Retail Assortment: How Pros Decide What to Keep, Drop, or Replace

I use a simple test: does the product earn its spot through repeat purchases and margin, or only through first-time sales?

At Simply Noted we sell handwritten notes at $0.89 each, and over the years we have tested add-ons like custom card stock, gift inserts, and different note formats. Some looked good on sales volume alone. When I added margin and repeat rate, the picture changed. One premium option sold decently but was complicated to produce, slowed our robots, and generated support questions. I dropped it and nobody missed it.

The simplicity test matters as much as the sales number. Ask: can a new customer understand this in one sentence, and can we make it consistently? If a product needs a long explanation or manual work, it costs more than it shows on the spreadsheet.

My rule: keep what sells repeatedly and is easy to explain, replace what sells once but confuses people, and drop anything that is neither. We review quarterly with three numbers: repeat rate, margin, and support tickets per hundred orders.

Every SKU on my shelves has to earn its spot against forecast demand, and the ones that can't get cut before they become a replenishment problem. I run a rolling review where I pull sell-through rates by product and compare them against my replenishment forecasts. If a SKU is consistently underperforming its forecast over two restock cycles, it goes on a retirement list. That keeps my buying team from automatically reordering slow movers just because they're still in the system.

The other half of this is clustering. My customers in different channels and regions don't want identical assortments, so I group stores and online segments by purchase patterns and keep a tighter core set of high-demand, high-margin SKUs in every cluster, then flex the remaining slots for local or seasonal preferences. That gives me cleaner shelves without stripping out products my core shoppers depend on.

When assortment decisions feed directly into replenishment this way, I stop accumulating dead inventory and free up capital for the products that move. If a product can't justify its reorder against current demand data, I take it off the shelf no matter how long it's been there.

A slow-moving product may still complete an important customer order. I would test a proposed substitute with regular buyers before removing the original from the standard range. Track accepted swaps alongside lost or delayed orders, then clear shelf space only where the replacement meets customers' needs. Keep exceptions available to order where practical, so a simpler range still protects customer commitments.

When retail teams shift from looking at sales volume separately and look at products from the point of view of gross margin return from inventory investment and the speed of the supply chain, advancement in assortment optimization is realized. By determining costs hidden in the operations that slow moving goods cause in the warehouse and on the shelves, it becomes possible to find the best way to balance the two opposing aims of maintaining sales figures and of keeping the operations efficient. A practical rule of thumb known as the cross-contribution test applies remarkable results. According to this test measures any product located in the lowest quartile of inventory turnover must have a gross margin considerable enough, meaning at least double the mean in the category earned by other products in that category, enabling the retailer to consider that product as worthy of the space it occupies in warehouses and on shelves.

In my practice of providing enterprise resource planning systems to retail and distribution companies, I constantly observe that retail teams directly face the consequences of assortment bloat. The root cause of it is the habit of not getting rid of the products in stock that bring minimal return, since there is a fear of losing the loyalty of a few customers. To illustrate the issue, I have to analyze stock movement speed and complexity and use the cross-contribution test, which enables efficient contention for space on the shelves based on the processing of such financial data as marginal profit earned from the least effective product in stock. The cross-contribution test stands quite helpful for the operations and procurement teams in deciding whether to remove a certain product from the shelves. Thus, the process of removing products from the shelves becomes much more secure due to the company using valuable information about stock profitability.

When deciding which products to keep, drop, or replace, I balance sales protection with simpler operations by looking at repeat demand rather than sales volume alone. In our family business, I've learned that a product can look important on paper because it sells occasionally, while actually creating extra handling, inventory, and scheduling complexity for very few customers. My rule of thumb is to never remove something based on low volume alone—I first check whether a small group of core customers buys it consistently and would have a difficult time substituting another option. We've had items that weren't major revenue drivers but were important enough to long-standing customers that keeping them protected the relationship. On the other hand, when demand is sporadic and there's a practical substitute, simplifying the assortment usually makes operations cleaner without creating meaningful customer friction. That repeat-customer test has helped me distinguish between products that are genuinely valuable and products we're simply accustomed to carrying.

When deciding which products to keep, drop, or replace, I balance sales protection with simpler operations by looking at whether each item earns the space and attention it requires. My rule of thumb is to never remove something based on sales volume alone; I also look at repeat demand and whether customers regularly ask for that specific option. I've seen a similar principle play out when helping homeowners choose pool finishes, equipment, and replacement parts—more choices can actually make decisions harder without producing a better result. On one project, narrowing several nearly identical equipment options down to the few that had proven reliable made the decision easier for the homeowner and simplified future servicing. I'd test removing or consolidating low-volume, highly substitutable products first while protecting items with strong repeat purchasing or a distinct customer need. That approach creates cleaner shelves and simpler operations without taking away the products core shoppers genuinely depend on.

Assortment management is quite a complex task, and most retailers use quite a simplified approach, just evaluating products to keep by sales, which is slightly misleading. The key decision factor here is incrementality of every SKU. It means that our products are not sold in a vacuum, they can be substitutes of each other. So if you delist one product, you can lose the customer, or demand of this product can be reallocated to the existing products. And vice versa, if you add a new SKU to the assortment, you can get more demand, more customers, increase of consumption, or this new product can just cannibalize demand of the current assortment. So the most important analytics we should do here is to analyze transactions and planogram changes, assortment changes, for some period of time, and understand which products, which categories really add incremental demand instead of cannibalizing existing demand.

I've seen in my experience a very good approach when a retailer is making assortment decisions not based purely on sales performance but on real customer reviews and feedback for those products. They have a rule: if the NPS for the SKU is low after two weeks, they immediately delist it, so that their assortment has the highest rating among their customers.

We don't have retail shelves in the traditional sense, but with thousands of promotional products online, we face the same assortment problem. Keeping every product forever sounds like more choice for the customer, but eventually you create duplicate products, harder navigation and more work for the team when several items are basically doing the same job.

One rule we've started using is "replace before you remove". Before dropping a slower or older product, we ask whether we can confidently point the customer to an equal or better alternative at a similar price point, lead time and branding capability. If the answer is yes and the product isn't serving a distinct purpose, there's a much stronger case for taking it out.

In retail this is often called "SKU rationalisation," which simply means reviewing whether each product still earns its place in the range. We don't judge that purely on sales either, because a lower-volume product might still matter if it fills a particular price point, has an unusual branding option or is important to a certain type of client.

That test has helped us simplify parts of the catalogue without making customers feel that we're reducing their choices. It also makes life easier for our account managers because they're recommending between a smaller number of genuinely different options rather than explaining five products that are almost identical.

The tradeoff is that sometimes we'll keep a slower seller because there isn't a good substitute. I'd rather carry a little extra complexity than remove something customers genuinely need just because it sits near the bottom of a sales report.

For me, the goal isn't the smallest possible range. It's making sure each product gives the customer a reason to choose it and, when it doesn't, having a better option ready before we take it away.

On Amazon the shelf is not physical, but Amazon charges you for clutter anyway. The aged inventory surcharge starts biting at 181 days and climbs steeply past 271 and 365 days, and weak sell-through drags your IPI score, which is what governs your storage capacity. So a bloated catalog does not just sit there quietly. It steals capacity from the ASINs that actually move.

The test we use now is three numbers side by side, pulled from the Inventory Age report and the Business Report: trailing 90 day units, contribution margin after FBA fees, and the share of that ASIN's sales that came from advertising. If a product is low volume, thin after fees, and more than half ad driven, it is not a product, it is a subsidy. That is a drop candidate.

The piece that protects sales is the exception you check before cutting. Look at whether the ASIN is a child inside a variation family that ranks. Children that look dead on their own often carry the parent's review count and keyword relevance, and pulling them can cost you more than they ever earned. Same with the odd size or color your repeat buyers quietly depend on.

The rule of thumb I would keep: cut by margin after fees, never by revenue. Revenue only tells you what shoppers clicked. Margin after FBA fees and ad spend tells you what you were allowed to keep. When we run that test, the ASINs that fall out are almost never the ones anyone came for, which is exactly why nobody misses them.

I drop a SKU when it cannot earn a named job in a normal wash-day basket, even if it still sells. Protecting revenue without that test just fills the shelf with jars the weekday inbox has to apologise for.

Every keep has to sit in one clear seat among cleanse, leave-in, define and seal, and it has to survive a week of product questions without me rewriting the reply script. If the jar only works as a special case, or needs a founder note every time porosity is wrong, it gets replaced. Core shoppers rarely miss the SKU that never earned its seat. They notice when the edit stops contradicting itself. We hold the range at twenty-eight jars across four brands because that test keeps biting.

In The UK Wash-Day Report 2026, https://zenvy-beauty.com/blogs/news/uk-wash-day-report-2026, the average UK curl routine used 5.2 products. An assortment that ignores that number creates support work the P&L will not forgive.

Before a brand reorders anything, I ask one question: would you place this exact order again tomorrow, same style, same quantity? If the honest answer is "yes, but," it isn't a keep. It's a replace.

The hard calls are the styles that sell well and still cause trouble, usually fit returns or a fabric that pills, because dropping a best seller scares people and the core shoppers who love it would notice it's gone. So we usually fix it on the next run instead: same look, better pattern or better fabric. Regulars keep buying and only notice that it fits better.

For anything new, test small. A new style starts at 50 pieces, not 500, so a miss never turns into a garage full of stock.

I start with understanding the replenishment rate of each SKU carried. Then I separate, hero/core products from seasonal products and products that haven't sold through in 4 months. You need different categories to keep the relationship between the consumer and the product desirable and intriguing, so having' seasonality with SKUs is important. I account seasonality before deciding if something is underperforming. My rule of thumb is if we haven't had to replenish a nonseasonal product in four months something needs to change. Sometimes a product that is sitting on the shelf and hasn't moved, isn't the problem; it may be how we are positioning it operationally.

In fast fashion, we decide what to keep, drop, or replace by sell-through velocity while a piece is in stock, tracked with Shopify automations rather than gut feeling. A style that sells through fast at full price earns a reorder and more sizes or colors when available; one that stalls gets cleared and not repeated, even if we loved it. The rule that most improved our results was buying to the sizes and styles that actually move instead of spreading evenly across the board, which freed up cash and shelf space for proven winners. Let the customer's buying, not your taste, decide the assortment.

-Whitney O'Keefe, co-owner, Avery Mae Boutique (averymae.com)