Current Issue
This Month's Print Issue

Follow Fast Company

We’ll come to you.

Behind The Scenes At Maple, The David Chang Backed, Delivery-Only Restaurant Of The Future

Inside Maple's $26 million quest to reinvent delivery—healthy, affordable, chef-worthy meals, fast—in the land of sad desk lunches.

[Photos: Courtesy of Maple]

"If I got this for delivery, I’d be like, 'Holy fuck!'" says Momofuku founder David Chang, lifting up a forkful of arctic char that has just arrived at the business office of Maple, a delivery-only restaurant in which he is an investor. "It’s the perfect temp. This is very hard. You just don’t see even a really good restaurant cook fish like this."

It's three weeks prior to Maple's April 28 launch, and Chang is scooping up forkfuls of food from the first test delivery run. The service, at first, sounds like any of the dozen or so apps that already promise restaurant delivery in New York City or San Francisco: Press a button, food shows up. But while Seamless simply conveys your order to a restaurant, and startups like Postmates and DoorDash dispatch delivery people to pick it up, Maple controls the entire process. It cooked the food (three $12 options for lunch, three $15 options for dinner, including tax, tip, and delivery), built the app you use to place your order, and hired the delivery guy who brings it to you. It is, if you will, a "full-stack" restaurant. Or, as Chang puts it, "the future."

Caleb Merkl

The spread looks tasty. There are vegetarian green chili enchiladas, made with tortillas from a tortillaria in Queens and house-made enchilada sauce; slow-cooked baked arctic char with an olive relish; and pepper-crusted pork tenderloin with sautéed Tuscan kale. "Hopefully this doesn’t taste gross," CEO Caleb Merkl worried as he pulled the boxes form delivery bags. "It’s the first day." Maple’s executive chef, Soa Davies, in addition to writing two cookbooks and starting her own business, spent years as the head of menu research and development at three-Michelin-starred French seafood restaurant Le Bernardin. The food definitely does not taste gross.

But it’s not, Merkl decides, quite good enough yet. "It’s probably a little bit colder than it should be," he says.

Down With Sub-Par Takeout

Delivered meals, according to the Restaurant Marketing and Delivery Association (RMDA), account for a 7% share of American food spending, with 26% of consumers ordering delivery or takeout at least once a week. The Internet, mobile phones, and an "on demand" expectation have helped feed the growing category. Seamless started taking online orders for delivery back in 1999, and its competitor GrubHub launched in 2004. The two companies merged in 2013 and together field about 200,000 orders in 800 cities every day. Delivery startups like Postmates, Wunwun, and DoorDash have further transformed the experience from a process that once involved a phone number and a paper menu to just a few taps of a smartphone. Sprig and SpoonRocket in San Francisco operate similarly to Maple.

But as much as ordering food for delivery has evolved over the last couple of decades, the restaurant itself has not, which might be one reason that many people—especially those outside dense urban areas—still associate delivery mostly with pizza and Chinese food. "When you order in the hole-in-the-wall Chinese restaurant, it's the same experience sitting on the couch in your apartment or eating at the restaurant," Chang says. "But would you order fish at any restaurant? It’s usually not done well. And you’re ordering from a place that is making it to be served on a plate."

By only cooking for delivery, Maple can develop recipes specifically to withstand a Manhattan bike commute—and thus, significantly enhance the potential deliciousness of the meal that arrives on your doorstep or stoop. Fried food, in Maple’s early tests, failed to arrive crispy. Loose sauces migrated; tacos arrived looking more like burrito bowls. So it adjusted its offerings accordingly. Maple’s focus is on getting the food as quickly as possible from oven to your door, because delivery is not a side business, it's the only business. Maple also avoids paying for expensive real estate, instead housing its inventory and most of its kitchen space in an old Pfizer factory in Williamsburg. It can serve more people with fewer staff, and the number of dishes it sells is not limited by a line that spills onto the sidewalk.

The BBQ Berkshire pork sandwich, one of Maple's dishes

The tall order it has set for itself, in a city brimming with sub-par takeout and soggy noodles, is healthy, affordable, food that tastes good, delivered within 30 minutes. This is—as Fast Company discovered over the course of several weeks shadowing the company before its launch—no small undertaking.

Starting in July, Maple tested recipes, built apps, and hired a team that grew from 50 to about 70 in the weeks before launch, including programmers, branding experts, delivery guys, and kitchen staff. It built a 5,000-square-foot industrial kitchen in Brooklyn, complete with rapid cook technology ovens and blast coolers, where a chef de cuisine, two sous chefs, eight line cooks, and a porter will work to prepare all of the food. And it opened a distribution kitchen in the Financial District, where a team of about 30 will cook, assemble, and dispatch meals. For the three weeks prior to launch, it ran the entire operation, with real food, at full-scale, twice every day. A computer program placed fake orders for customers and requested delivery to randomly selected addresses from 30 nearby buildings in under a half an hour. "Anything longer that, and you’re like, Where is my food?" explains Maple’s cofounder, Akshay Navle.

Akshay Navle

Navle, Maple’s COO, scrutinized this entire production for cracks in the system—any one of which can cause the familiar annoyances of delivery (cold food, long waits, homely dishes). And there were plenty.

The problem that led to that lukewarm fish, for instance: Davies, who handled everything from writing and testing the 270 recipes Maple developed before launch, to stylizing the food for photographs, says the equipment in the new kitchen was, as expected, working differently than the equipment Maple used before the live tests, creating about an 8-degree difference from the previous test kitchen. A measly 8 degrees is enough to throw everything off.

"His expression is the exact same expression that is on the face of any GM [general manager] that is about to open a restaurant," Chang says, pointing to Navle. "It says, Oh my god, this is happening."

Tilting At Chipotle

Eleven days before launch, Navle is leaning on the table, looking at his laptop's on-screen agenda with the intensity of someone watching a tennis match. Maple’s management team has gathered early for a meeting in the Williamsburg kitchen where Maple prepares food before driving it to the smaller distribution kitchen. Its members grip coffee purchased from three different vendors as Navle begins running through a number of issues he’s discovered during the first week of tests.

Navle seems to always have a list like this one running in his head. He knows everybody’s name, everybody’s job, and everything that needs to be fixed.

Would deliveries be faster if Maple hired someone to "butler" bikes when delivery guys are running inside to collect orders, or if it asked them to lock the bikes up? Can there be a break to flip the kitchen between lunch and dinner? The sheets with instructions for how to pack food (13 steps for a bacon lettuce avocado sandwich, 10 steps for tacos) are "killing it," but they really need to be "brain dead" simple. Delivery backpacks, ordered from Beirut, just got released from customs. There are four printers, but none of them work. "What about the tongs?" Navle asks the team. When they used them to box broccoli rabe, portion sizes were all over the place.

Merkl leans back in his chair, fiddling with a purple rubber band, and, as usual, wearing bright sneakers ("Caleb dresses in 95 layers of Nike," Navle once noted, while wearing a matching pair of sneakers that Merkl gave him as a gift). Merkl’s concerns seem to be more on the scale of a city plan than a laundry list. He has the charisma of a high school class president and a fondness for comparing Maple to Chipotle. (Sprig’s CEO and a representative from Munchery both made similar comments, unprompted, when I called to ask about their businesses.) It’s a very flattering comparison, and one that the entire genre seems to be hanging its hopes on.

As sales at fast food restaurants like McDonald's decline, Chipotle has led the trend toward better quality convenience restaurants. The chain’s sales grew 27.8% last year, with same-store sales increasing 16.8%. But Chipotle has almost 50 locations in Manhattan alone, and 1,783 stores worldwide—an expense that ate up about 5.6% of Chipotle’s revenue last year. "A company like Chipotle took years to build into a [billion-dollar business]," Merkl says. "Relative to that we’ll scale more quickly. Relative to another startup that is launching in 10 cities in one month, it’s going to be slow."

Soa Davies

For now, Maple is only comparable to Chipotle in the optimistic vision of its CEO. It is launching with a minuscule footprint, delivering within a five-minute bike ride of its distribution kitchen in the Financial District. But it is building around the vision of serving 100,000 meals per day in New York. To put that into perspective, Le Bernardin did about 400 meals on its very busiest day, according to Davies.

The kitchen that surrounds us is more commensurate with Maple’s potential than its launch reality: walk-in coolers, industrial chillers, fridge-sized programmable ovens with rolling racks. It’s all part of the plan. Sprig maintains its short delivery times by charging more during busy times, like Uber’s surge pricing. Munchery asks customers to finish cooking the meals themselves, which means it doesn’t have to keep them warm and can deliver more during one route. Maple’s strategy is to set up small local hubs like the one in Financial District in every neighborhood. But all the food will be prepared here, in this kitchen.

The question is whether good food can scale the way investors in startups like Maple, Sprig, SpoonRocket, and Munchery hope. Venture capital has taken interest in food before: think Starbucks, Blue Bottle, and Melt. But, says Semil Shah, who is an investor in Instacart and DoorDash, "The marginal cost of selling an extra piece of software is zero. The marginal cost of opening in a new city is a lot." When I spoke to GrubHub Seamless CEO Matt Maloney recently, he was (unsurprisingly) skeptical about the ability of "full stack" restaurants to scale like tech companies. "You’ll see some of the better ideas become great local businesses," he said in November. "But I think it will be challenging for most of these guys to operate at the scale we are now."

In some sense, there is no need for Maple to rush to scale. "Everybody eats lunch," as Chang puts it. "Whether it’s the CEO or the assistant to the CEO, most people would rather eat well and affordably." It’s a big market, and just like there is room for both Panera Bread and Chipotle in the same city, there will be room for more than one app-restaurant.

Even so, this early in the game, none of the "full-stack" restaurants that have set out with dominance in mind have achieved it on more than a regional scale. SpoonRocket has $13.5 million in funding now, but for its first three months after it started in June 2013, it was just the two cofounders, cooking and delivering food themselves. Now it’s in San Francisco, San Diego, and Seattle. Sprig, launched in November 2013 in San Francisco, has raised $57 million and served around 500,000 meals. It’s available in San Francisco and Palo Alto, with plans to open in Chicago soon. Munchery, which has raised $39.9 million since opening in April 2011, has opened in the Bay Area, Seattle, and New York, with plans to launch in Los Angeles and Brooklyn soon.

Maple has already raised $26 million from investors including Thrive Capital, Primary Venture Partners, and Greenoaks Capital (along with Chang himself), which allows it to launch at full speed. "Can this be a big business?" Navle remembers an early investor asking. "Then why beat the local Chinese restaurant? Why not beat Chipotle?"

Branding And Celebrity Sauce

Maple’s business office is technically an apartment, one that has been zoned for both live and work. There is still one room shoved full of the previous tenants' bikes and furniture. But aside from this, the only remnants of the previous tenants’ occupation are two bookshelves that flank the entrance. "We will eventually pull down the books and put Maple history in there," says Merkl. "Once we have Maple history."

With the number of well-funded food delivery startups already in the mix, Maple's strategy is to compete long-term on taste, brand, and story. Merkl’s focus on the brand comes through in the photos he stores on his camera roll: pictures of early photo shoots in Navle’s apartment, behind the scenes of a promotional video, and the many paint chips that the company collected when it was deciding on its logo color.

About those paint chips: Maple wanted the brand to be rooted more in the idea of culinary excellence than agriculture. Its brand team didn’t want to make everything green just because the food was fresh. They didn’t want to photograph everything in front of a barn just because they plan to source local foods. So could they still convey all of that while building a less stereotypical brand?

"The answer was yellow," Merkl says. "It conveys a tastiness, a comfort, and also isn’t used by many people. If this business is successful, it’s important to be distinct."

Getting to a specific shade of yellow, though, was a bit of a process.

"Have you read The Odyssey?" Greg Hathaway, Maple’s director of brand, asks. "Have you read it twice?" chimes in Zach Sniderman, Maple’s director of content strategy (disclosure: once upon a time, Sniderman and I worked together).

Other aspects went through similar levels of agony. What size should the packaging be? Should it have paper handles or fabric handles? What was the tone of the emails to be sent out? At one point, the brand team went through what it describes as 100 bell peppers shooting an accent video that will run within the app. When they finally nailed it, a chef informed them that they had used an improper cutting technique. So they shot it again.

Regardless of the brand or the logistics, like any restaurant, Maple won’t work without good food. And that’s where Chang, a superstar chef with a cult following, comes in. Maple’s main investor, Thrive Capital, introduced the team to Chang, who had already been thinking about food outside of the restaurant as the next frontier when he launched a boxed lunch program with coworking space WeWork, because, frankly, "real estate is just stupid right now." Soon after, he was writing them a check (two checks, technically—both he and Momofuku are investors) and introducing them to his contacts in the restaurant industry, including Davies.

Maple’s food will be nothing like the pork-laden menu of Momofuku. It’s focusing on simple, healthy meals that you could eat every day (and it hopes you will) without shaving years off of your life. The consumer app includes photos of every ingredient in every meal, chefs only uses antibiotic-free and hormone-free meat, and they use locally and sustainably sourced ingredients when possible. Chang, by contrast, once held it a point of pride that there was only one vegetarian dish at Momofuku Noodle Bar. But even so, it’s hard to overstate Chang’s influence. The New Yorker called him the "one of the most celebrated chefs in the country" in 2008, even before he had expanded his restaurants beyond New York, opened a line of bakeries, or launched a Momofuku product line. Sprig and Munchery have similar promises to Maple, but they do not have Chang’s recognizable name and reputation for good food.

Maple doesn't lean too heavily on Chang in its marketing—neither his name nor face are on the homepage—but it certainly doesn’t hurt to have him on board. "If we would have been like, Caleb and Akshay are launching a food company," Merkl says, "nobody would have talked to us."

Apps To Feed Them All

The distribution kitchen is a nondescript, cramped, chaotic hole-in-the wall in the Financial District. Delivery guys wearing yellow vests filter in and out of its unmarked door like bees returning to a beehive.

Maple’s CTO, Dan Cowgill, originally built five apps to get food from Maple to the customer. When an order comes in, it is added to a cooks' app displayed on tablets in the kitchen. Expediters take the food from the chefs, put lids on it, put it in a branded cardboard sleeve, and used a second app to check that the order was done. Packers refer to a third app that tells them, based on destination, which orders should be sent out in a bundle. They enter the bag's ID and sets it down for a delivery person to pick up. When the delivery person grabs it, he enters the bag ID into a fourth app, and receives instructions for the optimal route to take when dropping off the orders. A fifth app, which the manager of the distribution kitchen uses, gives the overall view of all of this, to spot the gaps in the process and adjust. All of these apps talk to each other, and all of this action fits inside a space that feels like the size of a city bus.

Four days prior to launch, the system and apps have been tweaked and streamlined, and the kitchen has delivered 800 meals over two hours—a new record. "Let’s just say it’s been a long day," Navle says, patting the back of the kitchen manager as he passes.

Merkl and Navle have agreed to let me follow a delivery guy on a route. His name is Ricky DeBushea, and he was the first hire on the delivery team, though the deliveries he’s doing today are for my benefit (he’s since been promoted to the kitchen). We walk through the angled streets of Tribeca, past people speaking in different languages, TV crews, and all the bustle of busy people out for lunch. We drop food at the Maple office, one of the only locations where it actually gets delivery during tests. When DeBushea takes out his tiny, company-supplied Android phone and opens the delivery app, we can see the delivery took nine minutes. He swipes "order completed." Then we’re on to 99 John Street, where he takes a selfie within 19 minutes of the order (selfies are how the delivery teams "complete" a fake order during the tests). The last bag he takes to 2 Gold Street. He "drops it" at 22 minutes. On a bike (and without me tagging along, though he doesn’t say this) he says it would have only taken him nine minutes.

Today, the average delivery was 27 minutes, the median was 23 minutes, and the standard deviation was eight minutes. And that was with a bug in the delivery app that slowed the process down. So far the fastest that the kitchen has run is 500 meals per hour, which represents an unlikely number of customers for a startup on its first day. It's four days to launch, and Maple is ready.

Delivery, Day One

When I open Maple’s ordering app on the first day of delivery, I’m greeted by a video montage: Soa huddled with a group of chefs (or at least, people wearing chef attire), bright green asparagus apparently diving into boiling water, and a sizzling prong.

By 11:51, I’ve decided on Spicy Shrimp Stew.

Sixteen minutes later, I get a call to let me know it has arrived.

Maple's roasted beet and avocado salad

There’s a small crowd of deliverymen waiting in the lobby: one from Chop't, another from some unknown, smiley-faced bag establishment. Uber announced earlier that day that it would begin delivering food from restaurants like Num Pang, Sweetgreen, and Scarpetta. Its competitor, Gett, announced a similar service. I spot my Maple delivery person by his recognizable yellow vest.

"Thanks for ordering," he tells me, on script. "Cheers, have a good one." He hands me a brown bag with cloth handles, like the kind you get at a fancy department store.

Inside are two plant-fiber boxes, tastefully wrapped in blind-debossed branded sleeves decorated with the perfect shade of yellow (106U at a 70% tint, FEE7EE if we’re talking Hex code).

It’s delicious. And still piping hot.