Mastering Coffee Delivery, One Tray at a Time

Starting last week, customers using the Starbucks mobile app in Seattle saw a new option at checkout: delivery by Postmates, available in select neighborhoods. We are excited to finally launch our collaboration with the largest coffee company in the world.

Postmates now fulfills nearly a million deliveries a month, and 5 percent of them are coffee beverages. Getting you your favorite espresso drink, right when you want it, still hot and in perfect shape, is a responsibility we don’t take lightly. We worked closely with Starbucks to hone the process, and along the way, we found ourselves somewhere we didn’t expect.

The goal: design, produce, and assemble 200 trays that will stabilize coffee in transport.

The timeline: 3 weeks.

Let me set the record straight: Postmates is a logistics company, not a manufacturing company. In true startup fashion, people from various parts of the company pitched in to make it happen. I, for one, am on the business development team, but because of this project, I had the chance to return to my mechanical engineering roots.

This is the story of how we engineered the Postmates coffee tray.

The final design: a tray surface with holes for 4 cups, supported by plastic “legs”.

Research and Inspiration

The first step was to see if any existing designs fit the bill. We hit the streets, delivering coffee using standard trays.

Left: foam tray, included with insulated bag | Center: cardboard carrier | Right: “egg carton” tray

When all failed, we asked our friends over at Starbucks what they’d recommend. They sent us the tray they used to deliver coffee within the Empire State Building, and we loved it.

Starbuck’s coffee delivery tray.

The tray held any size cup, Tall to Venti, snuggly in place. It turns out that all Starbucks cups have the same 3 inch diameter, 3.4 inches from the bottom. The trick was to match those dimensions.

Getting Started

While Starbucks’s tray was the right design idea, it wasn’t exactly what we needed. They were too small for our specialized insulated bags and would not survive regular daily use.

Bastian Lehmann, our CEO, decided it would be best for us to make our own larger tray, and that’s where I came in. Because of my mechanical engineering degree, this project was right up my alley.

I was brought up to speed on the task at hand. Since the Starbucks design used flat pieces, I immediately knew it was perfect for laser cutting and reached out to some local shops.

Meanwhile, I worked on the new measurements and sketched up the dimensions we needed for our tray — it felt great to be holding calipers again.

Since laser cutting shops can’t work off of hand drawn sketches, and I no longer had the right software, I roped in one of our graphic designers to help. SatAmrit Khalsa usually spends her time making the Postmates app beautiful, but used her ninja Adobe Illustrator skills to draw up DXFs of the design. Before sending it off, I printed and assembled a full-scale paper mockup. Building with my hands again was a welcome break from emails and spreadsheets, but I definitely missed control+Z.

Full-scale paper mockup — built from printer paper, staples, and FedEx envelopes.

The design was now ready to graduate from paper to something more durable. It was time to get cutting.

First Prototype

I was so excited when the box of parts arrived that I ripped it open like a kid on Christmas morning. That joy was short-lived: I quickly became frustrated by how hard it was to assemble. The slick pieces slid across each other and it was hard to center the tray “legs”. After getting everything in place, I had to hold the parts still until the glue set.

To make it easier, I decided to try a tab-and-slot design on the next prototype. The tabs would help position the legs against each other and the tray surface. I was also hopeful that if we picked the slot width just right, the pieces would stay together without glue (a press fit).

Tab-and-slot design.

Assembly completed, we inserted the prototype into the insulated bag — the corners of the “legs” were too sharp and cut into the lining. Luckily, this was a straightforward fix: we rounded those corners for future versions.

Material Selection

From earlier tests, it was clear that cardboard wouldn’t hold up. What we needed was a durable, easy to clean, and high temperature withstanding plastic.

To make sure the plastic can handle holding hot coffee, we put a few samples to the test by exposing them to steam. I placed the samples over cups of boiling water and left them in the insulated bag for 30 minutes.

Four of the samples, after 30 minutes exposure to hot steam.

Only two emerged triumphant: acrylic and ABS. We chose acrylic because it is food safe, less expensive than ABS, and came in a black color and matte finish we liked.

Second Prototype

For the second prototype, the updated design was made of matte black acrylic, had rounder “leg” corners, and used tab-and-slot design.

It turns out that the tab-and-slot design was better in theory than in practice because acrylic is too brittle for a press fit. While trying to assemble the second prototype, we broke off some of the tabs and sent them flying across the room.

In the end we realized that the tabbed design didn’t make sense: we would still need glue for assembly, and they are more expensive to cut.

Tab-less design: shared cuts. Tabbed design: longer cutting line.

At this point, we were only a week from deadline. We needed to start the laser cutting process so there’d be enough time to ship the pieces to Seattle, where they’d be assembled and glued. It was time to hit pause on trying new ideas and start producing 200 of the best design we’d gotten to.

Production and Assembly

To make 200 trays, I needed 180 square feet of black acrylic, which comes in large sheets of stock sizes. I had a lot of fun figuring out the best way to arrange our parts to use every corner of the plastic — it felt like a game of Tetris.

Three of our laser cutting layout drawings.

Without tabs, the production design would have been just as slow and cumbersome to assemble as the first prototype. I wanted to make a jig (a tool that helps place things correctly for assembly) to speed things up. It would’ve been impossible to get the jigs done in time were it not for two strokes of luck: 1) one of my friends is a carpenter, and 2) our operations manager for the West Coast was wrapping up a visit to San Francisco and heading to Seattle. Like magic, my friend made the jigs overnight, and the next day they flew to Seattle.

Two-piece jig: one holds the legs at 90-degree angles from each other, and the other makes sure the legs don’t protrude beyond the tray surface.

I shipped the parts in batches (~50 each) as soon as they were cut. The first of these batches arrived on the same day as the jig and we had a team ready to assemble, glue, and insert the finished trays into the insulated bag.

It’s go time — let the coffee deliveries begin! Each order in Seattle will fly from your phone to the barista and then travel safely back to you in our insulated bag and custom-made coffee tray.

Up Ahead

We are proud of all the work we’ve done to make your Starbucks delivery experience awesome, and this is only the beginning. We see these 200 trays not as a finished product, but rather as a third “prototype” — every Starbucks delivery is another round of field testing. What we learn here will influence future designs.

We’ve already begun to explore improvements we know we’d like to implement, such as reducing the number of parts in the assembly, joining the pieces without glue, and fitting more coffee cups per tray. We’d love to hear your ideas on what else the next iteration should include!

Discuss on Hacker News.