Most restaurants test menu knowledge with a written quiz handed out on a training day, marked by nobody, and never mentioned again. It produces a sheet of paper and no change in what happens at tables.
Testing is worth doing. It is one of the few management interventions with a solid evidence base behind it, because testing is not just measurement, it improves retention. But the format most restaurants use gets almost none of that benefit.
Why the written quiz fails
Three reasons, and they are all about the gap between the test and the job.
It tests recognition. Multiple choice lets people identify the right answer among four, which is a much easier task than producing an answer to a guest who asked an open question.
It tests the wrong direction. Quizzes tend to ask “which dish contains walnuts”, while guests ask “does the salad have nuts in it”. Practising in the wrong direction transfers poorly.
It tests silently. The job is spoken. A server who can write down that the bream comes with fennel may still be unable to say a clean sentence about it while a table watches.
Add that written quizzes are usually one-off, and you have an exercise that measures nothing useful and teaches nothing either.
| Written quiz | The actual job |
|---|---|
| Multiple choice | Open question from a guest |
| Dish to ingredient | Ingredient to dish, usually |
| Written | Spoken, under mild pressure |
| Once, on a training day | Every table, every night |
| Marked in private | Answered in public |
What a good test looks like
Spoken, open, short, frequent, and in the direction a guest would ask.
That is it. In practice it means a manager saying “describe the lamb” and listening to the answer, thirty seconds, a few times a week, which fits inside a pre-shift meeting you already hold.
Ask open questions, not closed ones. “What is in the dressing” rather than “does the dressing contain anchovy”.
Ask for the answer they would give a guest, not the technically complete one. The test should produce the sentence you want at a table.
Ask in front of other people, which sounds harsh and is the reason it works: everyone present rehearses silently, so one question tests one person and teaches ten.
The five questions that cover a menu
You do not need a question bank. Almost everything worth knowing sits behind five question shapes.
Describe this dish in one sentence. Tests the core skill of a spoken description.
What comes with it. The single most complaint-preventing fact on any menu, and the one most often unknown.
Does it contain X. Allergen recall, asked the way a guest asks it.
Which would you recommend for someone who wants something light. Tests judgement rather than memory, which is what answering a recommendation request actually needs.
What is the special tonight. Tests whether anyone listened at the briefing.
Rotate those five shapes across the dishes and you have covered a menu properly without ever writing a quiz.
| Question shape | What it tests |
|---|---|
| Describe this in one sentence | Production under mild pressure |
| What comes with it | The most-missed fact on the menu |
| Does it contain X | Allergen recall, guest-shaped |
| What would you suggest for X | Judgement, not recall |
| What is on tonight | Attention at the briefing |
Frequency beats depth
A thirty-second question three times a week produces far more than a forty-minute exam once a quarter, because distributed practice outperforms massed practice across essentially every study that has compared them.
This also solves the scheduling problem that kills most restaurant training. Nobody has to be rostered for it. It happens inside a meeting that already exists.
And it changes behaviour before the test rather than after it. Once a floor knows a question is coming most days, people arrive having looked at the menu, which is the actual goal.
How to test without making it unpleasant
The difference between a floor that likes this and one that dreads it is entirely in the manager’s handling.
Ask everyone, including yourself. A manager who occasionally answers a question from a server changes the tone completely.
Accept “I do not know” without comment, then answer it and move on. Punishing a wrong answer teaches people to bluff, which is the opposite of what you want at a table.
Never use it to single out the person who is struggling. Rotate genuinely.
Praise the specific thing that was good in the answer, briefly. “That is the sentence, exactly that” is worth more than general encouragement.
And keep it fast. The moment it becomes a five-minute discussion, it stops being a daily habit.
Testing new starters
New starters need a different cadence, not a different method.
Ask them daily for the first two weeks, but ask about the dishes they have already covered rather than the whole menu. A test on material nobody has taught is a demoralising experience with no learning value.
Test the direction of the job from day one. Dish to description, dish to allergens, guest question to answer.
Let them use their own cards while learning and then without. Practising with the deck is study; practising without it is the test, and a deck built from a photo of the menu makes the study half take minutes rather than an evening.
Set a clear finish line: by the end of week two you can describe every main, name what each comes with, and answer allergen questions from memory or say you will check. That is a target people can hit, unlike “know the menu”.
What to do with what you learn
The point of testing is to find out what the floor does not know, which means acting on it.
Two people missing the same fact is a training gap, not an individual problem. Fix it at the next briefing.
Everyone missing what a dish comes with usually means the menu is unclear, and the better fix is a change to the printed menu rather than more drilling.
Repeated allergen gaps are the one thing to treat as urgent rather than as a training theme, since providing accurate allergen information is a legal duty for food businesses and not a service nicety.
And if the same person struggles across weeks, that is a conversation to have privately, with support rather than with a quiz.
Testing knowledge that is not the menu
The same method applies to the rest of what a floor needs to know, and these areas are tested even less often than the menu.
Table numbers and seat positions. Ask a new starter to name the tables in a section from memory, or to say which seat is one. This takes fifteen seconds and catches a problem that otherwise shows up as plates being called out across a table.
The wine and drinks list. Ask for a pairing rather than a fact, because a pairing tests judgement and a vintage tests nothing useful.
Procedures. What to do with a returned dish, how to handle a card that declines, who to tell about a spill. These get taught once and then assumed forever.
Guest language. “A guest says the pasta is bland, what do you say?” This is the closest thing to a real test of service, and almost nobody asks it.
| Area | A question that works |
|---|---|
| Floor plan | Name the tables in your section |
| Wine | What would you pour with the lamb? |
| Procedure | A card declines, what happens? |
| Guest handling | A guest says it is bland, what now? |
| Specials | What is on, and what does it come with? |
Why testing feels harsh and is not
Managers avoid this because it feels like putting people on the spot, and that instinct deserves a straight answer.
Being asked a question you cannot answer in front of colleagues is mildly uncomfortable. Being asked a question by a guest that you cannot answer, at a table, while they wait, is considerably worse, and it happens to untested floors constantly.
The choice is not between testing and no testing. It is between a controlled thirty seconds where a wrong answer costs nothing, and an uncontrolled moment in front of a paying guest where it costs the table.
Framed that way, most staff prefer the briefing question, and saying the framing out loud once tends to end the resistance.
What to do about the person who always knows
Every floor has one or two people who answer everything correctly, and they are an underused resource rather than a scoring problem.
Ask them the harder version. “What would you pour with that” rather than “what is in it”.
Let them ask the question sometimes. A server asking the floor to describe a dish changes the dynamic and works just as well.
Use them to set the standard by going first. Hearing a good answer before being asked gives everyone else a target, and it makes the format feel like practice rather than examination.
Spacing still matters for them too, because knowledge decays without repetition even in people who learned it well the first time, and memory holds fewer items at once than most people assume once a menu grows past a certain size.
Testing yourself as a server
None of this has to wait for a manager who runs it well, and plenty do not.
Ask a colleague to fire five dishes at you before service and answer out loud. It takes two minutes, it is more useful than reading the menu again, and the colleague benefits from hearing the answers.
Cover the menu and describe each main from memory, then check. The dishes you stumbled on are your list for tomorrow, and there are usually only three.
Test in the direction guests ask. Given a dietary requirement, name the dishes that work. Given a mood, name two recommendations. Those are the questions you will get, and practising the reverse direction transfers poorly.
Do it on a break rather than at home, because short repeated sessions during the week beat one long one on a day off, and because a break is time you were spending anyway.
Frequently asked questions
How should managers test menu knowledge?
Spoken, open questions asked in front of the team, thirty seconds at a time, several times a week. Ask people to describe a dish in one sentence rather than to identify a fact from four options, because the job is spoken production under mild pressure and a written multiple-choice quiz tests neither.
Are written menu quizzes worth using?
Rarely. They test recognition rather than recall, usually ask in the opposite direction from how guests ask, and are silent while the job is spoken. They also tend to happen once and never again, which throws away the main benefit of testing, since repeated low-stakes testing is what actually improves retention.
How often should staff be tested on the menu?
A short spoken question a few times a week beats a long test once a quarter, because distributed practice retains far better than massed practice. Fitting it inside the pre-shift meeting removes the scheduling problem that stops most restaurant training from happening at all.
What questions should I ask?
Five shapes cover a menu: describe this dish in one sentence, what does it come with, does it contain a given allergen, what would you suggest for a guest who wants something light, and what is on tonight. Rotate those across dishes rather than building a question bank.
How do I test people without it feeling like an exam?
Rotate genuinely, answer questions yourself sometimes, accept “I do not know” without comment and then give the answer, and keep each question under a minute. Punishing wrong answers teaches staff to bluff at tables, which is a worse outcome than the gap you were testing for.
