Delivery volume is easy to get. Profitable delivery volume is not, and the gap between those two things has closed a lot of local food businesses that were, on paper, busier than ever.
The trap is that delivery revenue looks like revenue. A $50 order on a marketplace and a $50 order at the counter land in the same sales report. One of them has already given away fifteen dollars before you’ve paid for food, packaging or the driver. That’s the whole problem in one sentence.
Food delivery for local businesses is a margin question before it’s a logistics question. The good news is the margin question has a definite answer, and it’s arithmetic rather than opinion: what a stop costs you, what an order is worth, and how many stops you can fit into an hour of paid driving.
The Bottom Line
- Marketplace commissions run 15% to 30%, and Rezku’s 2026 breakdown puts the all-in cost at 30% to 40% per order once processing, marketing and adjustments are counted. Against a typical 3% to 5% net margin, that’s the entire profit on the order.
- Route density does more for your cost per delivery than any negotiation. EasyRoutes’ cost analysis finds that raising drops per stop from one to five cuts labor and vehicle cost by more than half.
- Most local operators end up with a mix, not a choice: a marketplace for discovery, own channel for repeat, and a courier partner for the peaks.
- The number to manage is contribution margin per delivered order, not delivery revenue and not order count.
Boost customer satisfaction with just a few clicks
Most-Loved Features:
- On-demand drivers
- Real-time GPS tracking
- Delivery confirmation photos
- Over 50% of customers report a smoother delivery experience
Why food delivery for local businesses is a margin problem first
A restaurant, bakery or caterer usually starts delivering for defensive reasons. Customers asked. A competitor did it. A marketplace rep called. The decision gets made as a yes-or-no, when the decisions that matter are all downstream of it: who drives, what you charge, how far you’ll go, and how many orders you need in the same direction at the same time.
Three structural facts shape all of them.
Delivery adds cost per order rather than per month. A new oven is a fixed cost you spread across output. A delivery is a variable cost that repeats on every single order, so the unit economics have to work at order one.
Food is also the hardest thing to deliver. It’s temperature-sensitive, time-critical and worth less per pound than almost anything else in a van. That combination is why grocery and prepared-food drops run around $10 to $20 per package when general urban parcel delivery costs less.
And the channel takes its cut off the top. Commission applies to the full order value, before your food cost. On a low-margin business, a 30% commission isn’t a 30% reduction in profit. It’s usually more than all of it.
Delivery is still worth doing. It just has to be priced and routed deliberately, which most operations never quite get around to.
The three ways a local food business can fulfil a delivery order
There are only three, and most businesses end up using two of them.
In-house drivers. Your employees, your vehicles or theirs, your rules. Lowest cost per order at volume, highest fixed commitment, and you own hiring, insurance, scheduling and the vehicle.
Third-party marketplaces. DoorDash, Uber Eats, Grubhub and their regional equivalents. They bring demand you didn’t generate and drivers you don’t manage, and they charge for both.
A courier or delivery partner. You take the order on your own channel and hand the trip to an outside driver network, paying for the delivery rather than a percentage of the food. Variable capacity without the payroll.
| In-house drivers | Marketplace | Courier partner | |
|---|---|---|---|
| Typical cost basis | Wage + mileage + insurance per hour | 15–30% commission, 30–40% all-in | Per-delivery fee, often distance-based |
| Brings you new customers | No | Yes | No |
| Who owns the customer data | You | The platform | You |
| Scales down on a slow day | Poorly, you’re paying anyway | Perfectly | Well |
| Scales up on a holiday peak | Poorly, you’re capped | Well | Well |
| Control over the handoff | Full | Minimal | Partial |
| Best fit | Steady daily volume, tight delivery area | Discovery and overflow | Variable volume, wider area, own-channel orders |
The honest read: in-house wins on cost only when you have enough consistent volume to keep a driver busy. A driver at $22 an hour all-in who completes two deliveries an hour costs you $11 a drop. The same driver at five drops an hour costs $4.40. Nothing else in this article moves the number that much.
What a delivery order actually costs: a worked example
Take a $50 order and run it both ways. Assume food cost at 30% and packaging at $2.50, the same for both.
Through a marketplace, at 30% all-in:
- Order value: $50.00
- Platform cost: −$15.00
- Food: −$15.00
- Packaging: −$2.50
- Left: $17.50, before rent, labor and everything else
On your own channel with an in-house driver at five drops per hour:
- Order value: $50.00, plus a $5.00 delivery fee
- Payment processing at ~3% + $0.30: −$1.95
- Food: −$15.00
- Packaging: −$2.50
- Driver cost per drop: −$4.40
- Mileage, three miles at the IRS 2026 second-half rate of 76 cents: −$2.28
- Left: $28.87
Eleven dollars of difference on one order. At thirty delivery orders a day, that’s over $120,000 a year. So the discovery a marketplace buys you has to be bringing in people who didn’t know your name, rather than taking a cut of the ones who did.
Run this calculation with your own numbers before you change anything else. Most operators have never done it, and the result usually reorders their priorities immediately.
How route density decides your cost per delivery
Cost per delivery is mostly a function of how close together your stops are. That’s it. Everything else is rounding.
EasyRoutes’ breakdown of delivery transportation costs puts it plainly: raising drops per stop from one to five cuts labor and vehicle cost by more than half. The driver is paid for the hour either way; density decides how many orders that hour is divided across.
Four levers change density:
Shrink the zone. A tight delivery area produces clustered stops. A generous one produces a driver on a highway. Most local businesses would earn more by cutting their radius and raising their minimum than by adding a driver.
Batch by time, not by order. Holding orders for a 15- or 30-minute dispatch window so three can go out together is the difference between one drop per trip and three. Customers accept a stated window; they don’t accept a surprise late order.
Group by geography at dispatch. Assigning the next order to whichever driver is already heading that way, rather than whoever’s free, is worth more than it sounds. Route optimization tools report 15% to 25% reductions in cost per delivery largely from this effect.
Steer demand toward density. A free-delivery threshold in your strongest neighborhoods, a scheduled delivery day for the far edge of your zone, a small fee band for the outskirts. Pricing shapes the map.
A practical benchmark to aim at: five or more completed drops per paid driver hour. Below three, in-house delivery is usually costing you more than a per-delivery courier fee would.
Delivery pricing: fees, minimums and zones that protect margin
Pricing is where most of the recoverable money is, because it’s the cheapest thing to change.
Charge a delivery fee, and don’t apologize for it. Free delivery is a marketing choice, and it has to be funded by something: a higher minimum, a higher menu price, or a threshold. Unfunded free delivery is just a discount you didn’t decide to give.
Set the minimum from the cost of a stop. If a drop costs you roughly $7 all-in, a $20 minimum on a 30%-food-cost menu doesn’t clear it once labor is counted. Two to three times the cost of the stop is a workable starting point.
Band the fee by distance or drive time. One flat fee across a wide zone means your near customers subsidize your far ones, and your far customers order more because they’re getting the better deal. Two or three bands fix it.
Use a threshold as the upsell. “Free delivery over $60” set just above your current average ticket raises order value and improves density at the same time, because bigger orders justify the trip.
Price the peak. A small surcharge on Friday and Saturday evenings is standard practice now, and it funds the extra driver those nights require.
Taking the order: your own channel versus the marketplace
Where an order originates decides most of its economics, which is why the ordering channel is a profit lever rather than an IT decision. Customers are not resisting the switch either. Restolabs reports 67% prefer ordering from a restaurant’s own site or app, and digital orders average 23% higher than in-person ones.
The short version: keep a marketplace presence for discovery, treat its commission as an acquisition cost, and work to move repeat customers onto your own ordering page with package stickers, inserts and price parity. Your own channel also gives you the structured address, window and access-code data that makes a route plannable in the first place.
Getting that channel right takes work of its own: zone logic, minimums, order throttling, POS integration, and which fields the form has to capture. That has its own guide, online ordering for businesses that deliver.
The handoff: contactless drop-off is now the default
The last sixty seconds of a delivery are where the margin you protected upstream can still evaporate. A wrong-door drop is a refund. A bag that sat in the sun is a refund plus a review. A disputed delivery with no evidence is a refund you can’t contest.
Four things cover most of it: collect the unit number and drop spot at order time, seal the bag with a tamper-evident closure, photograph the placed order with the door in frame, and notify the customer the moment it lands. California has required tamper-evident sealing on ready-to-eat orders handed to third-party drivers since 2021, and other states have been following.
Temperature is the constraint that gets underestimated. USDA FSIS allows two hours in the 40°F to 140°F danger zone, one hour above 90°F ambient, and that clock includes drive time. The full set of rules, including driver procedure and when an order shouldn’t be left at all, is covered in the guide to no contact delivery.
What packaging and temperature control cost per order
Packaging is a small line that behaves like a large one, because it drives refunds.
Budget $1.50 to $3.00 per delivery order for a food business doing it properly: a leak-resistant container, a tamper seal, a label and a bag. Insulated bags and reusable cold packs are capital rather than per-order cost, and they pay back quickly, since a single remade catering order usually costs more than a season of hot bags.
Where operators overspend: branded custom packaging before the operation is profitable. Where they underspend: containers that survive a car. A cup that seals, a container that doesn’t sag when the sauce is hot, and a bag that stands up on a car floor are worth more to your reviews than a printed logo.
The five delivery metrics to track
Delivery revenue is a vanity number. These five tell you what’s happening.
- Contribution margin per delivered order. Order value minus food, packaging, channel cost and delivery cost. If this is negative on any channel, stop that channel or reprice it.
- Completed drops per paid driver hour. The density number. Target five or more.
- Cost per delivery, by channel. Marketplace, in-house and courier, tracked separately. Blended averages hide the channel that’s losing money.
- On-time rate against the promised window. Measured against what you told the customer, not against the order time.
- Refund and remake rate on delivery orders. Compare it to your counter rate. A big gap points at packaging or the handoff, not the kitchen.
Review them monthly, by channel. A delivery operation that looks fine in aggregate and is bleeding on one channel is the single most common pattern in this business.
Which delivery mix should you run?
A rough map, by where you are:
Under about 10 delivery orders a day, don’t hire a driver. Use a marketplace for discovery and a per-delivery courier partner for your own-channel orders. You can’t fill an hour of driving, so you shouldn’t be buying hours.
At 10 to 30 a day in a tight area, in-house starts to win. One driver on your densest shifts, courier partner for overflow, marketplace kept on for acquisition while you migrate repeat customers.
Past 30 a day, run in-house for the core zone and core hours, courier for peaks and the outer bands, and a hard look at whether the marketplace is still producing new customers or just charging you commission on regulars.
Catering and wholesale-style scheduled drops are a different animal entirely. Scheduled windows and repeat addresses mean density is achievable by design, and marketplaces are largely irrelevant.
Whichever mix you land on, revisit it twice a year. The volume that justified a driver in November may not survive February, and the reverse is just as common.
Frequently asked questions about delivery for local food businesses
Is it cheaper to deliver in-house or use a third-party service? In-house is cheaper per order once you can keep a driver at four or five drops an hour. Below that you’re paying for idle time, and a per-delivery courier fee or a marketplace is usually cheaper. The crossover is about density, not about order count alone.
How much should I charge for local delivery? Enough to cover driver time, mileage, packaging and payment processing on a typical stop, with margin left. For most local food businesses that’s somewhere between $4 and $9, banded by distance, with a minimum order two to three times the cost of the stop.
Do delivery orders actually make money? They can, and they reliably do on your own channel with decent density. On a marketplace at 30% all-in, a delivery order often contributes little more than covered overhead. That’s a fair reason to keep the channel for discovery, and a bad reason to build the business on it.
How far should I deliver? As far as you can reach in 15 to 20 minutes of real drive time, then price the outer edge separately or restrict it to scheduled days. Radius on a map is a poor proxy for cost; time behind the wheel is the actual expense.
Start with the arithmetic, then fix the route
Every profitable local delivery operation looks roughly the same underneath: a zone tight enough to cluster stops, a minimum and a fee that cover the trip, a direct ordering channel carrying the repeat customers, a marketplace earning its commission on discovery only, and a handoff that doesn’t generate refunds.
None of that requires new technology. It requires knowing what a stop costs you, and refusing to take orders that don’t clear it. Do the per-order math this week on your last thirty deliveries, split by channel. The answer usually tells you exactly which of the five levers above to pull first.
