Which Products Should You Kill? 5 Tests Before You Cut a Product Line

In 2020, Coca-Cola looked at a portfolio of more than 400 master brands and concluded that roughly half of them were single-country products with little or no scale, together contributing about 2% of the company's revenue. It killed them. Tab, a brand with a 57-year history and a genuine cult following, went with the rest.

That decision is worth sitting with, because most owners will never face a portfolio that size and still find the same decision impossible. The reason is not analysis. It is that adding a product is somebody's job and removing one is nobody's.

Every product line grows a tail, because launches have owners and discontinuations do not.

Bain & Company's consumer research puts numbers on what that tail costs: complexity of this kind can inflate supply chain costs by as much as 25%, while reducing it has been associated with sales growth improvements of two to five percentage points and margin gains of 100 to 400 basis points. McKinsey has estimated that product complexity costs U.S. food and beverage manufacturers around $50 billion in gross profit a year.

I run a group founded in Azagra in 1890 that now operates eight divisions in more than 75 countries. Wine alone gives you a permanent education in this problem: every vintage, every format, every label variant for every market is a decision that survives long after anyone remembers why it was made. Below are the five tests I make people pass before a single item comes off the list.

Test 1: Contribution, not profit

Here is the trap that ruins most rationalization programs. You run a profitability report by product, discover that the bottom 30% of your range loses money, cut it, and the following year your margins are worse.

The reason is that the report lied to you. Product profitability is calculated by allocating fixed overhead across the range, usually by revenue or volume. When you delete the products, the overhead does not leave with them. It redistributes onto the survivors, and a new group of products now appears to be losing money. Cut those, and it happens again.

Judge a product on contribution margin, the revenue it brings minus the costs that genuinely disappear if you stop making it, and treat every allocated cost as guilty until proven avoidable.

So the working question is not "does this product make money?" It is "what cash actually walks out of the building if I stop making this tomorrow, and what cash stays?" Raw materials, direct labor and freight leave. The plant, the salespeople, the ERP licence and the finance team stay. A product with a positive contribution and no avoidable overhead attached to it is not a loser. It is paying rent on capacity you have already bought.

A product only becomes a genuine candidate for removal when the honest answer to that question is one of three things: contribution is negative, contribution is small and it consumes a scarce resource that something better could use, or removing it releases a real, identifiable cost.

Test 2: The customer basket

The fastest way to destroy revenue with a rationalization program is to analyze products in isolation.

Before you remove anything, pull the list of customers who buy it and ask what else they buy. You are looking for two patterns. The first is the orphan: a customer whose relationship with you begins with that one item, where cutting it does not cost you one line but the whole account. The second is the anchor: a low-margin product that gets you shelf space, a listing, or a place on a wine list, from which everything else follows.

In distribution this is brutally concrete. A restaurant that lists your entry wine and then buys three reservas is not four separate decisions. It is one relationship with four line items, and the cheapest of the four is the one that opened the door.

A product's value is the margin of the basket it appears in, not the margin of the line it occupies.

The discipline is simple: no item gets cut on its own numbers alone. It gets cut on its numbers plus the revenue at risk in the accounts that buy it. Sometimes the right answer is not to delete the product but to reprice it, or to keep it for two customers and delist it everywhere else.

Test 3: Does anything actually get released?

Cutting a product creates value only if it frees something you are short of. So name the resource before you cut, in writing.

The usual candidates are real and measurable. Production capacity lost to changeovers, where a plant running twelve formats instead of thirty gains hours it can sell. Working capital locked in slow inventory, which is often where the tail does its real damage: a long tail of products typically accounts for a disproportionate share of slow-moving stock while producing a fraction of the margin. Warehouse space. Planning and forecasting effort, which is a finite management resource that gets consumed by the items that need it most and deserve it least. Sales attention, the scarcest resource of all, since a salesperson with forty items to present sells the five they like.

If you cannot name the resource, you are not doing rationalization. You are doing housekeeping, and housekeeping does not show up in the P&L.

Test 4: The strategic role test, honestly applied

Some products earn their place without earning margin. An entry product that recruits customers who trade up. A halo product that sets the perceived quality of everything beneath it. A defensive product that keeps a competitor out of an account. A product that fills a gap in a range a buyer wants covered before they will list you at all.

These roles are legitimate. They are also the single most abused argument in any portfolio review, because every product manager can construct a strategic story for the product they own.

A strategic role is only real if you can state the mechanism, name the products the customer trades up to, and show that the trade-up actually happens in the data.

Ask for the number. What percentage of customers who buy the entry product go on to buy something else, and over what period? If nobody has measured it, the role is a hypothesis, and it should be given a deadline: measure it this year, or it is treated like any other line.

Test 5: The cost of leaving

Deletion is not free, and finance rarely models the exit properly.

The bill usually contains four items. Obsolete inventory and components that have no alternative use, which have to be written off or discounted. Contractual commitments: minimum purchase agreements with suppliers, and customer contracts that promise continuity of supply. Tooling, moulds, dry goods and packaging inventory bought for that item alone. And the shelf slot or listing you surrender, which is often the most expensive item on the list because you will not get it back cheaply if you change your mind.

None of these are reasons not to cut. They are reasons to sequence the cut. Run down the inventory, honor the contract to its end date, then delist, rather than announcing a decision that triggers a write-off in the same quarter you were trying to improve.

How to actually run it

Three moves, in this order.

• Freeze the tail. Before you delete anything, stop the inflow: a rule that no new item launches without a named item leaving. This alone stabilizes the problem while you work on it.
• Cut in one visible wave, not in a permanent trickle. A single, explained, well-communicated cut lets customers plan and the sales force sell the story. Continuous quiet deletions teach customers that your range is unreliable.
• Prove the release. Six months later, go back to the resource you named in Test 3 and measure it. If capacity, inventory or sales attention did not move, the exercise was cosmetic and you should say so out loud, because the alternative is that the tail regrows and nobody notices.

The uncomfortable part of this work is that it is subtraction, and subtraction never feels like progress. Nobody gets promoted for the product they discontinued. But a range that has been pruned deliberately sells better than a range that has never been questioned, and the companies that do it well do it on a schedule, not in a crisis.

Key Takeaways

• Every product range grows a tail because launching has an owner and discontinuing does not. Fix the inflow first with a one-in, one-out rule.
• Product-level profit reports mislead, because allocated overhead does not leave when the product does. Decide on contribution margin and avoidable cost only.
• Never evaluate a product alone. Check the customer basket: a low-margin item that opens an account is worth the margin of the whole account.
• Name the scarce resource the cut will release, in writing, before you cut. If you cannot name it, you are tidying, not rationalizing.
• Strategic roles like entry products and halo products are real, but they require a measured trade-up rate. An unmeasured role is a hypothesis with a deadline.
• Model the exit: obsolete inventory, supply and customer contracts, dedicated tooling, and the listing you surrender. Sequence the cut around them.
• Cut once, visibly, with an explanation. A permanent trickle of quiet deletions teaches customers that your range cannot be relied on.

Frequently Asked Questions

What is SKU rationalization?

SKU rationalization is the structured review of a product range to decide which items to keep, reprice, consolidate or discontinue. It looks at each item's contribution margin, the customers and channels it serves, the operational complexity it creates, and the resources its removal would release. Done properly it is a recurring discipline on a fixed calendar, not a one-time cost-cutting event.

How do you decide which products to discontinue?

Start with contribution margin rather than allocated profit, then apply three filters: whether the item's customers buy anything else from you, whether removing it releases a resource you are genuinely short of, and whether it plays a measurable strategic role such as recruiting customers who trade up. An item that fails all three filters and shows thin or negative contribution is a candidate. Then price the exit before you act.

Why do margins sometimes get worse after cutting unprofitable products?

Because most product profitability reports allocate fixed overhead across the range. When products are deleted the overhead they carried is redistributed onto the remaining items, so a new tier of products appears unprofitable and the cycle repeats. This is the classic death spiral of cost allocation, and the way out is to measure avoidable cost, the cash that genuinely leaves the business, instead of allocated cost.

What percentage of SKUs should a company cut?

There is no universal figure, and any target set in advance is a signal that the analysis is being reverse-engineered. Published consumer-goods programs commonly land near a quarter of items, but the right number falls out of the tests, not the other way around. A more useful target is a rule that constrains growth of the range, such as one item out for every item launched.

Does cutting products always reduce revenue?

Not necessarily. Some revenue is lost, but the freed capacity, working capital and sales attention frequently offset it, which is why disciplined programs have been associated with sales growth improvements as well as margin gains. The revenue that does not come back is usually the revenue lost by cutting an item without checking the customer basket first, which is the single most common execution error.

One thing to do this week

Sort your product list by contribution margin, not by revenue and not by reported profit, and draw a line under the bottom 20%. Do not cut anything. Just pull the customer list for those items and see how much of your revenue sits in accounts that buy them. That one page will tell you whether you have a pruning problem or a pricing problem, and they are not fixed the same way.

See how the eight divisions of Manzanos Enterprises fit together, and for the decisions that sit either side of this one, read why focus beats opportunity, how to raise prices without losing customers and how to cut costs without cutting muscle.