Food Delivery Business: Complete Guide 2026

A food delivery business in 2026 looks nothing like one from 2018. DoorDash, Uber Eats, and GrubHub finished consolidating. Commission rates settled in the 18-30% range. Restaurants that started as 'delivery only' rediscovered margins by adding a pickup window. And independent operators figured out — through painful trial — that running on third-party platforms alone is a great way to lose 25 cents on every dollar.

This guide is for three audiences: existing restaurants adding delivery as a serious revenue line, ghost kitchen / virtual brand operators, and independent food delivery startups. The numbers, the math, and the software — without the influencer optimism.

Every restaurant operator should know exactly what each platform takes. The 2026 commission landscape (US, standard tier — rates negotiated for higher-volume operators):

- DoorDash: 15% (basic), 25% (preferred placement), 30% (premier — top placement + delivery radius extension). Optional ad spend on top. - Uber Eats: 15-30% depending on tier, plus marketing spend. - GrubHub: 15-30% standard, plus optional marketing fees that can stack to 35%+. - ChowNow: Subscription-only ($199-249/mo flat) — no commission. Limited consumer marketplace, mostly used as your own branded ordering layer. - Toast Online Ordering: ~$0 if you're already on Toast POS, or modest monthly fees with no per-order commission. Your own online ordering, no marketplace.

The math that bites operators: a $40 order on DoorDash at 25% commission nets the restaurant $30 before food cost. If food cost is 30% ($12), labor is 28% ($11.20), other variable costs (packaging, delivery zone fees, processing) take another $2-3. The restaurant is left with $4-5 of contribution margin on a meal that took the same kitchen labor as a $40 dine-in ticket worth $20+.

Delivery is not a bad business. But it requires a different pricing strategy, menu engineering, and a relentless focus on first-party orders.

The operating principle: use third-party for acquisition, convert to first-party for retention. Every DoorDash order should arrive with a flyer, a QR code, or a packaging insert offering 15% off the same order on your direct site. A 35% conversion rate of third-party-to-direct, over two years, is the difference between a delivery operation that drains cash and one that compounds.

Most operators set their delivery radius by guesswork. Don't. The math is straightforward and changes the business.

A driver's economic round-trip is roughly 6-10 minutes per mile in suburban traffic, 8-15 minutes per mile in urban. A driver doing 3-mile-radius deliveries with a 5-minute restaurant pickup turnaround and 5-minute customer dropoff completes about 3 orders/hour in suburban, 2-2.5 orders/hour in urban.

At $20/hr loaded driver cost, your cost per delivery is $6.67-10. If a typical order ticket is $35, your delivery cost is 19-29% of revenue. That number determines whether you can run in-house delivery profitably.

The radius decisions to make:

- Hot food integrity radius. Most food quality degrades materially past 20 minutes in a delivery bag. That implies a 2-4 mile primary radius in urban density, 3-6 in suburban. - Economic radius. Below 3 orders/hour per driver, in-house delivery loses money. Limit radius to keep utilization up. - Marketing radius vs delivery radius. You can be discoverable to people you won't deliver to (great — they become pickup customers). - Different radius by daypart. Lunch needs tight (people want food fast). Dinner can stretch further. Late night usually tightens again because of fewer drivers.

If you're running in-house delivery, drivers are your biggest hidden complexity.

Classification. W-2 employee vs 1099 contractor. The default for in-house drivers is W-2. If you classify drivers as 1099 incorrectly, you'll lose a Department of Labor audit and owe back wages + penalties. The Borello / ABC test rules in California (and similar elsewhere) make 1099 increasingly hard to defend.

Insurance. Hired non-owned auto coverage on top of the driver's personal policy. A standard personal auto policy excludes delivery use — meaning if a driver hits someone while on a delivery, the personal policy denies the claim. You must carry the business-side coverage. Budget $1,500-3,000/year for a small fleet.

Pay structure. Hourly + tips is cleanest legally. Per-delivery flat ($4-7) can be fine but is harder to defend against minimum wage requirements during slow shifts. Many operators do hourly base + per-order bonus + 100% tip passthrough.

Vehicle. You provide reimbursement ($0.60-0.70/mile in 2026 to cover gas, depreciation, maintenance, and insurance), or you provide the vehicle. Reimbursement is the norm at small scale; fleet vehicles start making sense at 6+ full-time drivers.

Routing. Even 4 stops is hard to optimize by hand. Use Onfleet, Circuit for Teams, or Bringg. Cost: $25-90/driver/month. Saves more than that in fuel and time the day you start using it.

Packaging is 4-8% of delivery revenue. Not negotiable, and often undermeasured. Decisions to make:

- Hot vs cold containers. Polystyrene (cheap, terrible for brand), molded fiber (compostable, $0.30-0.70/unit), aluminum tin + cardboard lid (works for hot items, scalable). Choose based on brand and unit economics. - Branded sleeves and stickers. A $0.12 branded sticker on every bag turns a third-party order into a marketing touchpoint. Single biggest first-party conversion lever. - Tamper-evident seals. Now expected by most customers. A simple sticker across the bag opening solves it. - Insert collateral. A small flyer with a QR code to your direct ordering site, a punch card, or a thank-you note. Inserts are how third-party customers become repeat first-party customers.

The most painful operational reality of delivery is the proliferation of dashboards. DoorDash Merchant Portal. Uber Eats Manager. GrubHub for Restaurants. Your POS. Your online ordering site. Your driver dispatch tool. Your CRM. Your loyalty platform. Your marketing email tool. Eight tabs open at any given moment.

The stack that works:

- POS that handles dine-in + takeout + first-party online + integrates with the third-party platforms. Toast does this well. Square for Restaurants does this acceptably. Deelo's POS app + Inventory + CRM + Bookings approach handles direct ordering deeply and lets the team operate from one workspace. - Order aggregator (optional but lifesaving). Otter, Cuboh, Chowly, or Deliverect consolidates the three third-party tablets into one stream that fires to your POS. $50-150/month per location. - First-party online ordering. ChowNow, Toast Online Ordering, Square Online, BentoBox, or a custom layer. Don't go without one — that is where you bring third-party customers home. - CRM with email + SMS marketing. Customers who ordered on Wednesday at 7pm should get an SMS the following Wednesday at 6pm. This is the simplest, highest-leverage retention lever in food and almost no independent restaurant does it. Deelo CRM + Marketing handles this natively. - Routing and dispatch. Onfleet, Circuit, or Bringg if you run drivers. - Analytics. A single dashboard that pulls in third-party performance + first-party orders + costs. Operators using one are dramatically better at margin discipline.

The Deelo pitch: CRM, Marketing, POS, Inventory, and Invoicing in one workspace, with the third-party order integration as a managed connector. For a single-location operator running both dine-in and delivery, having all of that under one per-seat fee meaningfully reduces both software cost and operational complexity.

If you're running a ghost kitchen or virtual brand (no dine-in), the rules differ in a few important ways.

Marketing is your entire customer acquisition cost. No walk-by, no signage, no neighborhood familiarity. Every customer comes from a third-party platform algorithm or paid search. Expect to spend 8-15% of revenue on marketing forever.

Menu engineering is more constrained. You're optimizing exclusively for delivery, no dine-in cross-subsidy.

Brand surface area on packaging is the only physical touchpoint. Spend on it.

Multi-brand strategy is the norm. Many ghost kitchens run 2-5 brands from the same kitchen (e.g., a 'wings' brand, a 'salads' brand, a 'comfort food' brand) targeting different search demand. This requires inventory and menu engineering that can support all of them with shared ingredients.

Real estate is the lever. A $2.50/sqft warehouse-zoned kitchen is the difference between a profitable ghost kitchen and a marginal one. Locate for delivery economics, not foot traffic.

Food delivery is not a money-printing machine. It is a high-volume, low-margin extension of an existing food operation that becomes profitable only when you treat it as a marketing channel that builds a first-party customer base.

The operators winning in 2026 are the ones with three things: (1) a delivery-engineered menu that travels and costs in line with channel economics, (2) a relentless first-party conversion strategy that turns DoorDash customers into direct-site customers within two orders, (3) a software stack that lets one operator run dine-in + takeout + first-party delivery + third-party delivery + driver dispatch without losing track.

Get those three right and delivery is a real business. Get any one of them wrong and you'll convince yourself, against the math, that 'volume will fix it'. It won't.