Café Menu Clarity: Test What Customers Understand

Define one question the menu must answer

Choose a question that could change a label, description or grouping. For example: can a person identify what is included in a breakfast set without a member of staff explaining it? This is narrower than asking whether the whole menu is good.

Record the exact printed revision or digital screen, language, device or paper size, and service occasion. Define an observable completion criterion before the session. For the fictional test below, the person must select the intended set and correctly describe its included toast and small coffee without a hint.

Check the test menu against the approved recipe, portions and prices first. A clarity exercise cannot establish whether required food information is correct. Refer ingredient, allergen, nutrition or other regulated information to the responsible reviewer rather than asking participants to validate it.

Give a realistic task without teaching the answer

Use a situation someone in the intended audience could recognise. An original prompt might be: “You want breakfast for one person with coffee and toast. Use this menu to show what you would order, and explain what you expect to receive.” The facilitator’s answer key stays out of view.

Do not point at the set, read its description aloud or tell the person which heading to look under. Invite them to use the menu as they normally would. Record their selection and explanation before asking follow-up questions.

If the person asks a question that a server would normally answer, record the question and any help given. Assistance is useful evidence about the interaction between menu and service. It must not be silently counted as independent completion.

Keep a record for every selected session

Use anonymous participant references and explain the purpose of the exercise. Record relevant experience, such as whether the person has seen this menu before, without collecting unnecessary personal details. Recruit around the service occasion and include people beyond the team that wrote the menu.

Use one first attempt per person for the stated task and version in a cohort. Keep a different task, language, screen format or returning-participant group separate when it changes the interpretation. The worksheet detects duplicate references within its register; it cannot detect the same person entered under a new reference.

Give each session one outcome: Independent, Assisted, Not completed or Not attempted. A person who never saw the menu because the session was interrupted is Not attempted. A person who tried but did not meet the defined criterion is Not completed. Retain the observation behind the label.

Worked example: five sessions, four attempts

In this fictional cohort, the approved “Morning Pair” contains toast and a small coffee. Everyone receives the same task and menu revision. The facilitator records the first attempt, including help, before discussing improvements.

Read the denominator before the percentage

There are five recorded sessions and four attempted tasks. Two were completed independently, one with assistance and one was not completed. Independent completion among attempts is 2 ÷ 4 = 50%. The unattempted session remains visible and is excluded from that denominator.

This percentage describes the selected first attempts under one criterion. It does not measure the share of all customers who understand the menu, prove an increase in orders or establish a sample-size requirement. Read the actual misunderstanding and the assistance beside the counts.

The worksheet shows counts only after the register is confirmed complete and its rows are valid. An empty confirmed register shows zero counts and a blank rate. An unconfirmed register leaves the summary unavailable. Missing outcomes are never treated as successful attempts.

Change the specific source of confusion

Link the proposed revision to the observed problem. For the fictional toast-only selection, the owner might test clearer separation between the standalone toast item and the breakfast set, together with an explicit list of included items. Keep the approved portion and price promises unchanged unless those receive their own review.

Save the original menu, observation record and decision. Give the revised menu a new version and repeat the task in a separately labelled cohort. Returning participants may remember the answer; record that exposure rather than combining their later attempt with the original first attempts.

Review the new observations before releasing the revision, then listen for the same questions during service. Different cohorts, conditions and learning can affect the results. A higher completion share by itself does not establish that the redesign caused the change.

What usability guidance supports

Nielsen Norman Group’s task-scenario guidance supports realistic, actionable activities that avoid giving away the steps or answer. This guide adapts those general usability principles to an original café menu exercise. The menu example, outcome categories and worksheet are Keystonne Grow’s teaching model, not a quoted research finding or an industry benchmark.