Part 3 Conversations: how to use the printed questions before the audio starts

You get 39 questions, but only 13 conversations to earn them

Part 3 is built from 13 conversations, each carrying exactly three questions, for 39 total. That ratio matters: you are never listening for one answer at a time, you are listening for three answers inside a single 30-45 second exchange, in order. Unlike Parts 1 and 2, the questions and all four answer options are printed in your test book or on screen — the entire tactical advantage of Part 3 comes from using that printed text in the seconds before the audio starts, not from having sharper ears.

The pre-read sequence, in the order that actually pays off

• Read all three question stems first — not the answer options yet. 'What problem does the woman mention?', 'What must happen by 5 PM?', 'What does the man say he will do next?' tells you the shape of the conversation before you hear a word of it.
• Only after the three stems are locked in, glance at the options for question 1 if you have a second or two left — options 2 and 3 are rarely worth the time, since you won't need them until later and the conversation will already be moving.
• Do not try to read all twelve options (three questions times four choices) before the audio starts. That is the single most common way test-takers run out of pre-read time and end up hearing the first line of the conversation with their eyes still on the page.
• If a set includes a graphic, look at the graphic itself in your remaining seconds, not the questions again — you already know question 3 will send you there.

Gist, then detail, then 'what happens next'

The three questions in a set are not random — they track the conversation from start to finish and from general to specific. Question 1 is almost always gist: who is speaking, where, or what the problem is (compare 'What problem does the woman mention?' or 'Why does the man contact the restaurant?'). Question 2 asks for a specific fact stated partway through — a deadline, a feature, a number, a cause ('What must happen by 5 PM for the brochures to be reprinted?', 'What feature of the Lily Room does the man mention?'). Question 3 is the forward-looking one: what a speaker will do, say, or send next, and its answer sits in the conversation's final lines. Answering in that sequence as the conversation unfolds — rather than waiting to the end and trying to reconstruct all three — is what the 45-second window is actually built for.

Graphic questions: match the spoken detail to the visual, not the wording

When a set includes a graphic — a price list, a floor plan, a schedule, a seating chart — the correct answer is never read aloud as a full sentence. Instead, one speaker states a detail (a name, a time, a code, a row) that you must cross-reference against the graphic yourself to find the answer the question actually wants. If the graphic is a delivery schedule and the woman says 'the one that's delayed a day is the truck leaving on Thursday,' the question won't ask about Thursday directly — it will ask something like 'Which shipment will arrive late?', and you have to look at the graphic to see which shipment corresponds to that Thursday truck. Skimming the graphic before the audio starts, so you already know what its columns or labels represent, is what makes that cross-reference possible in real time instead of after the fact.

The scenarios Part 3 keeps returning to

• Office logistics gone slightly wrong: a booked room falls through and a backup has to be found (a room reserved is unavailable, an alternative like a larger room down the hall gets proposed with its features listed as the detail question).
• Scheduling conflicts: a meeting or interview clashes with something else already on the calendar, and a new time gets proposed and confirmed, often pulling in a third person who needs to be looped in.
• Vendor and delivery problems: a shipment is delayed for a stated reason (customs, a supplier issue), and the fix — a partial shipment, an expedited option — is what question 2 or 3 is built around.
• Customer service and reservations: confirming or changing a booking (a restaurant table, a hotel shuttle time), often with a follow-up request layered on — a dietary restriction, a luggage question, a room count that changed.
• IT and facilities requests: something isn't working, the caller reports the symptom, and the fix or next step is what closes the conversation.
• In all of these, distractor options recycle vocabulary from the conversation (the restaurant name, the room name, the word 'shipment') attached to something that was never actually said — the trap is recognizing a familiar word, not recognizing a true statement.

'What will the man/woman do next?' has a specific failure mode

This question type rewards listening to the very end of the conversation, and it punishes stopping early. The correct action is usually the last commitment made — 'I'll call them now,' 'I'll follow up with a written confirmation once it's booked,' 'I'll let the driver know in advance.' The classic wrong answer takes an action that was proposed and then dropped (a second shuttle that turns out not to be needed, a refund nobody actually offers) or assigns the right action to the wrong speaker — the man commits to calling Marketing, not the woman, even though she is the one who raised it. Track who says 'I'll' last, not just what gets discussed.

Two speakers versus three

• Most sets use two speakers, and options label them simply 'the man' and 'the woman' — track who owns each fact as you go, since a detail question can hinge on who said the number, not just what the number was.
• Three-speaker sets add a second man or woman, so options may say 'the first man' and 'the second man,' or use a name once it's given. The extra voice usually exists to be looped into a decision the other two already made — a colleague from another department, a specialist being consulted — so listen for whose turn it is to speak and why they were brought in.
• Three-speaker sets don't add a fourth question, but they do compress more identification work into the same 45 seconds, so the pre-read matters even more: know before the audio starts whether you're tracking two voices or three.

Managing the gap between conversations

You get only a few seconds between the end of one conversation's three questions and the start of the next one's audio — not enough to relax, but enough to start the pre-read cycle again if you move immediately. The moment you've locked in your answer to question 3, flip to the next set's three stems rather than reviewing what you just answered. Losing that gap to hesitation is how test-takers who understand every conversation still end up rushing the pre-read on set six, seven, and eight.