Which Delivery Apps to Keep and Which to Drop

Most owners talk about "the apps" like they're one thing. They aren't. One of them might be bringing you people who have never heard of you. Another might be charging you thirty percent to deliver food to the guy who lives above your shop and used to call the landline. Same commission, completely different value.

So stop judging delivery as a category and start judging each platform as its own vendor. Four numbers do the job. You can pull all of them out of the merchant dashboards you already have logins for, and you can do it in about an hour on a slow afternoon.

Number one: how many orders it actually brings

Start with raw order count over the last 90 days. Not sales dollars — order count. Dollars hide things. A platform with a high average ticket but only a handful of orders a week isn't a channel, it's a rounding error with a tablet on your wall.

Write the number down per platform, per week. Look at the shape too. Is it flat, growing, or has it been sliding since the last time they changed their fee structure? A platform that's been drifting down for six straight months isn't going to turn around because you kept it.

And be honest about the tablet. If one platform sends you four orders a week but your staff has to watch a fourth screen every shift, you're paying for that in attention during the rush, which is the most expensive thing you own.

Number two: your real commission rate on your plan

The headline percentage is not what you pay. Every platform has tiers, and most restaurants are on a plan they chose two years ago and never revisited. Marketing credits, promo participation, delivery radius upgrades, the "we'll boost your listing" programs — those come out of the same pot.

Do this instead of reading the rate card: take your payout statement for a full month. Take the gross customer-facing subtotal for the same month. Divide what you kept by what the customer was charged for food. That's your real rate on that platform. It's almost always worse than the number you have in your head, and the gap between platforms is usually bigger than owners expect.

If you want the full breakdown of where the money leaks — packaging, the menu markup you may or may not be charging, the orders that get refunded — we walked through it in why your delivery app orders cost more than you think. Do that math once per platform before you make any decision.

Number three: new customers versus your own regulars

This is the number that actually decides which app stays.

A delivery platform is worth its commission in exactly one situation: when it puts your food in front of someone who would never have found you. That's a customer acquisition cost. Expensive, but it buys something.

It is worth nothing when it re-serves people who already know your name. If a customer has eaten your food, likes it, and is ordering again, you are paying a finder's fee to find someone you already found. That's the whole argument.

Most dashboards report some version of "new customers" or "first-time orders." The definitions vary and none of them are perfect — a new-to-that-platform customer may well be an old regular of yours who switched. That's fine. You're comparing platforms against each other, not chasing precision. Pull the share of first-time customers for each one over 90 days.

Then add the thing the dashboard can't tell you: ask your staff. Front-of-house knows the repeat names. If the same six addresses account for a big chunk of one platform's volume, that platform is a very costly phone line.

Number four: refund and adjustment rate

Pull the adjustments report. Refunds, order errors, cancellations, customer credits charged back to you. Express it as a percentage of orders on that platform.

This number separates platforms more than people realize, because dispute policies differ. Some platforms refund a customer on request and debit you without a conversation. Some actually look at it. Some have driver pools that lose orders more often in your neighbourhood.

A platform with an elevated refund rate is costing you twice: the food, and the review that comes with a bad experience you didn't cause. Add that rate to your real commission rate from number two. Now you've got the true cost of that channel.

Lay the four numbers side by side

One row per platform. Orders per week, real commission rate, first-time customer share, refund rate.

The pattern is usually obvious once it's written down. One platform does most of your volume at the worst real rate and mostly serves people you already know. One does less volume but a healthy share of it is new names. One is barely there at all.

The decision follows:

• Keep the one that finds new people. Even at a painful rate. That's marketing spend with a receipt.
• Drop the one that mostly re-serves your regulars. You're paying rent on your own customer list.
• Drop the one with trivial volume. It's a screen, a login, a menu to update, and a source of errors. Cut it.
• If one has a bad refund rate and no new customers, it goes first. No argument needed.

Before you delete anything, pause it

Don't make this a permanent decision on the first pass. Most platforms let you go offline or pause your store. Turn the candidate off for two full weeks, including two weekends, and watch what happens to the others.

Sometimes the orders move. The same customers show up on the platform you kept, or on your own ordering page, and you keep more of each one. That's a clean win, and it tells you that platform was never bringing anything — it was just splitting the volume you already had.

Sometimes they don't move, and you feel it. Turn it back on and reassess. Two weeks of data beats a year of guessing.

Before the pause, give the customers somewhere else to go. Printed cards in every bag. A note on the receipt. Your ordering link in your Google profile and your Instagram bio. The sequence for doing this without a volume dip is in move your regulars off delivery apps, in order — run that play for a few weeks first, then pause.

What to do with the channel you keep

Treat the surviving platform as what it is: a place strangers try you for the first time. That changes how you use it.

Trim the menu you list there. Dishes that travel well, items with margin that survives the commission, nothing that arrives sad after twenty minutes. The app menu does not have to be your full menu, and a shorter one is easier to keep accurate.

Put something in every bag that makes the second order come direct. A card with your ordering link, a small offer, a QR code that goes straight to your own checkout. The platform charges you to acquire the customer. You only have to pay that once if you make the handoff.

And make sure the place you're sending them actually works — fast menu, no PDF, orders that land in the kitchen without someone retyping them. That's the whole point of an ordering setup you own. The apps are rented traffic. Your own page is the one that keeps the customer's name and phone number.

Rerun it twice a year

Fee structures change. Plans get restructured. A platform that was feeding you new customers can quietly turn into a regulars-delivery service as your neighbourhood catches on to you.

Put it on the calendar for January and July. One hour, four numbers, three rows. That's the whole review. It pairs well with the Monday check described in the three numbers to check every Monday morning — weekly for operations, twice a year for the channels you're renting.