Bundling slow stock is often better than marking it down, but not for the reason usually given. The argument that a bundle "preserves perceived value" while a markdown destroys it is true and unmeasurable. The argument worth making is arithmetic, and it only works under conditions you can check before committing.
The comparison, with the missing term
Product A sells steadily at £40. Product B is stuck at £30. Both cost £15.
Markdown: B at 50% off sells at £15 — you recover the cost and nothing else. Margin £0.
Bundle: A + B at £56, a 20% discount. Cost £30, margin £26.
Against a markdown, the bundle looks decisive. But that is the wrong comparison, because A was going to sell anyway. The honest one is:
A sold alone at £40: margin £25, and B still in the warehouse.
A + B bundled at £56: margin £26, and B gone.
The bundle earned £1 more and cleared a unit of dead stock. That is a genuinely good trade — dead stock costs money to hold and eventually gets written off — but it is a £1 trade, not a £26 one. Anyone selling you on bundling for clearance is quoting the first comparison.
The condition that decides it
Everything turns on one question: would the customer have bought A on its own?
If yes, the bundle costs you £14 of discount to move one unit of B — which beats the markdown, but only just, and only while B's holding cost is real.
If no — if the bundle is what made them buy at all — you gained a £40 sale and cleared B, and the offer is excellent.
Which case you are in depends on how much of A's volume the bundle captures. If nearly every A buyer takes the bundle, you are discounting a product that was selling fine. If a modest share do, you are mostly converting people who would otherwise have bought nothing.
This is measurable. Run the bundle for two weeks and watch A's standalone sales. If they fall by roughly the number of bundles sold, the bundle is cannibalising. If total A volume rises, it is recruiting.
Four structures, and when each fits
Pair the slow mover with a bestseller
The default, and the one most exposed to the cannibalisation above. Use it when the slow product genuinely complements the fast one, and cap it — a limited number of bundles, or a limited window — so it does not become the default way your bestseller is bought.
Discovery bundles
Two or three lesser-known products at a meaningful discount, framed as exploration rather than clearance. This is the structure with the least cannibalisation risk, because nothing in it was selling well anyway, and it carries a second benefit: it generates data on which of the slow products people actually like, which is worth more than the margin on the bundle.
The failure mode is transparency. Customers recognise a bundle of things nobody wanted. It works when the products are genuinely good and simply undiscovered; it does not rescue products that are slow because they are poor.
Seasonal transition bundles
End-of-season framing explains the discount without implying the product is worth less. "Last chance summer" is understood as timing, not quality, and the customer accepts a lower price without recalibrating what the product is worth next year.
This is the closest bundling comes to the "perceived value" argument actually holding, because the discount has a stated reason that expires.
Gift sets
Repackaging changes what is being sold. A product that is unremarkable alone can work as part of a considered gift, and gift buyers are less price-anchored because they are not buying for themselves.
The constraint is that it needs real presentation — packaging, a photograph of the set, a description written for the gift buyer. A gift bundle that is three products in a normal box with a badge is just a discount with extra steps.
Before you build any of them
Work out why it is slow. Bundling fixes discovery and it fixes price resistance at the margin. It does not fix a product that is genuinely wrong — bad fit, poor reviews, a size nobody wants. Bundling those attaches your problem to a product that was selling.
Check the holding cost is real. The case for bundling rests on dead stock costing you money: warehouse space, capital, eventual write-off. For a small, shelf-stable item in a corner of your own unit, that cost is close to zero and waiting may genuinely be cheaper than discounting.
Look at the units, not the value. Twelve units of a £30 product is a slow line. Four hundred units is an inventory problem, and it needs a different instrument — a properly discounted clearance, or a wholesale conversation.
The alternative you should price first
Before building any bundle, work out what the straightforward options cost, because sometimes one of them wins.
A quantity break on the slow product alone. If the item is a consumable and the problem is that nobody buys more than one, three-for-15% may clear the stock without involving your bestseller at all, and it leaves no cannibalisation to measure.
Moving it in the merchandising. A product buried on page three of a collection is not slow, it is hidden. Promoting it for a fortnight costs nothing and tells you whether you have a demand problem or a visibility one. Doing this first means a later bundle is aimed at the right problem.
Writing it off. For a small quantity of something genuinely unwanted, the cheapest outcome is often to donate or liquidate it and stop spending attention on it. Attaching it to a working product costs you discount on that product and postpones the decision by a quarter.
Running it without training your customers
The risk in any recurring clearance mechanism is that regulars learn to wait. Three habits prevent it: give every clearance bundle a stated end date and honour it, never put the same product into a clearance bundle twice, and keep the discount attached to a reason — season, discovery, gifting — rather than to the product itself.
Make the offer automatic rather than code-based, too. A code-based clearance bundle blocks the customer's own code at checkout, which turns a stock-clearing offer into a choice between your discount and theirs.
What to measure
- Units of the slow product cleared per week, against the rate before.
- Standalone sales of the fast product over the same period — this is the cannibalisation check and the one most people skip.
- Margin per order, not revenue and not AOV.
- Whether the discovery bundle produced repeat buyers of the slow product. If it did, the product was undiscovered. If it did not, it was simply unwanted, and the next lot should be cleared rather than bundled.
Where to go next
Bundle builds curated and mix-and-match offers as automatic discounts, with scheduling so a clearance bundle ends on its own. For the underlying margin arithmetic, see bundles and average order value; for the build-your-own structure that suits discovery bundles, mix-and-match bundles.