June 2016

Ordering in with
a unicorn
Delivery Hero CEO Niklas stberg describes how his company
creates value.
Niklas stberg, an energetic 35-year-old Swede, is the CEO and cofounder
of Delivery Hero. Based in Berlin and financed with venture-capital money,
the company is built around an online platform that matches restaurants
with hungry customers. Delivery Hero has grown to operate today in 33 markets
across five continents, processing 14 million takeout orders each month and
offering customers recommendations, as well as peer reviews of restaurants.
With a valuation of $3 billion, Delivery Hero is also one of about 170 unicorns:
start-ups with valuations above $1 billion. Given the number of new
companies that crashed when the turn-of-the-century tech bubble burst,
many executives and investors have cast a skeptical eye on the unicorn
phenomenon. stberg recently discussed with McKinseys Thomas
Schumacher and Dennis Swinford the start-up landscape, the importance
of innovation grounded in data, and his companys role as a disruptor of an
inefficient restaurant industry.
The Quarterly: Valuations of pre-IPO tech companies have come under

scrutiny lately, particularly the emergence of so-called unicorns. Whats going

on, in your perception?
Niklas stberg: Im sure a number of those unicorns shouldnt be unicorns.

As always, earlier-stage businesses come at a higher risk. But I am also

sure that the next Google or Apple is among themand if only one or two

of the current pool of unicorns get to that level, it justifies their valuations,
collectively, from an investor point of view.
But a lot has changed in the 15 years since the tech bubble of 2000. At
that time, many valuations were based on what the future might look like,
particularly in the Internet space, rather than on the returns a business could
demonstrate. The supposition was that the world was changing and would
probably change for the better as people went online. And although people
did eventually go online, that happened much more slowly than predicted.
Today, theres no doubt that online and Internet businesses are taking over.
Some of the biggest businesses in the world, including Facebook, Amazon,
Google, and Apple, are solidly grounded in the new world of technology. A lot
of other companies also have large, tangible revenue growth and earnings.
They dont buy users or customers with the hope of making money when,
maybe, those users eventually change their behavior. Delivery Hero, too,
generates a lot of revenueand earns a lot of profit in many markets. So
valuations dont depend on imaginary future earnings but on actual returns
and EBITDA.1
The Quarterly: How does your business model work?
Niklas stberg: Were a place where users and restaurants meet. The core

of our business is an online platform that allows us to map users to the

restaurants around them. Users are attracted to the platform and become
very loyal to it because it helps them identify which restaurants are available
and which ones are good. Its also convenient because they can pay online,
review past orders, and chart their savings.
Its a good model for restaurants, too. We channel more business to them and
they increase their orders. And because the variable cost of food is pretty
low, adding incremental customers is pretty lucrative. A restaurant that
serves 100 orders a day might not make a lot of profit, for example, but if it
boosts that to 110 orders a day, it would make good profits. Boost that to
200 orders a day, and it will make loads of money. So restaurants want to be
on our platform, and we charge them a fee for transmitting orders. If they
decide they no longer want to be on the platform, customers can order from
other restaurants.
1

Earnings before interest, taxes, depreciation, and amortization.

NIKLAS STBERG
Education
MSc in industrial
engineering and
management, KTH Royal
Institute of Technology
and ETH Zurich
Career highlights
Delivery Hero Holding
(May 2011present)
Cofounder and CEO

OnlinePizza
(November 2007
May 2011)
Cofounder and chairman
Fast facts
Provided capital and
advice to several European
start-ups as an angel
investor, including
Beekeeper, GetYourGuide,
and Peakon

Everything is automated and online, so our gross profitability per order is

around 90 percent. That also comes back to why we want to growbecause if
you have 90 percent gross profitability and low variable costs, the closer you
get, in theory, to 90 percent net profit. This compels us to build scale to add
those incremental users and get closer to that 90 percent EBITDA margin. In
some markets, we have already reached over 60 percent.
The Quarterly: Who are your competitors?
Niklas stberg: The usual way of ordering food is to pick up the phone and

call, so our biggest competitor is still the phone. And most people also still
cook, though only some of them actually like doing it. So why shouldnt we
get the many who dont like cooking? At a societal level, is it efficient for every
little household to do its own cooking? For everyone to go to the supermarket
and shop for groceries individually, versus buying groceries and preparing
meals for 100 people at once? More and more, people dont cook as long
as they can get the healthy food they want when they want it. Thats our
challenge, thento improve the inefficiency of that industry, to make it more
accessible and available.
The Quarterly: Youre talking about disrupting the entire social network of

how people eat?

Niklas stberg: I think we should, over the long term. Of course, you cant

do that all at once, but if you look over ten years, why not? Our focus is first
to attract those customers who order by phone and then to keep attracting

new customers by making the service better. Every small, incremental

improvement takes us one step closer. And at some point, maybe well have a
service thats so good, why would anyone cook?
The Quarterly: So if home cooks and the telephone are your major competitors,

who really worries you?

Niklas stberg: We do also have competitors in our own space. Uber, for
example, and Amazon and Yelp have similar efforts under way. Its a big
space, so why wouldnt they try? Even Facebook could enable online food
ordering via chat bots, which could completely change the industry yet again.
And, indirectly, guys like Facebook could become our competitors because
they could connect to someone else who provides restaurant info to their
chat bots. And Google, continuously offering better access to information, is
already offering restaurant data, including restaurant menus. So if we dont
stay innovative and dont stay the best and dont offer access to the best
and fastest food, then in the long term we are in trouble. Thats why we can
never relax.
The Quarterly: Do the restaurants get more value out of this than just reaching

more customers?
Niklas stberg: We try to give them as much value as we can, and its part

of our vision to do so. Besides attracting more customers, we reduce their

operational costs, since they dont need to have someone answering the
phone, for example. We also provide them with a point-of-sale system
replacing the cash register and we compile useful statistics. That will
not only save some thousands of euros per year but also help them provide
better food and service to their customers.
And while we expect to do more in the next year or so, were already able to
tell restaurants which menu items are likely to work. We can say, for example,
it looks like theres no one in your area providing a bacon burger. Why dont
you add a bacon burger to your menu? We can say which dishes always
bring customers back. Conversely, we can also tell which menu items draw
customer complaints or have very low reorder rates. Customers order, but
never return. Every time someone buys that dish, the restaurant loses
a customer.
The Quarterly: Innovation is most successful when it disrupts what already

exists. Who are you disrupting?

Niklas stberg: I would say that we are a disruptor of an inefficient

restaurant industry. Were disrupting bad service, inefficient
manual processes. Were disrupting inefficiencies in how restaurants
connect with customersnot every restaurant can build its own online foodordering platform. Were disrupting inefficiencies in delivery. It makes no
sense for every small restaurant to try to have its own delivery fleet with its
own drivers, given the cost of maintaining a fleet and coordinating deliveries.
After all, if a restaurant five kilometers away delivers to someone in one place
and then goes five kilometers in another direction to deliver to someone
else, its expensive. Its bad for the environment. And its bad for customers
because it takes so long.

Were also disrupting the inefficiency of a system that doesnt serve the food
customers want. If you were to ask people on the street, a lot of them would
say, I dont like delivery because I dont eat pizza or its just bad quality and
bad food. Combined, those inefficiencies raise costs and reduce quality.
The Quarterly: How are you using all the data you generate to improve

your business?
Niklas stberg: Big data should actually be big, meaning it should be

available to the entire organizationespecially at the front line of the

business. Thats where companies make tens of thousands of decisions every
day, some of which can be handled automatically. These can be very small
things, like shall we do this kind of promotion for our users? Is that a good
channel for our advertising? How do we improve our relationship with a
specific customer? If a restaurant has very bad delivery on Sunday evenings,
we can downgrade it on Sunday evenings. If the system detects fraud, we can
trigger people to stop ordering.
We also monitor our restaurants to maintain relationships. We know,
for example, that a restaurant is likely to cancel its contract if it starts
contacting us more frequently or gets negative feedback from customers.
The data automatically trigger a pop-up to one of our sales agentscall this
restaurant, see whats wrong, and do what you can to help. This involves
decisions that are made both automatically and independently by sales
agents, as long as they have the right information, and saves a lot of money.
The Quarterly: Do data also help inform investment decisions?
Niklas stberg: Data help us to be a little faster at managing our investments.

Say you make an investment with a one-to-ten probability that youll be

rightbut if youre right, youll make a 100-to-1 return. Thats a very good
investment to try. The problem is that if youre wrong in nine out of ten cases,
you need to have a very fast way of figuring that out. Then, when you do find
the one investment with high returns, you can put a lot of money on it.
For example, while the main part of our offering is the online platform, weve
also invested in separate businesses to handle delivery for independent
restaurants. That is part of building up our logistics to enable a better service.
Restaurants still do the cooking, naturally, but we track their orders. We offer
quality assurance through metrics like user ratings and reorder rates. And
we tell restaurants which dishes on their menus are good for delivery. We also
make much more money on thataround 10 per order, less the cost of delivery.
For investments like that, we track the data and optimize performance,
shutting them down quickly if it becomes clear they cant meet our
expectations. We spent nine months on an earlier delivery-space investment,
based on a different concept and setup, for example. We did as much as we
could to improve its performance and invested close to 10 million in the
project. But it wasnt meeting our expectations, so we shut it down and took
the loss. Now, maybe we could have realized that sooner and lost just
6 million, but other companies might have dragged out the investment and
spent 100 million on it. The point is, if youre going to fail, you want to fail
fast. You invest to validate or invalidate the concept and then shut it down
if necessary.
The Quarterly: You appear to have a highly federated business model with a

number of CEOs of individual delivery businesses. How does that work?

Niklas stberg: Centralization is always more efficient, in a way, because

you can do one thing and multiply it across units. On the other hand, giving
people autonomy and authority and responsibility also has an amazing value.
What rarely works is to be 100 percent one approach or the other. The trick is
finding the right balance.
We give local CEOs autonomy and authority to encourage entrepreneurship
and they fight with blood and sweat to win in the market. But you have to set
the rules of the game. And you have to set the culture of your company. That
balance can be fragile. For example, if you set the wrong incentive scheme
and you place autonomy at the local level, people are more likely to optimize
for their incentive schemes rather than for their businesses. And, suddenly,
youre sitting there on a conference call wondering, is this the right decision
that hes suggesting or is this the right decision for him? And you dont really

know. Thats why, first of all, its important to find people with an owner
mentality rather than a manager whose career and financial interests are the
top priority. Then give them an incentive scheme that reflects ownership as
closely as possible.
Finally, were a data-driven culture. Decisions based on data are the glue
that holds us together. And if data are your starting point, then a CEO in
Argentina, for example, cant just argue that we should do it this way because
every Argentinians doing it this way. We might not agree, but we can do
the A/B testing and see what the data tell us. CEOs get the final decision, but
if they cant prove that their way is better and still do things their way, its a
question of judgment. You can be wrong many times as long as you address
the issue.
The Quarterly: If we look back in our imaginations five years from nowsay,

after an IPO or acquisitionwhat would have to happen for Delivery Hero to

fail? And what must happen for it to justify its considerable valuation today?
Niklas stberg: Were in an economy that moves fast. It would be terribly

dangerous to think that something cant go wrong or that we cant be

disrupted. That could happen, especially if someone comes along with an
innovation and were not already there. So we are alwaysand I think you
have to beon the edge of innovating and on the edge of moving fast. Thats
whats required of companies at our stage.
In terms of revenue, were in a good position. This is true even if I dont argue
that we can grow over 50 percent five years in a row, though I think we could;
even if I dont assume that we can improve our unit economics, though I think
we will; and even if I dont assert that we can increase our pricing, though
I think we can. Todays valuation is not built on some utopian assumption
that the world will change and people will suddenly start ordering food in
a certain way. People already order food onlineand we have the data. We
are the market leader in at least 25 markets. We have a business model that
people like. And every second of every hour, we deliver 16 meals globally,
hundreds of millions of orders a year. I think weve proved we can make a
profit out of that.
Thomas Schumacher is a partner in McKinseys Dsseldorf office. Dennis Swinford is
a senior editor of McKinsey Publishing and is based in the Seattle office.
The interviewers would like to thank David Cogman, Philipp Hillenbrand, Carsten Hirschberg,
and Tim Koller for their help preparing for this interview.
Copyright 2016 McKinsey & Company. All rights reserved.