This story plays out a thousand times a week. The details change. The math doesn't.
She built the site on a Tuesday. Three hours between lunch and dinner service, a laptop balanced on the pass, a setup wizard walking her through menus and photos and pickup windows. By Thursday it was live. Her own ordering page. Her own customer data. Her own margin.
Friday night she checked the dashboard. Four orders. DoorDash had done forty-three.
She knew the math before she started. Twenty-five percent of forty-three orders was bleeding her. Three percent of four orders wasn't saving her. The escape route was right there on the screen. The customers hadn't found the door.
The delivery platforms take 15-30% of every order. Nearly half the customers who receive the food can't remember whose restaurant made it. That problem is real. The question underneath it is harder: what can an operator actually do?
The answer is more honest than most people selling direct ordering software want to admit. The tools exist. The math works. For some.
The other door
The operators getting out aren't leaving the platforms entirely. They're changing the balance. Treating the marketplace as one channel, not the channel. And building what they need to own the customer directly.
The tools are not new. ChowNow has been selling commission-free direct ordering since 2012. Owner.com bundles a website, an app, marketing automation, and its own driver network. GloriaFood and Square offer free ordering systems with standard processing fees. The technology works. Twenty-two thousand restaurants are on ChowNow alone, recovering an average of $16,000 a year in commissions they used to hand to a platform. And the industry hasn't moved.
So why are most operators still on the marketplace?
Because the tools solve the ordering problem. They don't solve the customer problem. A direct ordering page with no traffic is a menu nobody reads. The platform's commission was never just for processing. It was for the 43 people who searched "Thai food near me" at 7pm and tapped the first result. The tools replace the fee. They don't replace the demand.
That math is real. For the operator with $40 average orders and a name people already search by. The owner staring at four orders on a Friday night has access to the same tools and the same math. She's not missing the technology. She's missing the customers who don't know her name yet. The gap between "available" and "viable" is where most operators live.
Direct ordering solves half the problem. The customer orders through your site. But who delivers it?
The delivery problem Panera already solved
In 2017 Panera built their own delivery operation. Own drivers. Own app. Own customer. Full control. The playbook every consultant recommends.
By 2021 they shut it down. Driver economics killed it. Hiring, insuring, scheduling, and managing a fleet across 2,000+ locations cost more per delivery than the marketplace commission they were trying to escape. Panera had a national brand, a loyal customer base, and $5 billion in revenue. If they couldn't make owned delivery work, a 15-unit regional chain wasn't going to figure it out either.
Here's the part nobody talks about.
DoorDash sells two products. Most operators only know the first one: the marketplace, the app, the customer, the search ranking, the 25-30% commission. The second is DoorDash Drive. Logistics only. A driver picks up the order and delivers it. No marketplace listing. No customer capture. No commission on the food. Flat fee, roughly $8 per delivery. The restaurant keeps the customer. DoorDash keeps the delivery fee. Uber does the same thing with Uber Direct. White-label. The customer orders through the restaurant's own site, pays the restaurant directly, never sees the Uber brand.
Read that again. The escape route was built by the same company the operator is trying to escape. They just don't put it on the homepage.
But that $8 delivery fee has to come from somewhere. Most operators pass it to the customer. Some absorb part of it. And that's where the math gets honest.
The breakeven most operators miss
Casual dining and fast casual operators typically run orders of $30 to $45. The math works for them. The median QSR delivery order sits closer to $22 to $28. For those operators, the logistics-only model doesn't save money. It shifts the cost and calls it progress.
Even when the math works, the fee creates a different problem. Marketplace customers don't see a delivery charge; it's baked into the commission. Direct customers see the $8 at checkout. Some close the tab and open DoorDash. The fee is visible in one model and invisible in the other. That's not a math problem. It's a behavior problem.
The owner on the pass? Her average order is $34. Just above the line. The math says she can do it. Whether her customers will is a different question.
Who can actually leave
The logistics separation changes the game for some operators. Not all. The honest answer about which ones isn't comfortable.
The owner checking her dashboard is somewhere between "strong independent" and "discovery-dependent." She has regulars who would order direct if they knew the option existed. She also has Tuesday night's forty-three DoorDash orders, most of which came from people who searched "Thai food near me" and picked the first result. Those customers aren't following her to a new site. They'll order from the next restaurant in the feed.
About ten years ago I sat in a room with One Off Hospitality in Chicago. Paul Kahan, Donnie Madia. Operators behind Blackbird, Avec, The Publican. Some of the best restaurants in the country. They had the brand, the kitchens, the infrastructure, even catering vans sitting idle that could have run delivery. They had everything they needed to own a direct channel before the platforms existed. They chose to stay pat.
When Covid hit, Blackbird closed after 22 years. Avec closed. The brand that should have been untouchable wasn't. I don't tell that story to blame anyone. These were world-class operators who made the same calculation most world-class operators made. The tools were available. The urgency wasn't. By the time it was, the window had closed. Having the tools and using the tools are different things. Most operators already know that.
Knowing isn't the hard part. Doing it while keeping the lights on is.
The cost of walking out
She tried running both for a month. DoorDash orders on one tablet, direct orders on another, two pricing structures because the direct menu couldn't carry the 15% markup she'd built into the marketplace menu to cover commissions. A regular noticed the gap between her site and DoorDash. Posted about it. She spent the next morning on the phone explaining why the same pad thai cost $16 on her site and $18.40 on DoorDash. The answer was honest. The conversation was exhausting.
That's the hybrid model. It works in theory. In practice it means two menus, two pricing structures, two sets of customer expectations, and a marketing budget the operator never asked for. The platform's commission included demand generation. Going direct means replacing it: Google ads, social media, email campaigns, loyalty programs, and QR codes on every takeout bag.
For the owner on a Tuesday night who just built her site, "going direct" now means learning digital marketing between prep and service. That's the real cost. Not the processing fee. The time.
Platform customers repeat at 15-25%. Direct customers repeat at 35-55%. Those numbers only matter if the operator has the brand and the capability to build the relationship. Without them, the repeat rate is academic. There is no painless exit from the platform economy. The operators who make it work don't escape the platforms. They outgrow them. One regular at a time.
The next landlord
Everything above assumes the categories stay fixed. POS companies on one side. Delivery platforms on the other. The operator choosing between them. That distinction is dissolving faster than most operators realize.
Two companies that started on opposite sides of the restaurant are building toward the same thing. Toast began at the register and is pushing outward: inventory, labor scheduling, supplier management, and now a consumer discovery app called Toast Local. DoorDash began at the doorstep and is pushing inward: website builder, email marketing, loyalty programs, and a proprietary POS piloting in San Francisco. They're not competing on features. They're competing to become the operating system.
They're not the only ones. Qu is building a unified operating system for enterprise QSR from scratch, backed by Danny Meyer. Travis Kalanick's Atoms bundles a kitchen platform, a POS, and robotics under one roof. Everyone is building toward the same thing: a single system that touches every part of the restaurant.
And every one of them will eventually face the same question the delivery platforms face today. Toast Local doesn't charge commission yet. "Yet" is doing a lot of work in that sentence. The discovery layer and the ordering flow already live inside Toast's infrastructure. If DoorDash processes the payment, manages the loyalty, and owns the POS, the restaurant runs its entire operation on DoorDash rails. That's not a delivery company anymore. That's a landlord with better software.
The owner on the pass is watching this from her Toast terminal. The system she bought to own her data is quietly becoming the system that might take it back. When one company runs the ordering, the kitchen, the inventory, the marketing, and the payments, the restaurant doesn't have a vendor. It has a dependency. And the question that matters most now applies to more of the operation than it ever has.
That's the question underneath every vendor pitch, every product demo, every partnership proposal. Not "what does it do?" but "who does it serve when the incentives tighten?" And there's a second question that matters just as much: what do I own when I leave? If the answer is nothing, no customer data, no order history, no email list, you're not using a tool. You're renting one. The rent will go up.
The escape route is real. The tools exist. For operators with the brand, the volume, and the order value, the math works. For the rest, the platform is still the only way customers find them. That's not a comfortable answer. It's an honest one.
The owner who built her site on a Tuesday is still running it. DoorDash still does most of her delivery. But three months in, twenty-six regulars order through her page. They pay her directly. She has their names, their emails, their order history. It's not enough to quit the platform. Not yet. But last week one of them called to say the green curry was the best in the neighborhood. Called the restaurant. Not the app.
That's the escape route. Not a platform. Not a feature. A decision.