Ask most operators how to increase revenue and the answer is more covers or higher prices. Both are hard. Covers are capped by your floor plan and your catchment; prices are the first thing guests compare and the fastest route to a complaint.
Average order value is the lever in between, and it is the one most restaurants leave alone because it is less obvious how to move it.
Guests decide most of what they will spend in the first ninety seconds of reading a menu. What sits at the top of a category, what has a photograph, what is described well, and whether the guest can read it in their own language all shape that decision before anyone takes an order.
This is the cheapest stage to fix and the most commonly ignored. Reordering a category so your highest-margin dish is not buried in position seven costs nothing.
A menu that only lists dishes leaves the selling to staff, and staff sell when they have time. On a busy Saturday — when the most covers are in the room — they have the least.
That is the coverage problem at the heart of upselling. Training helps, but it cannot fix a constraint that is structural. Recommendations built into the menu reach every table at the same standard regardless of how busy the floor is.
The other half of this stage is relevance. A suggestion that fits what the guest is already ordering reads as good service; a generic prompt reads as a pop-up.
A returning guest costs nothing to acquire and typically spends more, because they are not spending part of the visit working out what the place is. Retention is where average order value compounds rather than resets.
Most restaurant technology stops at the transaction, so the relationship resets after every meal. Anything that carries a guest's history forward — what they ordered, how often they come, what they spend — turns the next visit into a warmer one.
Measure first. Most operators know their busiest night but not their average spend per head by service, and not which dish is genuinely most profitable rather than most popular. Without that, every change is a guess.
Then work the stages in order. A loyalty programme attached to a menu nobody reads properly has nothing to work with.