Contactless Ordering

No-Contact Delivery shipped quickly and did exactly what it was built to do. Then driver tips fell. The doorstep exchange had been the moment tipping happened, and removing the contact removed the moment, a consequence nobody had modeled, because the feature had been reviewed for customer and store safety and no one had asked what it did to driver earnings.

Reframe it from a missing button to a missing moment. The question was not “where do we put tipping in the app,” it was “when does tipping now happen, and what has to exist at that point.” That put the answer after delivery rather than at checkout.

Saying out loud, days after a well-received launch, that the launch had created a problem. In a crisis everyone is being asked to move; volunteering that your own shipped work needs follow-on work is not the popular contribution.

Drivers were working more than they ever had, earning less per run, and were not users of the ordering app. Nobody’s roadmap owned their income. Every metric the team reported on measured the customer.

Give drivers standing in the design as a stakeholder in their own right. Post-delivery prompt at the moment the order lands, preset amounts to remove arithmetic, a custom option for control, and stored payment so completing it is one action rather than a checkout.

We added an optional thank-you message. The doorstep exchange had never only been a transaction: it was also the only moment a driver was acknowledged by name. Restoring the money without restoring the acknowledgment would have solved the smaller half.

Curbside was being rolled out across quick service at speed, and most of it was bad. Customers could not tell where to park, had no dependable way to signal they had arrived, and were handed processes more complicated than the problem they solved.

Catalog the failure modes first and design against them by name. Under time pressure the instinct is to copy whatever competitors shipped; what we copied instead was the list of ways it went wrong, and treated each one as a requirement.

Days, during a period when days were the currency. It is a genuinely hard argument to make when a working competitor screenshot is sitting on the table.

Curbside is not an app feature. It is an app, a parking lot, a store workflow and a staff member walking out with a pizza. The part design controlled was the smallest part of it, and the failure modes we had cataloged nearly all lived outside the screen.

Design across the whole handoff. Car make and model captured at checkout so staff could identify a customer without asking. Designated spots mapped in the confirmation. Arrival signalled by push or phone, with one obvious path rather than several. Real-time order status so nobody sat in a parking space guessing.

Buy-in from store operations, who do not report to design and whose actual shift the change altered. The design was only as good as its adoption behind the counter, which made this a negotiation rather than a specification.