A 60-seat bistro paused delivery apps for six weeks. The P&L told the truth.
A Main Street operator went dark on Uber Eats, DoorDash and Skip for six weeks. Delivery revenue dropped 22%. Profit went up. Here's the math she ran and what she kept.
A 60-seat bistro on Main Street paused every delivery app for six weeks this spring.
No Uber Eats. No DoorDash. No Skip. Cold turkey, from a Monday in April to the middle of May.
The owner — call her M, it's her experiment to tell — didn't do it out of principle. She did it because the math had stopped adding up, and she was tired of arguing with a number she couldn't see.
Here's what she found.
Why she pulled the apps
M's delivery business looked healthy from the top. Roughly $14,000 a month in app sales, about 18% of total revenue, growing quarter over quarter. On a spreadsheet, killing it looks insane.
The problem was everything under the top line.
The three apps were averaging 28% commission once you blended promoted placement and the "free delivery" fees she'd opted into to stay visible. Packaging ran another 4%. And the orders landed in clusters — a 9-ticket dump at 7:15 on a Friday, right when the dining room was also slammed, pulling a line cook off the pass to assemble bags.
So the real question was never "is delivery making money." It was "what is delivery costing the room I actually own."
She couldn't answer that without turning it off.
The experiment
Six weeks. Every app paused, not deleted. A short note on each profile: "Pickup and dine-in only for now — see you soon."
She kept three things running so she wasn't just torching demand → her own website ordering for pickup, a phone number on every table card, and a Friday email to her list reminding them she was there.
Then she watched four numbers, not one → net contribution per order, kitchen ticket times during the dinner rush, average dine-in check, and how many of her delivery regulars showed up in person instead.
That last one was the bet. Were delivery guests loyal to her, or loyal to the app?
What the P&L did
Delivery revenue didn't vanish. It moved.
Pickup orders through her own site climbed from about 30 a week to 110. Those orders carried zero commission and the same food cost, so each one contributed more than three times what the equivalent app order had.
Total delivery-channel revenue still fell — down 22% against the prior six weeks. She lost the convenience layer the apps provide, and roughly a third of her pure delivery-only guests never resurfaced.
But profit went the other way.
Net contribution from off-premise dining rose 19% over the blackout, because the orders that remained weren't bleeding 28% off the top. Friday and Saturday ticket times dropped by an average of six minutes once the line wasn't assembling bags mid-rush, and her dine-in check average ticked up 4% — she thinks because the kitchen was finally cooking for the room in front of it.
The headline she wrote on her own whiteboard: sold less, kept more.
What she brought back, and what she didn't
M is an operator, not an ideologue. She turned the apps back on in May — but not the way they were.
She kept one app, the one that drove the most first-time guests who later booked tables. She dropped the other two.
She killed every "free delivery" and promoted-placement fee, accepting lower visibility in exchange for keeping commission near the contracted floor.
And she capped delivery hours → lunch and the early window only, off before the 6:30 rush starts, so the line never gets pulled off the pass during peak again.
Net result, eight weeks after restarting: delivery is now 11% of revenue instead of 18%, and it contributes more dollars than the 18% version did.
The lesson
Most operators track delivery as a revenue line. It's a margin line wearing a revenue costume.
A channel that's 18% of your sales and 6% of your profit is not an 18% channel. It's a 6% channel that's borrowing your kitchen during the only hours your kitchen makes real money.
You don't have to quit delivery. You have to be able to see it. Net out the commission, the packaging, and the labour it steals from the room — then decide which apps and which hours earn their seat.
M ran the blackout because she couldn't see the number any other way. You might be able to model it before you ever turn anything off. But if you can't, six weeks dark will show you the truth faster than any spreadsheet.
Sold less. Kept more. That's the whole report.
— Damon
Frequently asked
Should I drop delivery apps entirely?
Probably not. The point of this experiment wasn't to quit delivery forever — it was to find out what delivery was actually costing in margin, labour, and kitchen focus once you net out the commission. Run the six-week test, read your own P&L, then decide which apps and which hours are worth keeping. For most full-service rooms the answer is fewer apps and narrower hours, not zero.
Won't I lose customers who only order delivery?
Some, yes. In this case delivery revenue fell 22% during the blackout and about a third of those guests never came back to the apps. But delivery-only guests are the lowest-margin, lowest-loyalty cohort she had — they cost 28% in commission and almost never converted to a reservation. Losing the bottom slice of them was cheaper than feeding them.
What's the first number I should look at?
Net contribution per delivery order after commission, packaging, and the labour to assemble it — not gross delivery sales. Most operators track the top-line and never net it out. When she did, the average delivery order was contributing about a third of what a dine-in cover contributed, on more kitchen effort.
Know what every channel actually costs you.
DAMON AI maps your revenue by channel and daypart so you can see margin, not just sales, before you cut anything.
Try DAMON AI →