get shelf space.
SKU rationalisation: what you get
You see which SKUs earn their place and which cost more than they earn. You see which cuts are nearly free because the sale stays in your range, and which ones cost revenue. Each new line is judged only on the demand it adds, so the range stops growing straight back after every review. Reordering and stock-keeping get simpler and cheaper with every SKU you drop.
We build the model on your own sales data, basket data and shelf-space data. Basket data shows what each shopper bought in one visit. From your sales history we estimate where sales move when a line goes or arrives. Before it goes live, we check that against cases where your customers actually switched to another line. From then on, you carry the SKUs that earn their place.
- A model written around your own assortment and SKU rationalisation rules.
- Results measured in the numbers you already track.
- It slots into how you already work, and you can check every answer it gives.
How it works
Your sales, margins and shelf space go into one model. So does what each line costs you while you keep it in the range. It also shows where a line’s sales would go if you dropped it. You get a short list of lines to cut or keep, each with its reason. New lines face the same test before they get shelf space.
Contribution analysis by SKU
The foundation is contribution by SKU: what a line earns after all the costs it causes. Many retailers don’t calculate it. It starts from the margin, sale price minus purchase price. Say a pack sells for 10 francs and costs you 6. The margin is 4 francs. Then you subtract the full cost of carrying the line. That covers storage, handling and planning hours. It also covers the money tied up in stock and the risk that goods expire or go out of fashion. A line can look profitable in your accounts. Once you count what it costs to keep in the range, it can turn negative.
Space productivity
Space productivity asks the same question on the shelf. Every SKU takes shelf or slot space, and that space costs money. Lines that earn less per metre of shelf than that metre costs are candidates to cut. Their margin on paper doesn’t change that. A leaner range also makes your inventory policy simpler and cheaper. That’s the set of rules for how much of each line you hold.
Substitution analysis
Remove a SKU, and the sale doesn’t simply vanish. It moves, and the question is where. When a shopper buys another item instead, that’s called substitution. If that item is also yours, cutting the line costs almost nothing. You keep the sale, drop the cost and get a tighter range. If the shopper goes to a competitor instead, the cut costs you that revenue.
If a review can’t tell the two apart, it cuts too little because it fears every loss. Or it cuts too much and doesn’t see the sales it loses. The model estimates which SKU picks up the demand. It uses your basket data, what each shopper bought in one visit, and your sales data from weeks when an item was missing.
Reading substitution from your own stockouts
You don’t have to cut a line to learn where its buyers go. A stockout is a time when a product was missing from the shelf. Every stockout is a small experiment you’ve already paid for. When a product was off the shelf for a week, look at what sold instead. If similar items sold more that week, the demand moved to them. If sales of the whole category fell, those shoppers left.
The split can be large. In 2002, researchers at Emory, St. Gallen and Colorado ran a study on out-of-stocks for the grocery trade. They asked shoppers what they do when an item is missing. On the worldwide average, 31% said they go to another store and 26% switch brand. Another 9% buy nothing at all. In your categories these shares may be higher or lower. That’s why we estimate them from your till data, not from a survey average.
Cannibalisation modelling in retail
Cannibalisation means a line takes its sales from your own range. A new flavour takes much of its volume from similar flavours in your range. With two pack sizes that are almost the same, one lives off the other’s sales. The model estimates how much of each SKU’s demand is new and how much comes from a similar product in your range.
Without that, every new line looks better than it is, and every cut looks worse. With it, you score each line on net incremental contribution. That is the profit you gain because you carry the line. Sales it only takes from your other lines don’t count. We estimate substitution and cannibalisation from your sales data, and update them whenever you add or drop lines. The optimisation then decides the range.
New product screening
Every new SKU should pass a test before it gets shelf space. The test: the line must be expected to add profit. To work that out, you subtract two things: the sales it takes from your own lines and the full cost of its shelf space. This test before listing is what stops the range from growing straight back after each review.
Assortment optimisation with MILP
Assortment optimisation then looks at the whole range at once. Mixed-integer linear programming (MILP) writes the choice as equations with yes-or-no decisions: keep this line or not. Whether to keep, cut or add a line is decided in one calculation. The model chooses the range and the space per SKU that bring the highest net contribution. It stays within the shelf and warehouse space you have. And it keeps the range as wide as your customers expect.
This is product range optimisation built on your data, with every decision traced back to a number. It runs on data you already have: basket data shows substitution, till history shows cannibalisation, and space data shows productivity.
Common questions.
Something else? Ask us directly.
It’s the observation that a small share of your items usually brings most of the sales. A SKU is a single item in your range. That’s a useful first look and a poor rule for deciding what to drop. An item that sells little can be the reason a shopper comes in. One that sells a lot can earn nothing once you count what it costs to store and handle. So rank items by what they earn after costs. And ask: if the item were gone, would shoppers buy another of your products or go to a competitor? Then use the 80/20 list as a quick cross-check.
Review your weakest lines at a steady rhythm, say once a quarter. Review the whole range once or twice a year, when you rearrange the shelves anyway. The test for new lines matters more than the rhythm. Only add a new item if it brings more profit than it takes away from the items you already sell. Then fewer items that add nothing make it into your range, and the next review has less to drop.
Usually some of each, and the split decides whether the cut pays. You can estimate it before cutting. Look at weeks when the item was out of stock and check whether similar items sold more. Or take the item off the shelf in a few stores and compare their sales with the other stores. Then drop the items whose buyers switch to another of your products. Keep the items whose buyers would go to a competitor.
Sometimes it can, and if that’s the case, we’ll tell you. Standard software is built for an average operation, though. Yours runs on its own rules: your contracts, your exceptions, the things you know by heart. A custom model writes those rules in, and that’s where the better plan comes from.
You don’t need either to start. We build the model from the data you’ve already got, even if it lives in three spreadsheets and one person’s head. You get a system you run yourself. We train your team until they no longer need us.
You see a first result with the prototype, before you commit to anything bigger. We start with a free 30-minute call about your operation and the model that would fit it. Then we build a working prototype and run it on your numbers.
Similar pages.
Terms on this page
Every abbreviation explained in plain words, in our article “Metrics and methods”.
Tell us what you’re planning.
In a free 30-minute call with a specialist, you find out whether optimisation can help.
- Free, no commitment
- You talk to a specialist directly
- A clear go or no-go