Skip to Content

Unicorn Instacart Hopes Its Data Scientists Can Calculate a Path to Profits

Same-day grocery delivery is a notoriously difficult business, but Instacart believes it’s found a way to make money.
June 30, 2016

Why can’t grocery shopping be as easy and convenient as summoning a car on Uber?

A well-funded startup called Instacart is trying to prove that it can. In 24 metro areas in the U.S. you can tap an order into the company’s mobile app and have it delivered in as little as an hour. Instacart workers - some employees and some contractors - pull items from the shelves of stores including Whole Foods and Target and deliver them using their own vehicles. The startup says that not owning warehouses or vans like competitors such as Amazon and Safeway gives it an opportunity to make the grocery business more customer friendly, just as Uber gains advantages from not owning the cars it dispatches.

Instacart, which counts Whole Foods as a partner and investor, received $275 million in funding and was valued at a reported $2 billion last year, enough to earn the “unicorn” moniker applied in Silicon Valley to private companies worth over a billion dollars. Yet it is not profitable. The company recently increased delivery fees and reduced its number of professional shoppers, along with their pay—drawing comparisons to Webvan, a grocery delivery company that went public and then bankrupt during the dot-com bubble. Delivering anything within hours has punishing economics. Last year eBay gave up on same-day delivery service, and Google shuttered two Bay Area facilities set up for a competing service of its own.

Instacart delivers groceries in as little as an hour using workers who pull items from the shelves of regular supermarkets, guided by an app.

Jeremy Stanley, who joined Instacart as vice president of data science nine months ago, claims his team of data crunchers is disproving received wisdom about same-day delivery. By aiming new algorithms at its workers and customers, the company is slashing costs and increasing revenue, he says. And it’s building up a new business by charging consumer goods companies to target promotions at Instacart customers.

Profits are still lacking. But Stanley says his team’s efforts have helped make Instacart “gross margin positive,” meaning that on average the cost of fulfilling an order is less than the revenue it generates—after corporate costs such as marketing are excluded.

Stanley says Instacart’s average gross margin per order is measured in dollars, not cents, and that the company's average gross margin is in the red in less than 15 percent of the company’s markets.

One focus of Stanley’s team has been to make Instacart workers more efficient by improving the app that assigns them orders and guides them as they pull items from shelves and deliver them. It now takes 40 percent less time to fulfill the average order than it did nine months ago.

“Every time we reduce that number of minutes, we completely change the economics of our business,” says Stanley. He declined to disclose just how long an average order now takes, but says most customers ask for delivery in one to two hours.

One way the company has increased speed has been to improve the algorithms that assign workers to orders, says Stanley. The decision has to take into account where the order needs to be delivered, which items were ordered, how much stock is on hand at nearby supermarkets, and which Instacart workers are available.

Some of the largest efficiency gains have come from improving the delivery step, says Stanley. As orders come in, they are bundled into batches that will travel with the same driver in a way designed to minimize driving time and distance.

Instacart is also working to cut the time it takes for workers to pull items from shelves. One effort involves constructing rough maps to show where items are in stores, using motion-sensor data gathered from the phone app that Instacart shoppers use to scan items as they collect them.

Of course, profit also depends on growing revenues, something Instacart aims to do in part by building up a big business selling ads. Instacart's app already offers people promotions from brands such as Häagen-Dazs as they shop. Stanley says companies want to pay for those messages because they are proven to convince people to buy more, and that his team is building out more sophisticated ways to target ads and deals and track their effects.

John Deighton, a professor at Harvard Business School, says this opportunity could be crucial to the startup’s prospects. People have so much choice when it comes to items such as toothpaste that consumer goods companies must spend big on marketing. That can account for 30 percent of the cost of a supermarket basket, says Deighton: “It’s a very wasteful business.”

Instacart might be able to help cut that waste, he says. An app can target ads and coupons more personally than TV, billboards, or mail and convert them into a sale at the tap of a fingertip. Companies can know exactly who responds. Although that model is proven for many types of businesses, it hasn’t really shaken up consumer goods yet, says Deighton.

Instacart will need to move a lot more groceries to be the one to do that, though. Deighton suggests that the startup will need to build up a very large customer base if it is to become really profitable and make itself essential to consumer brands. That means Stanley and his data crunchers will have to control logistical costs well enough to buy time for that growth. The startup will also have to fend off Amazon and Google, which have large existing ad businesses and are also pushing grocery deliveries.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

The problem with plug-in hybrids? Their drivers.

Plug-in hybrids are often sold as a transition to EVs, but new data from Europe shows we’re still underestimating the emissions they produce.

Google DeepMind’s new generative model makes Super Mario–like games from scratch

Genie learns how to control games by watching hours and hours of video. It could help train next-gen robots too.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.