Range Architecture: How Many Products Should A Brand Have

Ask a three year old brand how it ended up with thirty four SKUs and nobody can tell you. There was no meeting where somebody decided thirty four was right. There was a series of individually reasonable additions: a variant a retailer asked for, a flavour a founder liked, a festive pack that never got retired. The range was not designed. It accumulated.

Range architecture is the discipline of treating the number of products as a decision rather than a residue. It asks what each product is for, what it takes from the others, and what it costs to keep alive.

A range is a set of deliberate choices

Start from the shopper, not the catalogue. In any category a shopper is answering a small number of questions: which one is for me, how much do I want to spend, and how much do I want to buy. A range exists to answer those questions quickly. Every product that does not help answer one of them is making the decision slower.

That is why the test for adding a product is not whether it will sell. Almost anything will sell something. The test is whether it makes the range easier to choose from.

Good better best, or need states

There are two clean ways to structure a range, and mixing them badly is where most catalogues get confusing.

Good better best is a price ladder. Same job to be done, three levels of specification and price. It works when the category has a genuine quality perception, when the shopper can see what the extra money buys, and when trading up is a realistic behaviour. It fails when the differences are invisible on a product page, because then the ladder just becomes a discount conversation and the shopper picks the cheapest rung.

Need state segmentation splits by occasion or problem rather than by price. Daily use versus travel. Oily skin versus dry. Morning versus night. Everyday versus gifting. It works when the needs are genuinely different and the shopper can self identify in one glance. It fails when the needs are actually the same need with different marketing on the pack, which the shopper detects immediately.

Pick one as the primary axis and let the other operate as a secondary layer at most. A range that ladders on price and splits on need state and also offers three sizes in each has produced a grid nobody can hold in their head, least of all a quick commerce shopper deciding in eleven seconds.

The hero SKU and what it earns

Almost every functioning brand has a hero. It is the product people name when they name the brand. It usually carries a disproportionate share of volume, it is the one that gets reordered, and it is the reason the brand has any distribution at all.

The hero earns three things the rest of the range does not. It earns the marketing budget, because acquisition spent on the hero converts best and creates the memory the rest of the range borrows from. It earns shelf and listing priority, because retailers and platforms allocate space to proven rate of sale, not to breadth. And it earns supply chain protection, meaning it is the SKU that must never go out of stock, even if that means someone else waits.

The common error is treating the hero as solved and spending attention on the newer, more interesting products. The hero is not solved. It is the asset. A brand that lets its hero drift on availability, reviews or content while it launches variant seven has misallocated its scarcest resource, which is attention.

Line extensions cannibalise more often than they expand

The pitch for a line extension is always incremental. A new flavour reaches shoppers the current flavour does not. A new size opens a new occasion. Sometimes that is true. More often the extension sells to people who were already buying, and the category share stays flat while your SKU count rises.

You can test this rather than argue about it. After launching an extension, look at total brand volume in the affected segment, not the extension’s own sales. If total volume is roughly flat while the new SKU builds, you have moved demand sideways. Then check whether extension buyers are new to the brand or existing customers who switched, and watch rate of sale on the original.

Cannibalisation is not automatically a failure. Sometimes you cannibalise deliberately, into a better margin format or a defensive position. The failure is cannibalising by accident and then counting the extension’s revenue as growth.

What every extra SKU actually costs

The per unit cost sheet does not capture the cost of existence. Each additional SKU carries a standing tax across the business.

In catalogue, it is images, copy, attributes, compliance fields and identifiers, maintained on every platform you sell on, forever, and re-audited every time a platform changes its requirements. In inventory, it is a separate forecast, a separate safety stock, a separate minimum order quantity, and a separate chance of both stockout and dead stock. Fragmentation is the quiet killer here: the same total demand spread across more SKUs needs more total buffer stock to hold the same service level.

In supply, it is another set of components, another artwork version and another quality standard. In demand, it is a share of a finite marketing budget. In support, it is more variants for customer service to know, more return reasons and more ways to send the wrong thing. And on the shelf, a range with too many faces gets pruned by the retailer or the platform anyway, on their logic rather than yours.

Decide what the new product replaces

The single most useful rule in range architecture is this: every proposal for a new product must name what it replaces. Not necessarily one for one, and not always a deletion. It might replace a size, absorb a variant, or take over the job of a slow seller. But the question must be answered before approval, because it forces the range to stay a set of positions rather than a growing list.

Pair it with a scheduled review where the bottom of the range is examined on contribution rather than sentiment. Ask what job each product does and what happens to the shopper if it disappears. The answers are usually clearer than people expect. The difficulty is never the analysis. It is the willingness to delete something somebody worked hard on.

There is no universal right number. It varies by category, by channel and by how much distribution you can genuinely support. What is universal is the shape of a healthy range: a hero that is protected, a small number of products with clearly different jobs, one primary organising axis a shopper can read instantly, and a tail that gets cut on schedule rather than when the warehouse runs out of space.