Back to Blog

Keys To Building An On-Demand Startup

By Christopher Steiner  •  Apr 21, 2017

Christopher Steiner co-founded ZRankings and Aisle50, YC S2011, which was acquired by Groupon in early 2015.

In 2011 my cofounder and I had a second meeting with a partner at Sequoia Capital who was interested in our startup, which partnered with big brands like General Mills and Procter & Gamble to give consumers discounts for subscribing to products. He asked us if we had ever thought about delivering single orders of groceries. We said we hadn't, and that the economics of such a mission seemed impossible to us.

We never had another meeting with Sequoia. A year later, the venture capital firm funded Instacart, which has gone on to reach a $3.4 billion valuation while proving that the unit economics for delivering groceries can work, when the solution is creative enough and the software behind it good enough.

On-demand businesses like Instacart's aren't the kinds of ideas that come out of an MBA-class brainstorm. The sightline to profitability isn’t obvious or clear. Being successful in on-demand requires a fine understanding of a market and creative approaches to solving delivery distances, inventory, warehousing and smaller orders.

People were quick to dismiss national-scale grocery delivery a decade after the flameout of Webvan, which blew through $1 billion before shutting down in 2001. But Sequoia and Instacart's founders saw a different way to hack the problem.

The web has trained human psyches to expect everything to be available at nearly any moment. Amazon's Prime service, and companies such as Uber and GrubHub, plus startups piling into all kinds of on-demand markets have fed the trend. It only seems a matter of time until everything is available within an hour.

But building one of these companies is especially tricky. It requires founders to not only solve the normal demand and product riddles, but also become experts in logistics, transportation and, in some cases, the financial and physical demands of warehousing goods.

For startups who can crack all of these problems at once, growth can come quickly.

Here are some quick takeaways on what's most important when building an on-demand company - there are more details below:

  • Barriers to entry. The best on-demand businesses involve some factor that keeps thousands of startups from piling and fragmenting the market.
  • Unit economics matter. It's understood that early on, many on-demand startup will lose money per order. But founders need to be able to show that things will turn positive on a per-order basis at some point.
  • Find ways to batch and schedule deliveries and services.
  • If your potential market is big, know that Amazon and others, possibly Uber, will come after you. Plan ahead and know why your product can win.

Build a moat

The world of tech startups has always been a game of copycats on some level. If a startup can't fend them off and out-execute the best of them, then it will fail. It’s better, however, if a startup needn’t depend simply on better operations to win. The best on-demand businesses carry with them barriers that will keep other startups out of the space.

For a team with a few good engineers, building an Uber copycat is eminently possible. But it's not the application that keeps others from piling in after Uber and Lyft. It's the challenge of getting a critical mass of drivers and users. This is where so much of Uber's cash pile of $15 billion—raised with debt and equity—has gone. Uber gives drivers big incentives to sign up and stay on, as its service simply isn't useful without a horde of cars available to users at any given moment.

And that fleet of ready drivers is what's going to make Uber a player in so many on-demand markets in the future, whether it's delivering food, flowers or goods purchased through the web. Uber Eats is already producing 20% of the revenue of the biggest player in the space, GrubHub, and it's likely to grow.

Instacart's moat comes from a number of factors, one of them being its close relationships, sometimes exclusive, with retailers. Getting that kind of cooperation and co-branding from a multi-billion-dollar brick-and-mortar retailer takes years. At Aisle50, our sales cycles with retailers could stretch anywhere from six months toward two years.

Amazon, the other major player that any on-demand business must cope with, covets Instacart's relationship with Whole Foods such that it considered buying the grocery chain outright. Whole Foods market cap is currently $11 billion.

Unit economics matter

On-demand businesses tend to be awash in losses early on. That's the nature of building out the infrastructure and user base to support such a dynamic service. But long before the overall operating losses disappear, the business should achieve some measure of profitability on a per-order basis.

The idea, of course, is that the volume of these profitable orders that will eventually overwhelm the macro costs of the business and put the books into the black.

Before charging into an on-demand business, founders should demonstrate to themselves, in an honest accounting exercise, how the unit economics will play out for the startup. If founders are banking on the occasional anomalously-sized order to make up for average or smaller ones, that's a bad sign. The average transaction should be a profitable one.

The founders of goPuff, which provides on-demand delivery of common grocery and drugstore items in 16 cities, did this accounting and came to the conclusion that their network of warehouses, and the margin they'd make on each item, since it would be bigger than those at an Instacart, which doesn't buy the goods from suppliers itself, would work.

GoPuff delivers purchases via a team of workers who use their own cars, just like Instacart, while expecting to earn gross margins on its sold items that are more in-line with a mainline convenience store.

"We're confident," says co-founder Yakir Gola

Gola and his co-founder Rafael Ilishayev have raised $8.25 million thus far.

Keeping overall infrastructure costs down is paramount for startups looking at this space, as well. Most investors will demur from funding something that will require multiple millions just to get rolling. Piggybacking on existing networks is ideal, says Jim Price, an entrepreneur in residence at the University of Michigan's Zell Lurie Institute for Entrepreneurial Studies.

Instacart did this, of course, by leveraging existing retailers and the cars of its delivery staff. Uber's first efforts were around existing networks of black cars and taxi cabs. Pitching investors on being the next Amazon—something that's fully integrated top to bottom—isn't likely to work unless the founders have a history of building billion-dollar companies.

Building an on-demand startup requires solving two critical problems, not just one

When building a SaaS platform, founders can focus on the software, build a good product and then bear down on their critical path: getting customers. With on-demand products, however, the challenges can be greater.

Founders not only need to build out a good software experience and find consumers, but they also need to solve all the logistics involved in an on-demand product. On one end, founders are writing software to solve the kinds of problems all tech companies face, and on the other they're using software, communications and people to solve the kinds of problems that FedEx and UPS deal with. That's acutely challenging.

And the issue of customer acquisition can be two-fold as well, as on-demand companies must attract consumers as well as the part-time workforce that will serve them.

As more gig-like opportunities become available through other startups such as Uber, Instacart and others, the competition for dependable people—who are often required to have their own vehicles—will become stiffer. So, in essence, on-demand companies usually have to market their product to two different groups: the end consumers, as well as the gig-focused workers, if a startup is indeed going to use them in its solution.

Founders choosing to build an on-demand startup should know from the beginning that they're taking on a bigger challenge than most startups whose focus and critical path may be simpler.

Figuring out a way to batch and schedule can defuse challenges and lower costs

A startup that's literally delivering orders as they come in, on an individual basis, will have a much harder time making the unit economics balance out in a favorable direction. Finding a solution that allows for orders to be bundled and batched up can drastically lower the cost of fulfillment.

Creating a solution that neatly does that and also allows for one-off deliveries when necessary requires more nuance and is more complex than building a logistics solution that simply metes out orders and deliveries as they arrive. But it's a necessary hurdle for companies who want to succeed as on-demand solutions.

Thinking out how batching and scheduling will work ahead of time, building out the logistical framework, at least in a rough fashion, should be done before writing any code or building out a nifty landing page full of Javascript widgets.

Have a good answer for the tough question regarding any on-demand business now: how will you beat Amazon and Uber?

This answer isn't just for smart investors who will prompt founders with this question - it's an answer that founders must provide themselves, if they're being honest about their chances of success.

Instacart, for one, has found an edge in forming proprietary relationships with grocery chains. And at this point, it's formed a good deal of trust with consumers. Now it's Uber and Amazon that have to woo customers away from Instacart if they want to gain share in the grocery delivery market.

Grubhub, too, has formed a loyal base of consumers, which will help it grapple with a rising number of challengers: Uber, Amazon, Groupon, DoorDash, Postmates.

Providing a defensible answer to this question is hard, though. One co-cofounder of a major on-demand company, valued at more than $1 billion, told me: "Uber and Amazon have sucked all the air out of on-demand. It's over."

He later allowed that getting into on-demand isn't impossible for startup, but it takes a special market position or proprietary advantage that Amazon and Uber can't readily match. Because if the market is big enough, both of those companies will certainly be there.