Case study — Technology · 11 min read · 5 questions
Why food delivery is never profitable
Published · Updated
The thesis
In 2014 the barriers to entry in food delivery were so low that a few thousand dollars and a website got you a company in weeks, which is exactly what happened here — the author left college to start one. The trap is that delivery is a marketplace, and marketplaces have a circular dependency you cannot fund your way out of at small scale: no customers means no order volume, no order volume means no restaurants, no restaurants means no customers. There are three ways to win — speed, price, or exclusive supply — and the first two need a war chest no startup has.
So the strategy was supply: food trucks nobody else served, one curated menu, photography and delivery provided free for 30% of net sales. It took 10 to 20 hours to onboard a single vendor. In seven months it delivered 3,400 orders and grossed $35,000 — which felt enormous and never paid anyone. Eventually every mobile food business within 75 miles had been onboarded, the menu went stale, and customers stopped ordering. There is cash flow in delivery and no profit.
Eight years and one pandemic later the giants proved the same thing at scale. DoorDash grew revenue 1,580% and still loses half a billion a year, and its take rate crawled from 10.35% to 11.65% — after a demand shock it still could not demand a bigger slice. Uber Eats states in its own filings that payments to Eats drivers have historically exceeded the delivery fees customers paid. When a model needs customer tips to pay a living wage, value creation is not the problem. Value capture is.
How do you think about this? 5 strategy questions this case raises and does not answer.
Read the comments, or add yours
The statistics
By the numbers — swipe or use arrows
Revenue, order counts, monthly active users, marketplace gross order value, net revenue margin (take rate), contribution margin, and net losses from DoorDash, Uber Technologies, Delivery Hero SE and Just Eat Takeaway.com annual reports, 10-K filings and investor presentations, 2018 through 2021; the delivery cost disclosure and driver pay model breakdown as published by the companies themselves; 2014 startup figures from the author's own company records
Key takeaways
There are exactly three ways to win and two are unavailable to anyone small. Speed requires enough drivers to out-route the competition; price requires swallowing a loss on every order until volume covers it; supply requires exclusivity. No investor approves the burn for the first two, so startups chase narrow supply niches that are too small to sustain.
The chicken-and-egg is the whole business. No customers means no order volume, no order volume means no restaurants, and no restaurants means no customers. Every new restaurant adds choice — and dilutes the average restaurant's sales, worsens discovery, and gives owners a reason to leave for better terms elsewhere.
The author's own startup took the supply route: food trucks and street vendors neglected by the apps for having no brick-and-mortar location, on one curated menu where customers could combine dishes from multiple vendors in a single order.
The numbers looked good and meant nothing. 3,400 orders and $35,000 gross in seven months off a $2,500 bootstrap — never enough margin to hire a single person or pay the founders, who did the photography, flyers, 4AM packing, driving, marketing and site updates themselves.
DoorDash won by going where nobody else went. While delivery was most visible in Manhattan, Chicago and Los Angeles, DoorDash took suburban and small metropolitan markets — fewer competing options, manageable customer acquisition costs, and residents who drive far enough that delivery is worth more to them than to city dwellers.
DoorDash does not hide that the tip is load-bearing. Driver pay is base pay (by time, distance and desirability), the customer's tip (kept in full), and promotions — which are too non-transparent and inconsistent to count as regular earnings, varying by when a driver signed up, what code they used and where they are.
Its own example order proves the point. On a $33 order — $22 food, $6 service fees, $2 tax, $3 tip — the driver earns $8, the restaurant gets $20 and DoorDash keeps $5. The customer's tip is nearly half of what the driver is paid.
The pandemic growth was extraordinary and changed nothing structural. Revenue grew 1,580% in three years: $290M on 83M orders in 2018, $890M on 263M orders in 2019, $2.8B on 816M orders in 2020, and $4.8B on 1.3 billion orders and 25M monthly actives in 2021 — against losses of $500M a year and an average −42% operating margin.
So the company asks investors to judge it on other metrics. Marketplace gross order value — everything customers pay, including food, tax, tips and fees — grew from $2.8B to $42B across those four years, while DoorDash's own revenue is only the 15-30% restaurant commission plus consumer service fees.
Take rate is the metric that actually matters, and it has barely moved: 10.35% in 2018, 11% in 2019, 11.7% in 2020, and 11.65% in 2021. It answers whether DoorDash can claim a bigger slice as the pie grows — and the answer is no.
That flatness reveals how fragile the position is. Raise commissions and restaurants leave; raise service fees and price-sensitive customers leave; cut driver pay and drivers leave. If a company cannot demand more during a pandemic that guaranteed its growth, when exactly can it?
Contribution margin is the friendliest number and it is still not enough. It was −20% in 2018 and 2019 — losing $2 on every $10 order — and turned positive from 2020, reaching about $2.22 per $10 order. But it excludes R&D and G&A, which together run over $1 billion a year and produce the half-billion-dollar net loss.
The strategy now is frequency, not margin. DashPass at $9.99 a month trades lower unit margins for higher order frequency — the Costco logic, where membership fees are a small share of revenue but drive regular large orders — with DashMart extending it into groceries and convenience essentials.
Uber Eats says the quiet part in its own filings: cumulative payments to Eats drivers have historically exceeded the cumulative delivery fees paid by customers. It grew from $1.4B to $8.3B in revenue on $51.6B of gross order value, and offsets a −2% contribution margin by taking a 16% slice — five points more than DoorDash.
Delivery Hero has the worst margins of the four. It grew from €687M in 2018 to €5.8B in 2021 and still posts €1 billion annual losses — losing 14-20% on every order even while taking 20% of the pie. Europe and Asia at least do not depend on tipping, and stricter labor law means it employs drivers directly.
Just Eat proves the take rate is the only lever and it does not hold. It grew from €2.7B to €5.3B partly by acquiring GrubHub, posting €200-300M annual losses. It only stopped losing money per order by hiking its take rate to 30% during the pandemic — and once forced back down to 19%, immediately returned to −7% per order. Investors, facing a 75% valuation drop, wrote an open letter demanding the CFO and board be fired.
Common questions
Is food delivery profitable?
Not for the companies running it. DoorDash grew revenue 1,580% in three years to $4.8 billion and still loses about half a billion dollars annually at a −42% operating margin. Uber Eats discloses in its own filings that cumulative payments to Eats drivers have historically exceeded the delivery fees customers paid. Delivery Hero loses 14–20% on every order despite taking 20% of the transaction, and posts €1 billion annual losses. Just Eat only achieved positive per-order economics at a 30% take rate during the pandemic, and returned to −7% when it fell back to 19%.
How much of a DoorDash driver's pay comes from tips?
Roughly half, by the company's own published example. On a $33 order — $22 of food, $6 in service fees, $2 tax and a $3 tip — the driver earns $8, the restaurant gets $20 and DoorDash keeps $5. That makes the customer's $3 tip close to 50% of the driver's earnings. DoorDash explicitly states that if customers consistently tip low or not at all, there would be material harm to its ability to provide delivery and retain drivers, because base pay alone is inadequate.
What is take rate and why does it matter?
Take rate, or net revenue margin, is the percentage of everything customers spend that the platform keeps for itself. It matters because it measures leverage: whether the company can demand a bigger slice from restaurants, drivers or customers without losing them. DoorDash's went 10.35%, 11%, 11.7%, 11.65% across 2018 to 2021 — essentially flat through a pandemic that guaranteed its growth. That flatness says the platform cannot raise commissions without restaurants leaving, cannot raise fees without customers leaving, and cannot cut driver pay without drivers leaving.
Why did DoorDash beat Uber Eats and GrubHub in the US?
By taking the markets nobody was fighting over. Delivery was most visible in dense cities like Manhattan, Chicago and Los Angeles, but DoorDash targeted suburbs and small metropolitan areas that were traditionally underserved — fewer competing options meant meaningful customers at manageable acquisition costs. The unit economics are also better there: suburban customers are families rather than individuals, so orders are larger, and lighter traffic with easier parking lets drivers complete deliveries faster than in cities.
What is contribution margin and why is DoorDash's misleading?
It is what remains after deducting the direct costs of generating and fulfilling an order — sales, marketing, delivery, payment processing, insurance, support and platform operations. DoorDash's was −20% in 2018 and 2019, meaning it lost $2 on a $10 order, and turned positive from 2020 to about $2.22 per $10 order. The problem is what it excludes: R&D and general expenses run over $1 billion a year and are the reason the company still posts a half-billion-dollar net loss. A metric that omits your largest costs is not a measure of profitability.
Why is it so hard to start a food delivery company?
Because it is a marketplace with a circular dependency you cannot escape at small scale. Without customers you have no order volume; without order volume no restaurant will partner with you; without restaurants you cannot attract customers. Winning requires competing on speed, price or exclusive supply — and the first two need capital no startup has, so small players chase narrow supply niches that are too small to sustain. Adding restaurants helps and hurts simultaneously, since more listings dilute each restaurant's sales and make discovery harder.
What is DashPass and how does it help DoorDash?
A $9.99 monthly membership giving unlimited free deliveries and lower service fees at eligible restaurants. It deliberately trades lower margin per order for higher order frequency — the Costco model, where membership fees are a small share of revenue but drive regular, larger, more frequent purchases. DoorDash's leadership has concluded that the path to profitability lies in frequency rather than in squeezing out margin points, since there is little they can do to influence order size and the remaining delivery inefficiencies are too expensive to chase.
How did inflation and fuel costs affect delivery companies?
They exposed how little pricing power these platforms have on either side. DoorDash introduced 10% fuel cashback and distance bonuses to retain drivers, and absorbed both costs entirely rather than raise consumer prices, explicitly to protect order volume and growth. But that only controls one variable — restaurants raise their own menu prices to cover ingredient and labor costs regardless, and any increase in the platform's commission gets passed to customers anyway. The stated strategy of keeping consumer prices unchanged is not actually within the company's control.
Discussion
DoorDash could not raise its take rate past 11.65% even during a pandemic that guaranteed demand. If maximum leverage is not enough to claim a bigger slice, is there any future state where a delivery marketplace has pricing power?
No answers yet — be the firstThe company states plainly that low tipping would materially harm its ability to operate, then declines to educate or require tipping. Whose problem has that decision made the tip, and is that sustainable?
No answers yet — be the firstContribution margin excludes the billion dollars a year of R&D and G&A that produce the net loss. When is a non-GAAP operating metric a genuine lens on the business, and when is it a way to change the subject?
No answers yet — be the firstJust Eat's margins only worked at a 30% take rate and collapsed the moment it fell to 19%. Does that tell you the correct commission is 30%, or that the business does not work at any rate restaurants will accept?
No answers yet — be the firstThe author's own startup died between supply and demand at $35,000 of revenue. DoorDash is dying of the same thing at $4.8 billion. What actually changes with scale here, and what provably does not?
No answers yet — be the first
Related case studies
TechnologyWhy AI is just big data againBig data was sold for a decade and produced nothing. The same players are running the identical playbook with AI.15 min read
TechnologyWhat happened to GoProIt invented the action camera and then spent a decade learning what an action camera is for.11 min read
TechnologyWhy Opendoor and Zillow failed at flipping housesIt needs to flip 60,000 homes a year to break even. It flipped 21,000. The rest is debt and vocabulary.12 min read