Bundles that earn their place: attach rate, margin, and the discount you did not need

Measure attach before you price the bundle

Attach rate is the share of main product orders that include the accessory, measured before the bundle, during it, and with the discount removed if you dare. That last test is the one that matters and the one nobody runs, because a positive result means admitting the discount was waste. Run it anyway, on one bundle, for a fortnight, with the bundle placement kept and the discount set to zero. The result tells you whether you are selling convenience or renting it.

Watch margin per order, not revenue per order. A bundle can raise order value while lowering contribution, once the discount, the extra pick and pack time, and the higher return complexity are counted. Returns are the quiet cost: a bundle returned in part creates a refund calculation your support team will meet weekly, and a policy nobody wrote down.

Build bundles from use, not from slow stock

The bundle that works completes a job: the device and the case and the cable the device needs to work at all, the kit that saves a second order and a second delivery wait. The bundle that fails is a warehouse clearance wearing a merchandising hat, a strong product lashed to a weak one shoppers did not ask for. They can tell. Attach rate tells you they can tell, within a week.

Keep the components buyable alone at their normal prices, and show the arithmetic plainly on the bundle page: each price, the bundle price, the saving if there is one. Hiding the comparison inflates short term clicks and teaches shoppers to distrust every bundle you show after.

Placement beats discount size

A bundle offered on the product page, at the moment the shopper is picturing using the product, outperforms the same bundle pushed in a cart popup at a bigger discount. Timing is information: on the product page the bundle answers a question the shopper is already asking. In the cart it interrupts a decision they already made. We have watched a zero discount product page bundle outsell a discounted cart offer on the same store, which should end the argument that the percentage does the work.

One bundle per product, chosen well, beats a carousel of combinations. Choice overload is real at the exact moment you want a simple yes. If two bundles genuinely serve two uses, name them by the use, not by tier names that mean nothing outside your office.

Review bundles like products: attach rate, margin per order, return rate against the main product alone. A bundle with a return rate well above its components is telling you the pairing confuses fit, compatibility, or expectation, and no discount fixes a misunderstanding. Kill it or rebuild it. Keeping a bad bundle because building it took effort is how a merchandising page becomes a museum of old ideas.