Running a gel pack returns programme that customers actually use

Two schemes, same product, same city. The first asks customers to scan a QR code, print a label, box the packs and drop them at a post office; it recovers 8% of packs. The second puts a pre-paid, flat-packed envelope in the delivery and collects from the doorstep on the next order; it recovers 61%. The difference is not the incentive. Neither scheme paid anything. The difference is that one required a printer.

Returns programmes are usually designed by people who are good at logistics and who have never tried to return anything while holding a toddler. The design question is not “how do we incentivise this”, it is “how few decisions does the customer have to make”.

Friction beats incentive, every time

Every step in a return is a place where a customer quietly stops. Find a box. Find the packs, which by now are at the back of a cupboard. Print something. Leave the house. Each step costs you between eight and fifteen points of return rate in our experience, and the print step alone is worth more than any deposit you can afford to pay.

The version that works has two steps: put the packs back in the bag they arrived in, hand the bag to the driver. That is it. The bag has to be designed for it, with a flat base so it stands up in a cupboard, a zip rather than a fold, and a printed return date that is a real date rather than a window.

The collection moment is the whole programme

Attach collection to the next delivery, not to a separate trip. The driver is already at the door. The marginal cost of taking a bag back is a few seconds and a little space in the van, and the marginal cost of a customer making a special journey is the entire programme. Where a customer has no next order within the return window, offer a collection-only slot at a low fee, and accept that most will not use it.

Then close the loop visibly. A short message the day the packs are received, with the count and what happens to them, raises the return rate on the following order by a surprising margin, because it converts an abstract obligation into a completed transaction. Customers who never hear back assume the packs went in a bin and stop bothering.

What comes back is not what went out

Plan for a return stream that is 15 to 20% unusable. Packs come back with torn film, with gel that has been frozen in a domestic freezer set to minus 24, with food residue on the outside, and occasionally with something that is not a gel pack at all. Sorting has to happen at a single point with a clear reject criterion, or the wash step becomes the bottleneck.

The reject criterion should be physical and fast: film intact, seam intact, no deformation beyond a set limit, label legible, pack weighs within tolerance of its dry weight. A pack that has lost 6% of its mass has lost water and will underperform; it goes out of the loop regardless of how good it looks. Weighing is faster than inspecting and catches more.

The economics, honestly

A returnable pack needs to come back at least four times to beat a single-use pack on cost, and six times to beat it comfortably once you have paid for collection, washing, re-freezing and the 15 to 20% attrition. That means the programme only works at scale on routes with dense delivery, where a driver passes the same door every week. On a rural route with fortnightly deliveries, the maths rarely closes and a single-use pack is the better answer.

Run the numbers per route before you launch, not per company. Most failed programmes we have seen were launched nationally because they worked in one city, and the losses came entirely from the routes where collection frequency was low.

If you are designing the pack that has to survive six or eight trips, the film and seam requirements matter more than the gel, and our SAP powder grades for reusable packs are specified by cycle count for exactly this reason.

Start with one dense route, one bag design and one collection moment. Measure the return rate for eight weeks before you write a national policy, and be prepared to find that the answer varies by postcode more than by product.