Oral Placement Test — eighteen tasks that show what a learner can actually say
Three tasks per CEFR level, A1 through C2, in a fixed ladder: greetings and introductions at the bottom, global strategy at the top. Each task carries three prompts for the candidate and three questions for the examiner, plus the grammar it is really testing — Present of Be at task one, Past Modals and Mixed Conditionals at fifteen. Fifty-four prompts, fifty-four observation points, no multiple choice anywhere.
Why a speaking test and a grammar test disagree
Most placement is done with a written grammar test, and then the school is surprised when a "B2" student cannot order lunch. The two measure different things: recognition versus production. This test only observes what the candidate produces under mild pressure, which is the thing that predicts how they will cope in a speaking class. It is slower than a multiple-choice sheet and it does not need marking.
The examiner questions are the actual rubric
Under each task sit three questions aimed at you, not the candidate — does the candidate use "be" correctly?, is the pronunciation clear?, can they answer where they are from?. These turn an impression into an observation. Without them, two teachers testing the same student will disagree by a whole band; with them, they are looking at the same three things and usually land in the same place.
Climb until it breaks
Do not start at the level you expect. Start one band lower, so the candidate's first minute is comfortable and they stop being nervous, then move up. Their level is not where they answer correctly — it is where the answers get shorter, the pauses get longer, and they start translating visibly. That is the ceiling, and the band below it is where they should be taught. Stop as soon as you see it; pushing two more levels only demoralises.
Report a profile, not a number
A learner is rarely one level. They can be B1 in fluency, A2 in accuracy and B2 in interaction, and the average of those three describes nobody. Write down all three. Then tell the student in plain words: "you keep talking even when you are missing a word, and that is a real strength; your past tenses fall apart as soon as you speed up, and that is what we will work on." That sentence is more useful to them than a letter and a number.
Running it with a full class
The test is one-to-one, which is the practical obstacle. Two workable arrangements: test individually in a corner while the rest of the class works through a written task or one of the other tools here; or have students record their answers on their own phones and listen afterwards, which loses the follow-up questions but scales to thirty students in a single lesson. Either way, budget six to eight minutes per candidate for a full ladder.
Re-test with the same instrument
The value compounds when you run it again. Use the identical eighteen tasks at the end of the term. Students recognise the questions, which sounds like a flaw and is the point: the comparison is clean, and they can hear the difference themselves. A learner who stalled at task seven in September and reaches task twelve in January has evidence of progress in their own voice, which no grade delivers as convincingly.
Goes well with
Further reading
CEFR Levels Explained — what A1 to C2 actually mean in a classroom
Error Correction in Speaking — when to interrupt and when to wait