Spring til indhold

Restaurant Analytics

Frequently Bought Together: Raising Restaurant Order Value

"Frequently bought together" means dishes that appear in the same order more often than chance would suggest — a burger and fries, a coffee and a pastry. Restaurants raise order value by finding those pairs in their own orders and offering the second dish at the moment the guest chooses the first. It works when the suggestion is useful to the guest, not when it is a push.

Finding the pairs in your own orders

  1. Take orders with more than one dish from a few complete weeks.
  2. For each pair of dishes, count the orders that contain both.
  3. For each dish, work out how often its companion appears with it: orders with both, divided by orders with the first dish.
  4. Ignore pairs that appeared only a handful of times — a pattern from three orders is noise.
  5. Keep the top two or three companions per dish; more than that is rarely useful.

Direction matters. Most orders with fries may contain a burger, while only some burger orders include fries. The useful suggestion is the one shown next to the dish guests choose first.

Obvious pairs and surprising ones

Many pairs confirm what staff already know. The useful ones are the surprises: a dessert that goes with one particular main, a drink that sells with breakfast but not lunch, a side that guests add to a salad. Those are where a suggestion on the menu can change behaviour, because guests may not think of the combination themselves.

Ways to use them

Turning a pair into a suggestion
ApproachWhereWorks best for
A "goes well with" lineOn the dish pageDrinks and sides
A suggestion in the basketBefore the order is sentDesserts, extras, a forgotten drink
A combination on the menuIts own item or sectionPairs that sell together very often
Staff promptAt the table or counterVenues where guests order from staff

Suggest, do not push

  • Only suggest dishes that are available right now.
  • Keep it to one or two suggestions; a long list is ignored.
  • Respect what the guest has told you — do not suggest alcohol where the venue serves none, or a meat dish next to a vegetarian choice.
  • Make declining as easy as accepting.
  • Check the effect: compare average order value and the share of orders with the suggested dish before and after.

How to read the number this changes: average order value.

Dishes ordered together in Thmenu

Thmenu's menu performance report counts which dishes guests order together in your own orders and shows, for each dish, up to three dishes most often ordered with it — once the period holds enough orders with more than one dish; below that it says there is not enough data yet. On the guest menu, a dish page can show drinks that pair well with it, chosen from your own available drinks, and while ordering is open the basket can suggest other available dishes from your menu; when AI is not available, the basket uses simple rules by category. Guests add a suggestion with one tap or ignore it.

See restaurant analytics in Thmenu.

About the author

Thmenu Team — Product team at Synaltix LLC, the company that builds Thmenu. The Thmenu team builds menus, ordering and payment tools for restaurants and cafés. Articles describe how things work in practice and, where Thmenu is mentioned, what the product does today.