When a broad assortment helps and when it hides weak items

A broad assortment is useful when it serves real and distinct demand: it brings in customers who cannot find what they want elsewhere, or it completes purchases of more profitable goods. Without such a role, it spreads stock money and the team’s attention across products customers do not choose. Research also shows that cutting slow sellers across the board can cost a store more than it saves. So the question is not “how many items” but “what role does this group play”.

Breadth has two faces

On the benefit side, the effect is well documented. A study of online booksellers in 2000 found they offered more than 23 times as many titles as a typical large bookstore, and that the value of this wider choice to customers was seven to ten times larger than the gain from lower prices. Follow-up research at a multichannel retailer found that online sales were spread across a wider range of products than catalogue sales, even though both channels offered the same products at the same prices. A higher share of niche products went hand in hand with the use of search and recommendation tools. Slower items sell when customers can find them.

On the cost side, an important correction came at the same time. An analysis of music and film services showed that niche titles are consumed mainly by heavy users, who on average rate them lower than popular titles. The author therefore advises strictly managing the cost of items that rarely sell and using popular products to win customers. For a store holding its own stock, this matters: a digital catalogue costs nothing to extend, physical inventory does not come free.

What happens when you cut the range across the board

The best available evidence comes from an online grocer that removed 24–91% of items in every category, mostly slow sellers, and compared the results with a control group of customers. Total sales fell. Customers shopped less often and bought less, and the lower shopping frequency cost more than the smaller baskets. The impact varied widely by category, with less frequently purchased categories hit hardest.

The same authors note that an earlier analysis of selected categories at the same retailer found no drop in sales. They explain the difference by that analysis covering only part of the categories, while the loss arose across the whole store. A study at a large Dutch retail chain adds the time dimension: after a quarter of a category’s items were delisted, there was a substantial short-term sales drop but only a weak long-term effect. The loss fell mainly on former buyers of the delisted items and was partly offset by new buyers the narrower category attracted.

Three lessons follow for an online store. Measure the impact across the whole store, not only in the category you narrowed. Track purchase frequency, not just basket size. And expect the first weeks after a cut to look worse than the long-term result. Both studies are foreign and dated, so do not carry over the percentages; the direction and mechanism are what transfer.

Judge groups, not single items

A single item with low sales tells you little. It is more useful to split the range into groups – by category, brand, price band or purpose – and set several numbers side by side for each: share of items, share of revenue, gross margin, stock value at purchase cost and availability. Then add the most important piece: a named role.

An illustrative example with invented numbers:

Judged on revenue alone, the add-ons would look weak, yet they carry a high margin and tie up little cash. In an off-season quarter, the seasonal range would look like dead stock. The real question is raised only by the last group: almost half of the items and the stock money, eight per cent of revenue and the lowest margin. Even here this is not a verdict yet – just the place to start investigating.

Low direct sales do not make an item weak

Before you label a group weak, rule out four common explanations.

Discoverability. If customers do not see the items in listings, search or recommendations, low sales measure navigation rather than demand. Compare product page views with sales: few views point to a discoverability problem, many views without sales point more to price, presentation or the offer.

Availability. An item that was sold out for half a year could not sell. Calculate sales velocity only for the period when it was in stock.

Role in the basket. Research on supermarket product selection showed that a product’s value should include the profit from the purchases associated with it. An item ranked last on its own profit can rank among the most important on total contribution. Check what share of orders contain the group alongside more profitable goods. Be careful, though: appearing in the same basket does not prove the purchase would not have happened without the add-on. The customer may simply have bought it elsewhere.

Season. Assess at least a full year, or you will cut seasonal goods at the time they are not meant to sell.

When breadth really hurts

The popular idea that more options always put people off buying does not hold. A meta-analysis of fifty experiments found an average effect of choice set size that was virtually zero, with large differences between studies. A later meta-analysis showed that choice overload appears under specific conditions: when options are complex and hard to compare, the decision is difficult, the customer is unsure what they want and wants to keep the choice as effortless as possible.

In practice, breadth hurts mainly in three situations. When a category holds many near-identical variants customers cannot tell apart. When stock without a role ties up cash that the core range or add-ons need. And when the team spends time managing, describing and buying items that bring nothing. Items made to order or shipped by a supplier without your own stock cost far less than goods on the shelf, so their breadth is easier to justify.

How to narrow the range without unnecessary loss

Narrow by group and in stages. First stop reordering items without a role. Then sell through the stock, ideally with a nudge to customers who bought them before. Only then hide them. Leave deletion until last, and only for items with no visits from search engines: Shoptet offers the visibility setting “Only via URL, cannot be ordered”, which removes a product from categories and search while keeping its page reachable, so a customer arriving from a search engine does not land on a 404 error.

Before you act, write down what you expect, and afterwards watch the whole store: revenue, number of orders, share of returning customers and sales in related categories. Research shows a loss can show up in less frequent purchases before it shows up in smaller baskets.

What to do now

1. Split the range into groups. Use categories, brands or price bands, depending on how customers shop.
2. Set five numbers side by side. Share of items, share of annual revenue, gross margin, stock value at purchase cost and share of days in stock.
3. Name the role of each group. Does it bring customers in, complete purchases, cover a season – or none of these? A group without a role is a candidate for a closer look.
4. Rule out false signals. For each candidate, check product page views, availability, co-occurrence with more profitable goods and seasonality.
5. Narrow gradually and measure the whole store. Stop reordering first, then sell through and hide; track the result in revenue and purchase frequency.

For stores that sell unique items, Korzaro’s Products section builds a buying map: categories and price bands with median time to sale, actual margin and the share of items that sold versus items that lingered or disappeared without a sale. Groups with too small a sample get no recommendation, and older stock tying up cash for a long time is shown separately. What role a category plays in the range, however, is the owner’s call, not the table’s.