One test separates a Task 7 answer with reasons in it from ninety seconds of confident noise. Could a reasonable person disagree with this sentence? If not, it is not a reason. "It would be better for everyone" fails that test. Every sentence in the answer below survives it.
The failure
You state your opinion. Then you state it again in warmer words. Then you add an example that turns out to be the same claim wearing a hat. Ninety seconds gone, one thing said three times, fluently. Content/Coherence and Task Fulfillment are two of the four published rating dimensions, and an answer with one idea in it is thin in both at once.
The same failure hides inside three reasons that are three angles on one idea: it saves money, it's cheaper for the city, taxpayers pay less. Real reasons sit on different axes — one about cost, one about fairness, one about what people will actually do.
And "it depends" is not a position. It describes the question instead of answering it. A middle position is fine, but it needs edges — it must rule something out. "I'd support this, but only if X stays" is a position. "There are good points on both sides" is not.
The task
Task 7 is "Expressing Opinions" — the seventh of the eight Speaking tasks, in a component that runs 15 minutes. You get 30 seconds of preparation and 90 seconds of speaking. Only Task 1 gives you that long — every other task gives 60. Aim for roughly 200–240 words — an answer that lands at 130 is half-empty, not crisp. Those clocks come from Paragon's Speaking study pack, not the test-format page, which publishes the task names but not the timings.
The prompt
(original AlmiWorld practice prompt)
Your employer is thinking about scrapping the annual performance review and replacing it with a short check-in between each employee and their manager once a month. Staff have been asked for their opinion. Do you agree with the change? Explain why.
Thirty seconds of preparation, honestly
You are not outlining an essay. You are settling two things.
First, the position — and if it is conditional, the condition, because that is the edge that makes it arguable. Here: yes to monthly check-ins, but keep one written record a year. That rules something out in both directions, which saves it from being "it depends".
Second, two reasons that are not each other. When feedback arrives is one axis — what people will say out loud is another.
The answer
I agree with the change, with one condition I'll come back to at the end.
The main reason is timing. My last review was in March, and half of it was about a report I wrote the previous April — eleven months earlier. And she wasn't wrong, actually. The feedback was fair. But by March that project was closed, the client had moved on, and there was nothing I could do with it. If she'd told me in May, I'd have fixed it in a week.
The second reason is a different thing entirely. It's about what people are willing to say. When the one conversation you get all year is also the conversation that decides your raise, nobody walks in and admits they're behind on the training, or that they don't understand half of what Finance sends them. You go in and you sell yourself. Twenty minutes a month with no money attached to it — people will actually put their hand up.
Now, the obvious objection is the paperwork. The annual review leaves a written record, and promotions need something on file. That's true, and I wouldn't want to lose it. So keep a one-page summary at year end, written up from the twelve check-ins.
Monthly for the real conversation, one page a year for the file. That's what I'd vote for.
Roughly 225 words for 90 seconds — about 150 a minute, unhurried, not a race.
Read against the four dimensions
Content/Coherence. The position lands in the first sentence and is never re-opened. The two reasons are genuinely separate — one about when information arrives, one about what people will admit — and both survive the disagreement test: a manager could argue that March feedback still helps, or that staff are candid in reviews. The objection is put in its strongest form ("promotions need something on file"), conceded, then answered.
Vocabulary. Nothing rare — precision that fits this situation only: "the client had moved on", "no money attached to it", "one page a year for the file". Concrete words about a real workplace beat impressive words about nothing.
Listenability. Officially, "How easy is it to listen to and understand the response?", with "Rhythm, pronunciation, and intonation" among its sub-points. Read it aloud: short units — "And she wasn't wrong, actually." — and audible signposts, "The second reason", "Now, the obvious objection". A listener hears the turn coming. Long subordinate chains flatten intonation.
Task Fulfillment. The verb was argue, and the answer argues: takes a side, supports it, handles the counter-argument, closes deliberately. Beneath this dimension sits the harshest official rule — a response giving insufficient information to assess scores 0 or 1, and levels 0 and 1 map to no CLB level at all.
What no one can tell you
- This page does not put a band on that answer, and no page honestly can. There is no answer key for Speaking, no published weighting across the four dimensions, and no rater in the room.
- Paragon's pages disagree even on who marks you — three to five speaking raters on the results page, at least three certified raters in the Score Guidance PDF. The FAQ's AI-human hybrid answer names Writing only — whether Speaking is AI-scored is not published either way.
- What is published is the four dimensions. Aim at those.
What this answer is not
It is not a structure to fill in. Three of its lines exist only because of this prompt: "My last review was in March, and half of it was about a report I wrote the previous April", "they don't understand half of what Finance sends them", and "keep a one-page summary at year end, written up from the twelve check-ins". Take away this review cycle and all three are false.
This answer works because of its details, and its details are the part you cannot borrow. Carry the wording into a prompt about charging for parking downtown and you are reciting furniture from someone else's house.
What travels is what the answer does: commits in the first sentence, puts its reasons on different axes, makes one concrete, answers the other side, stops on purpose. Habits, not lines.
Try one
[Try Task 7: Expressing Opinions →] — one full task, marked, no card required.