Last-mile e-grocery: gel packs at the doorstep hand-off

The photograph that arrives in customer service shows a box on a concrete step, lid open, one gel pack slumped against a tub of ice cream that has clearly given up. The delivery note says the box left the depot at 14:12 and was photographed at 15:04. Fifty-two minutes. The customer was home the whole time; she simply did not hear the bell, and the driver followed the no-answer protocol, which is to leave the box in the shade of the porch. There is no shade on that porch after two o’clock.

Last-mile e-grocery is the only part of the cold chain where the final twenty minutes are controlled by someone who has never read the specification. That is the design constraint, and it should shape the pack, not just the courier script.

The doorstep is a worse environment than the van

A refrigerated van holds 4 °C and cycles its door. A doorstep at 15:00 in August holds 34 °C in the air and closer to 45 °C on a dark stone step in direct sun. The box is not in that environment for the whole journey; it is in it for the eight to fifty minutes between the knock and the customer reaching the door, plus however long the box sat before the knock.

Most e-grocery pack specs are built for transit and treat the doorstep as a rounding error. Then the summer complaint rate triples and everyone blames the gel. The gel did what it was told. It was sized for a two-hour van leg, not for a fifty-minute step in the sun.

Sizing for the doorstep, not the depot

Add thirty minutes to your worst realistic doorstep dwell and design for that, not for the average. In practice this means a pack weight increase of 15 to 25% for ambient-sensitive lines, or a switch to a small insulated bag inside the carton, which is usually cheaper than more gel because it also blocks radiant heat from the lid.

The second change is placement. A pack sitting on top of the payload cools the air above it. A pack pressed against the lid fights the lid, which is the hottest surface in the box after twenty minutes in sun. Put one pack below the payload and one against the side that faces the sun at the delivery hour, and label the carton so the driver can tell which side that is without opening it.

The hand-off script is part of the pack

Three script changes cost nothing and beat most pack trials. Knock, then wait a full sixty seconds before the second attempt, because a customer upstairs needs that long. If there is no answer, do not leave the box on an exposed step; take it back and re-deliver in the evening slot, and price the second attempt into the delivery fee. And where the customer is home, hand the box over rather than leaving it, because a box that goes straight into a kitchen is a box that never sees the worst half hour.

There is a counter-intuitive finding here that surprises most operators: the customers who complain loudest about melted product are not the ones with the longest routes. They are the ones on routes where the driver consistently arrives at the same time as the sun on the front of the house, which is a scheduling problem disguised as a packaging problem. Move that customer to the morning slot and the complaint disappears without a single extra gram of gel.

What to instrument

Put a single low-cost logger in one box per route per week, taped to the payload face, and record the time from door-open to door-close from the driver app. Two numbers per delivery is enough: maximum payload temperature and doorstep dwell. After four weeks you will know which routes need a bigger pack, which need a schedule change, and which need a conversation with the customer about where to leave the box.

Then re-cut the specification by route, not by product category. A route with six-minute doorstep dwells does not need the pack that a rural route with forty-minute dwells needs, and paying for the bigger pack on every order is how e-grocery margins disappear.

If you are rebuilding a summer last-mile spec, the placement diagrams in our logistics ice pack user guide cover the two-pack doorstep layout, and the log sheet on the last page is the one we use for route-level trials.

Take your five worst summer routes, add one logger and one dwell timer each, and run them for a fortnight before you change a pack size. The data will point at scheduling more often than at gel, and that is a cheaper fix.