Show your menu to 100 people browsing a delivery marketplace and roughly 3 will order from you. Show the same menu to 100 people on a well-built ordering website and up to 15 will. Show it to 100 people who opened your branded app and up to 35 place an order. Same food, same prices, a 10× spread in the result. These figures come from more than 3M orders on our platform. Once you see why they split apart, they point you straight at where the next marketing dollar should go.
The benchmarks, and the caveat they need
Start with what "up to" means. The 35% and 15% figures are what well-run brands on our platform reach, not the median a new business hits on day one. A fresh ordering site with weak photos and a clunky checkout won't convert at 15%.
The bigger issue is that the traffic isn't the same temperature. Someone opening your app is almost always a returning customer who installed it after a good first order. Someone scrolling a marketplace is cold: they've never tried you, and your listing sits in a ranked feed beside forty competitors. So comparing the raw rates partly means comparing warm traffic against cold. Anyone who quotes these numbers without saying that is selling you something.
That caveat is really the strategy. The three channels do different jobs. Marketplaces reach people who don't know you yet. Your site closes the people you've already convinced. The app is where your regulars settle in. Using all three makes sense. What burns money is paying an acquisition-level price at every stage, including the one where the customer already belongs to you.
An app converts at 35% because the decision was already made
Nobody installs a restaurant app just to look around. Installing it is the commitment: the customer has already decided you're one of their regular places. After that every step is shorter than anywhere else. Address saved, card saved, order history one tap away, a reorder done in three taps. A push at 6:10pm lands right when dinner is still an open question.
Selection bias does part of the work here, and that's the point of it. A branded ordering app filters for the people most likely to buy, then strips out every reason not to. Which is why a delivery brand's repeat business belongs in the app rather than in a channel that charges 25–30% to re-sell you your own regular.
Bring your numbers to a 30-minute call. We'll benchmark them against brands your size.
Websites: 15% is intent, 3% is friction
Website traffic sits in the middle. People arrive from Google Maps, a "sushi near me" search, or the link in your Instagram bio. They already want food and they've already picked you as a candidate. That intent is worth 10–15% conversion when the path to checkout cooperates.
Most restaurant sites throw that intent away: a PDF menu that takes eight seconds to load, an "Order now" button that dumps the customer onto a marketplace where competitors sit one scroll away, a checkout that demands account creation before it will even show the delivery fee. That's how a site full of buying-ready visitors ends up at ~3%, level with a cold marketplace listing. An ordering website built as a store rather than a brochure is the whole distance between those two numbers.
The marketplace's ~3% is working exactly as designed
Inside a marketplace the platform itself converts almost every visitor: someone opens the feed hungry and orders from somebody. Your ~3% slice of that is a ranked-feed lottery of placement, ratings, your markup against the neighbours, and how much visibility you paid for. The platform is tuned to maximise its own total conversion, and your share of it is incidental. So a marketplace works well as a way to get discovered and badly as a way to keep customers. We ran the full numbers on that in what delivery apps really take from every order.
How to use these benchmarks without fooling yourself
Treat benchmarks as a way to diagnose your funnel, not a leaderboard to win. A few things keep them useful.
First, measure your own funnel before you compare it to anyone else. Sessions to carts to paid orders, split by channel, over at least 30 days. Reports and analytics should give you this without exporting to a spreadsheet.
Second, compare against the right band rather than the headline. Cuisine, ticket size, and city density all move the bands around. A 12% site conversion can be perfectly healthy for a $60-ticket steakhouse and a warning sign for a $14 poke shop where 18% is the norm. The trend matters more than the absolute number: if your site climbed from 6% to 9% in a quarter, you're doing the right things.
Third, fix conversion before you buy traffic. A site converting at 5% means ads triple your waste rather than your revenue. Better photos, faster loading, guest checkout, and an honest delivery fee all cost less than clicks.
Do this next
Pull 30 days of data and build one table: visitors and orders for your site, your app if you have one, and each marketplace you're on. If your owned channels are sitting near the bottom of the bands above, the quickest revenue you'll find this year is conversion work, not a bigger ad budget or another marketplace.
Branded app and ordering site with full-funnel analytics — launched in about two weeks.