The final two minutes of a lesson can tell a teacher what to do first on Thursday. One short response from every learner often reveals more than a general impression formed while the confident speakers were answering aloud. An exit ticket earns classroom time when its results alter the next lesson.
This article is for a teacher of four A2 learners who wants individual evidence without adding another graded test.
Ask for one observable performance
“What did you learn today?” produces memories, feelings and polite summaries. Those answers may matter, though they rarely identify the next teaching move. A useful ticket asks learners to perform the lesson target in a form the teacher can sort quickly. After a lesson on past questions, the prompt might be: Write one question you could ask a colleague about yesterday, using did. After a listening lesson: Choose which of two callers changed the appointment, then write the four words that gave you the answer. A pronunciation ticket can ask learners to mark the stressed word in one contrastive reply.
Keep it small.
One prompt should take 90 to 150 seconds. The teacher needs all four responses, including an explicit “I cannot do this yet.” Names are useful when the next task will differ by learner. Anonymous tickets suit a question about comfort or lesson pace, where identity adds little.
Choc Education's article on observable comprehension checks deals with evidence gathered during a lesson. An exit ticket has a different job: it survives after class and informs planning.
Evidence depends on what follows
Black and Wiliam's widely cited review of classroom assessment examined evidence on formative assessment and argued that assessment becomes formative through the use made of its evidence. A collection of slips stored in a folder has completed the measurement step. Teaching changes only when the answers affect grouping, explanation, examples or practice. This distinction protects the ticket from becoming routine paperwork. If Thursday's plan will remain identical under every possible pattern of answers, the Wednesday prompt has no decision attached to it. Drop it or change the question.
A 2026 quasi-experimental study in Taraba State assigned 240 secondary-school English students through intact groups and reported stronger achievement for classes using exit tickets than for conventional instruction. The published study used a non-equivalent control-group design, so assignment did not fully remove pre-existing class differences. Its setting and broad English achievement measure also prevent a direct claim about a four-person online class in Taiwan. It offers supporting evidence, rather than a guarantee. Research on university marketing courses adds a useful limit. MacDermott and colleagues found that exit tickets were associated with deeper engagement and greater use of course resources, while extended study time and overall motivation changed little. Their 2025 peer-reviewed report relied partly on student reports and came from another subject. A tiny end task will not create an evening study habit on its own.
Sort answers into the next action
Read the four tickets against a decision table prepared with the prompt. When two learners choose the right form for different reasons and the other two repeat the same error, the opening task should preserve the successful reasoning while giving the shared error one fresh example and another attempt.
| Pattern in the answers | Opening move next lesson |
|---|---|
| Four accurate responses | Start with a new context that uses the same form |
| Two or three share one error | Show two examples and let pairs repair the error |
| One learner has a distinct error | Give that learner a 3-minute adjusted task during pair work |
| Most responses are blank or off target | Re-teach with a different representation before adding difficulty |
The table keeps diagnosis connected to action. It also limits overreaction. One wrong article in an otherwise clear past question does not justify repeating a 40-minute unit. The teacher can mark the article for later and respond to the target actually sampled by the ticket. I prefer sorting into three piles over assigning marks: secure, partial, and needs a new explanation. A numerical score such as 7 out of 10 implies precision that a one-item sample cannot carry.
Return the evidence to learners
Thursday should open with visible continuity. The teacher might say, “Three answers used did with a past-tense verb, so we will repair that pattern before today's interview.” No names are needed. Show two anonymised sentences, ask pairs to choose a repair, and run a fresh question 6 minutes later. Learners then see that the ticket affected the lesson. Without that return, a request for honest evidence can feel like a miniature exam that disappears into the teacher's files. Individual feedback can remain brief. Circle the exact point sampled and add one instruction: Try this again with go, or Listen once more for changed to Friday. Avoid comments about effort, talent or general English ability. A 12-word response supports a local inference.
Protect the last minute from noise
Exit tickets fail when the room is already packing up. In an online class, learners may be finding the Leave button while instructions are still being spoken. Put the prompt on screen before the final activity ends, reserve two minutes in the lesson plan, and stop accepting new teaching points once the timer starts. The same prompt should not appear every week. A retrieval response, a choice with one line of evidence, a 14-second voice note and a sentence repair sample different abilities. Match the form to the lesson target. Voice notes are a poor choice after a reading lesson. Written sentences miss turn timing after conversation practice.
There is one awkward edge. A learner may understand the target and misread the ticket. If an answer conflicts with everything seen during the lesson, treat it as a signal to check. Withhold the verdict. Start the next class with a second example before assigning remediation.
For the next lesson, write the exit prompt beside the decision it controls. Collect one response from each learner, sort all four in under 5 minutes, and use the pattern to choose Thursday's first 6-minute task.