CELPIP Speaking Task 7: The Opinion Playbook

Here's the honest truth about CELPIP Speaking Task 7: the prompts are ordinary opinion questions, often about work or society — "should companies allow remote work?", "is social media good for friendships?" Nothing technical, nothing you need a news habit to answer. And yet this is the task where template-heavy speakers sound most robotic and prepared speakers sound spontaneous, all inside roughly 90 seconds. This post gives you the playbook: a position-picking rule that works in five seconds, the structure that survives tiredness, an energy plan, and a full band-9 sample on the remote-work question, annotated.

To practice the opinion task with a real clock and prompt audio, take the free mock → celpipuni.com/free-mock-test.

TL;DR

Task 7 is the opinion task — roughly 90 seconds of speaking after about 30 seconds of prep, usually on a work or society topic. The playbook: pick a side in the first five seconds of prep (never hedge both), build your answer with PREP — Point, Reason, Example, Point — and manage your energy, because this is the longest response in the section and it arrives deep in the test. Raters hear fatigue as slower pace and more fillers, so the fix is practicing full-length runs, not isolated tasks.

What Task 7 actually asks

The prompt gives you a statement and asks for your opinion — sometimes "do you agree," sometimes a should-question with two defensible sides. The topics lean toward work and society: remote work, working parents, city planning, social media, retirement age. You get about 30 seconds of prep and roughly 90 seconds of speaking — the longest single response in the Speaking section.

Notice what "harder" does not mean here. It doesn't mean a stricter rubric — same four criteria as every task. It means the prompt often weighs two genuinely defensible sides, so the lazy strategy of "pick the side with more to say" works less often. What replaces it: pick the side you can give a specific example for, fast. An opinion without an example is a band-7 opinion almost every time.

And here's the part nobody builds into their prep: by Task 7 you have been testing for close to three hours, and you've already spoken six responses into a headset. The two things the rater hears first are pace and fillers. A tired speaker slows down without noticing, and "umm" starts arriving between every clause. That's why this post spends as much time on energy as on structure — your Task 7 plan has to work at 80% battery.

The playbook: how to run prep and the 90 seconds

Prep, second 1–5: pick a side. Any side. Fast. The rule: choose whichever side you can attach a specific example to in five seconds — something from your work life, your family, your city. "Should companies allow remote work?" — if you (or someone you know) have actually worked remotely, that's your side, whichever it is. If you genuinely have an example for both, take the one you believe; belief gives your voice energy.

Prep, second 5–20: one reason, one example, in keywords. Not sentences. "commute → 2 hrs," "cousin's company → hybrid." Your prep notes are a map, not a script.

Prep, second 20–30: pick your concession. One sentence acknowledging the strongest point on the other side, and your counter to it. "Focus matters, yes — but that's a scheduling problem, not a location problem."

Then the 90 seconds, in PREP:

Point (10 seconds): "Yes — companies should allow remote work, at least most of the time."
Reason (20 seconds): the why — mechanism, not slogan. "Because commuting doesn't create value; it just moves you."
Example (30–35 seconds): the specific one you chose in prep. A person, a company, a number, a consequence. This is where bands are decided.
Point again (15 seconds): restate, with the concession folded in. "Focus concerns are real, but they're about how you schedule the day, not where you sit. Overall — yes, remote work, done properly."

Why PREP works for a tired brain: it's a loop, not a maze. If you lose your thread mid-task (you will, at this point in the test), you just arrive at the next letter. The structure catches you, which is exactly what a 7-aiming-for-9 needs at hour three.

One test-day reality: there's no pause button between tasks — the prompt appears on a countdown you don't control. Practice the transition: give an opinion for 90 seconds, then immediately take the next prompt. That handoff is where unpracticed test takers lose ten seconds of prep just orienting. (Formats get revised periodically — confirm current timings on the official CELPIP site.)

The numbers: Task 7 fresh vs Task 7 at hour three

	Fresh	At hour 3 (untrained)
Pace	Natural, ~140–160 words/min	Slows without you noticing
Fillers	A few per minute (normal)	Every clause ("umm, so, yeah")
Example quality	Specific person, place, number	Vague ("some people," "I think")
Position	Stated in first 5 seconds	Hedged — "it depends"
Structure	PREP loop holds	Middle collapses, ending rushed

The fix isn't talent — it's rehearsing tired. One full 8-task practice run per week, in order, at your real test time of day, trains the condition Task 7 actually arrives in.

What fatigue sounds like — and the mistakes it creates

Hedging both sides. "Remote work has advantages, but the office also has advantages, and it depends on the situation." Tiredness wants the safe answer; the safe answer is the band-capper. The task says should companies allow it — take a position.

The example that dissolves. Fresh: "My cousin's company in Mississauga went hybrid in 2022, and her two-hour round commute became a two-hour day-care pickup." Tired: "Many people work remotely and it's good for them, I think, in many ways." Same speaker, four tasks apart. The second one has no person, no place, no number — and no development.

Fillers multiplying. One "umm" per minute is human. One per clause is what raters hear as a retrieval problem — and deep in the section, that's exactly what it is: your brain is retrieving slower, and the fillers are the sound of it. The fix isn't willpower; it's the keyword prep above, which gives your mouth something to reach for.

The soft landing. Ending with "so, yeah, that's what I think" — the energy is gone and it shows. Your restate point is the last impression; make it a real sentence.

What good looks like: a band-9 sample, annotated

Prompt: should companies allow employees to work remotely? Give your opinion.

"Yes — most of the time, they should.

My cousin's company in Mississauga went hybrid in 2022. Before that, she commuted two hours a day — an hour each way on the 401. Now she goes in twice a week, and those two reclaimed hours go to picking her kids up from school instead of sitting in a car. Her output didn't drop; her manager's own words were that she's more available now, not less. That's the pattern I've seen everywhere remote work is done properly: the commute hours don't vanish, they get reassigned to actual life.

The strongest argument against is focus — some people genuinely concentrate better in an office. Fair. But that's a scheduling problem, not a location problem. Book your deep-work hours, mute the notifications — those fixes work at a kitchen table same as at a desk.

So yes — allow it, structure it well, and the company gets the same work plus a human being who isn't spending a tenth of their waking life on the highway."

Why this scores 9:

Task fulfillment: a clear position in the first five words, held throughout, with a concession genuinely answered.
Content/coherence: textbook PREP, invisible — point, mechanism, a developed example, restate with the concession folded in.
Vocabulary: "reclaimed hours," "reassigned to actual life," "a tenth of their waking life." Precision, not rarity.
Listenability: short punches after long sentences ("Fair."). Rhythm survives, which is the fatigue tell in reverse.
The example: a person, a company, a place, a year, a number. Five anchors — the biggest reason this sample isn't a 7.

Your Task 7 checklist

Side picked in the first 5 seconds of prep — no "both sides" openers
One specific example chosen before speaking: a person, a place, a number
PREP keywords, not sentences, in your prep
One concession + counter, prepared in the last 10 seconds of prep
Full 8-task practice runs weekly — in order, timed, at your test hour
Last sentence is a real restate, never "so, yeah"

Want to know if your Task 7 sounds fresh or fatigued to a grader's ear? Get an evaluation with band-level feedback → celpipuni.com/evaluation.

The deeper dive: your real Task 7 opponent

Here's an opinion built from watching a lot of speaking recordings: nobody loses the opinion task on structure. They lose it on battery. By the seventh task, almost every candidate we see still knows PREP, still knows to take a side. The knowledge is all there. What's gone is the fuel for it: the pace, the specificity, the voice. You can hear the moment the example was supposed to arrive and something vaguer came out instead.

So the highest-leverage Task 7 training is boringly physiological. Do your weekly full 8-task run at the same time of day as your real test, eat beforehand the way you will on test day, and don't pause between tasks — the section gives you no pause, so your practice shouldn't either. And treat the last two tasks as a sprint you've saved energy for: raters anchor hard on the first and last thing they hear, and your final tasks are the last thing. Some of our students consciously spend a little less on the middle tasks — good, not perfect — so the bookends get their best voice.

Tools and resources

The official CELPIP site — free sample opinion prompts and the current format. (Formats get revised; confirm there.)
CELPIP Speaking: the complete guide — the full task map and the daily practice loop: celpipuni.com/celpip-speaking-guide
CELPIP Speaking time management — pacing all 8 tasks without rambling or racing: celpipuni.com/celpip-speaking-time-management

Frequently asked questions

How long is CELPIP Speaking Task 7?

About 30 seconds of prep and roughly 90 seconds of speaking — the longest response in the Speaking section. Confirm current timings on the official CELPIP site; formats get revised.

Is Task 7 harder than the other speaking tasks?

The prompt often has two genuinely defensible sides, but the real difficulty is placement — you're deep into a three-hour test when it arrives. Structure knowledge doesn't change; energy does. Train tired.

Can I say "it depends" in Task 7?

Avoid it as your position. If the honest answer is genuinely conditional, make the condition your position: "Yes, with structure — here's what has to be true." That's still a stance, not a hedge.

Do I need real facts and statistics?

No — and don't invent them. A specific example from your own world (a cousin, a former employer, your city) is what the rubric wants. "Studies show" with nothing behind it is a band-7 filler move.

Are Task 7 topics always about work?

Work and society dominate — remote work, working parents, city life, technology, education, retirement. See the recurring themes: celpipuni.com/celpip-speaking-topics

What to do next

Today: answer the remote-work prompt out loud, timed, 90 seconds. Record it.
Listen back and count: a real example with a person and a number, or vague "many people"?
This weekend: one full 8-task run, in order, at your real test time of day. Notice what Task 7 feels like at that point in the sequence.
This week: get one evaluation, and ask specifically where your pace and fillers sit by the late tasks.

The bottom line

Task 7 is an ordinary opinion task that arrives at a demanding moment — deep in a three-hour test, with the longest response in the section. The playbook: pick your side in five seconds, prep keywords not sentences, run PREP, save one concession to fold into your restate. Then rehearse the actual condition — full-length runs, in order, tired. The rater hears energy as much as English. Make sure yours arrives.

See what your opinion task sounds like under real conditions — take a free mock test with the full 8-task sequence → Take your free mock test.