Here is the test you can run on any Task 4 answer, including your own: delete every future-tense word and see whether what is left is a description of the picture. If it is, you answered Task 3 twice. A prediction is not a description with "will" in front of it — it is a claim about something that is not in the frame yet.
Task 4 hands you the same image you have just spent sixty seconds describing, so the trap is built into the order of the test. Your mouth is still warm from naming things.
The failure
The failure is the sentence that sounds like a prediction and is actually an inventory. "The woman will be standing on the ladder." "The man is going to be pushing his cart." Nothing has happened. The clock runs out at eight things named and nothing predicted.
The sample answer below is built to make that impossible. Not one of its sentences states something you can see in the picture. Every sentence is a consequence.
The task
Task 4 is "Making Predictions", the fourth of the eight Speaking tasks. You get the same image as Task 3 and you say what happens next. The clock is 30 seconds of preparation and 60 seconds of speaking — roughly 130 to 160 words at a normal pace. Those per-task times are official but they live in Paragon's Speaking study pack, not on the test-format page, which is why so few sites publish them.
The prompt
(original AlmiWorld practice prompt)
The picture: the outdoor garden section of a hardware store on a Saturday morning in late spring. A man is pushing a flat cart loaded with four bags of soil and, balanced on top, a tray of marigolds hanging over the edge. A small girl is ahead of him with both hands on the front rail of the cart, leaning back and pulling. Along the aisle, a coiled hose is running; a thin sheet of water has spread out near the corner where the aisle turns toward the tills. To the left, a staff member in a green apron is on the second step of a stepladder with her back to the aisle, hooking up hanging baskets. Through the doors behind them a pickup truck waits with its tailgate down, and the sky past it has gone dark.
You have 30 seconds to prepare. Then tell a friend what you think will happen next.
Thirty seconds of preparation, honestly
You do not plan a structure in thirty seconds. You pick a first domino and you decide who does not know yet.
The first domino here is the tray, because it is the only thing already off balance. The person who does not know yet is the woman on the ladder. And the sky gives you a second, slower chain that lets you end somewhere other than the floor.
The answer
Okay — within about ten seconds that top tray of marigolds is going off the cart, no question. She's not going to slow down when he asks her to, and the water from that hose is going to reach the corner before they do, so the soil won't just sweep up, it'll smear. He'll let go of the cart to save the tray, which — actually, that's the worse move, because the second he lets go, the soil bags come down too. The woman up the ladder won't hear a thing until it's already on the floor, and she'll almost certainly turn round too fast on that top step. My guess is she comes down fine but the basket doesn't. And the rain's going to start before he gets those bags into the truck, so he'll be driving home with forty kilos of wet soil and a kid who's decided none of it was her fault.
About 150 words. Target: 60 seconds.
Read against the four dimensions
Paragon publishes four dimensions for Speaking: Content/Coherence, Vocabulary, Listenability, Task Fulfillment. Nobody in this room can put a level on the answer above — there is no key for Speaking, no published weighting for the four dimensions, and no rater here. What follows is only where the answer feeds each one.
Content/Coherence. It is one causal chain, not a list. The tray falls, so he lets go; he lets go, so the bags come down; the noise is what turns the woman on the ladder round. "The second he lets go, the soil bags come down too" is the joint that holds it together, and it is why nothing here has to be sequenced with "and then".
Vocabulary. The nouns are the store's own — flat cart, tray of marigolds, hanging basket, tailgate. Task 4's real vocabulary, though, is certainty: "no question", "almost certainly", "my guess is". Three different levels of confidence in 150 words is a prediction being graded on its likelihood, not a guess dressed up.
Listenability. Officially, "How easy is it to listen to and understand the response?", with rhythm and intonation among its sub-points. The clauses are short and the sentence that breaks — "which — actually, that's the worse move" — is where a listener hears a person thinking, not reciting. Accent is not what is being rated here; being followed is.
Task Fulfillment. Every main verb is a future one. Nothing in the answer could be checked against the picture, because nothing in it has happened yet — which is the task. This is also the floor: a response giving insufficient information to assess scores 0 or 1, and a minute of naming objects is exactly how a Task 4 response ends up thin.
What this answer is not
It is not a shape you can carry to another picture. This answer works because of its details, and its details are the part you cannot borrow.
"The water from that hose is going to reach the corner before they do" needs a running hose and a corner. "She comes down fine but the basket doesn't" needs someone elevated and holding something. "Forty kilos of wet soil and a kid who's decided none of it was her fault" needs a truck, a storm and a child — and it is the line that most obviously cannot be reused, which is precisely why it is the line that earns most.
Memorise those sentences and you will speak worse on a different image, because you will spend the minute steering toward a line that does not fit. What you can take is the method: find the thing already off balance, name someone who has not noticed, and let the weather give you an ending.
Try one
[Try Task 4: Making Predictions →] — one full task, marked, no card required.