Most delivery businesses fail for a boring reason: they launched with too much geography and too little supply. The software is the part everyone worries about, and it is the cheapest, fastest part to solve. What actually decides the outcome is density — enough restaurants and enough riders inside a radius small enough that a delivery takes 20 minutes, not 50.
1. Pick one zone, not one city
Draw a 3–4 km radius around a dense residential or office cluster. Count the restaurants inside it that already do takeaway. If the answer is under 25, the zone is too thin to launch; find a denser one. A platform with 30 restaurants in one neighbourhood feels alive. The same 30 spread across a city feels empty.
Your delivery promise should be a number you can hit on a rainy Friday: 30 minutes, not 20. Under-promise and over-deliver in the first month, because the first 500 customers decide whether there is a second month.
2. Recruit supply before demand
Restaurants say yes to a new platform for one of three reasons: no upfront fee, a lower commission than the incumbent, or a delivery service they cannot run themselves. Lead with the first two and be specific about the third.
What you need from each vendor on day one: menu with prices and photos, prep time per item category, opening hours, and one person who will answer the tablet. A vendor who will not answer the tablet in the first week will not answer it in month three.
Do not sign exclusivity. You do not have the demand to justify it, and it makes recruitment hostile.
3. Decide the delivery model honestly
Three options, and the right one changes as you grow:
- Vendor self-delivery — cheapest to start, inconsistent quality, and it excludes restaurants that do not want to run riders. Good for a pilot.
- Platform riders — the real product. You control the experience, but you carry fixed cost per shift whether or not orders arrive.
- Hybrid — platform riders in the core zone, self-delivery outside it. The usual end state.
Rider economics are simple and unforgiving: a rider needs to clear roughly one delivery every 12–15 minutes during a shift to earn a wage worth turning up for. If your zone cannot supply that density, the fleet will quit before the customers notice.
4. Set commission from your unit economics
Work backwards. Take a $10 average basket in a zone where a delivery costs you $2.50 all-in (rider time, fuel, support). If you charge 15% commission you collect $1.50 — you are subsidising every order. Either the basket has to be bigger, the delivery fee has to carry more of the cost, or the commission has to be higher.
Most new platforms start at 15–20% plus a delivery fee paid by the customer, and move to 25–30% only once they are delivering volume the vendor cannot ignore. Announce changes with notice; surprise commission changes are how you lose your best restaurants in a week.
5. Get the ordering stack right
You need four things on day one: a storefront customers can order from, a kitchen interface vendors will actually watch, a rider app, and an admin view of every live order. That is the minimum. Everything else — loyalty, wallets, referrals — is a month-two problem.
You can build that stack for $25,000+ and six months, rent it for a few hundred dollars a month, or install a productised platform such as Foodomaa for a one-time $299. The strategic question is not which is cheapest; it is whether you want to spend your first six months managing a development team or recruiting restaurants.
6. Launch small and measure three numbers
Ignore downloads. Track order frequency per customer, vendor rejection rate, and delivery time from "ready to "delivered". The first tells you whether the product is worth repeating, the second whether supply is reliable, the third whether your radius is honest.
A launch with 30 restaurants, 6 riders and 200 orders in the first week is a real business. A launch with 300 restaurants, 40 riders and 90 orders is a expensive logo.
7. Keep the customer relationship
The aggregator model works because the customer belongs to the platform, not the restaurant. If you are the platform, that is your advantage — use it. Own the app, own the push notification, own the reorder button. If you are a restaurant building your own channel, the same rule applies in reverse: the regulars you move to your own ordering app are the ones that stop paying commission.
Launch order, summarised
- Choose one dense zone and validate restaurant count.
- Sign 25–40 vendors with menu data and a named contact.
- Pick the delivery model you can actually staff.
- Set commission and fees from unit economics, not from what a competitor charges.
- Install the ordering stack and place ten test orders yourself.
- Launch to one neighbourhood, measure frequency, rejection rate and delivery time.
- Expand the radius only when the first zone is boring.