Sixty seconds. That's the whole long turn in CAE Speaking Part 2, and most candidates spend about forty of them describing photos nobody asked them to describe.
I sit in on enough mock exams to know the pattern by now. Candidate gets two photos, glances at the question printed above them, and then just starts narrating what's in each picture as if the examiner were blind. Person on the left is wearing a blue jacket. There's a mountain in the background of the second photo. It sounds reasonable in the moment. It's also not what Part 2 is testing.
The task is comparison, not description
CAE Speaking Part 2 gives you two photos connected by a theme and a specific question printed above them, something like "compare the photos and say why the people might have chosen these activities." You get about a minute to talk. Your partner gets roughly thirty seconds afterward to react to one small part of what you said.
That printed question matters more than most candidates treat it. It's not a suggestion. It's the actual task, and an examiner is listening for whether you address it, not just whether you can name what's in front of you.
Describing each photo separately, one after the other, isn't comparing. It might feel like comparing because you're talking about two images in sequence, but a comparison requires you to hold both photos in your head at once and talk about how they relate: similarities, contrasts, what's different about the mood, the setting, the people. If you describe photo one, pause, then describe photo two, you've produced two mini monologues glued together. That's the single most common thing I flag in feedback after mocks.
Quick fix
Practice starting your very first sentence with a comparative structure: "Both photos show people outdoors, but the atmosphere is completely different." That one sentence forces you into comparison mode immediately, instead of drifting into description and hoping you'll pivot later.
How the minute actually breaks down
There's no official stopwatch script Cambridge hands out, but after enough mock sessions a rough shape becomes obvious, and it's the one I coach toward:
The first stretch, maybe the first 35 to 40 seconds, should be spent directly comparing the two photos using the theme as your anchor. Not switching between them randomly. Actually contrasting: what's similar, what's different, why that difference might matter given the topic.
The last 15 to 20 seconds need to answer the actual printed question. This is the part candidates run out of time for, and it's also the part examiners weight heavily, because it's the bit that was explicitly asked for. If you spend fifty seconds comparing and then tack on "and I think they chose this because it's fun" in your final three seconds, that's technically an answer, but it's a rushed one, and it shows.
Some candidates split it more evenly, closer to 30/30. That's fine too. What doesn't work is spending the whole minute comparing and never circling back to the question, or spending so long on visual description that comparison and question both get squeezed into the last fifteen seconds.
What the second candidate is being scored on too
It's easy to forget the thirty-second follow-up is also marked. While candidate A is doing the long turn, candidate B isn't just resting. The examiner will ask B a short, related question, something like "which of these activities would you prefer to do?", and B needs to respond with a real opinion, not a restatement of what A already said.
I've watched candidates B just repeat A's comparison back with slightly different words, presumably out of nerves or because they weren't tracking closely enough. That's a wasted thirty seconds. The examiner wants a genuine reaction, ideally with a reason attached, even a short one.
Language that actually signals comparison
There's a small set of structures that do a lot of work here, and most candidates already know them but don't reach for them under pressure. Whereas, in contrast, unlike the first photo, both photos, similarly. These aren't advanced vocabulary. They're just underused in the moment because nerves push people back toward the simplest structure they have, which is usually plain description.
It helps to over-practice this specific move until it's automatic: photo, comparative phrase, photo. Not photo, full stop, new sentence, other photo. The comparative phrase is doing the grammatical job of linking the two images, and without it the examiner has to infer a comparison you never actually made explicit.
Hedge a little here too, because it's not a hard rule: occasionally a strong candidate will describe briefly before comparing, and it still works, as long as the comparison itself is unmistakably there by the halfway mark. What examiners are marking against is whether comparison happened at all, not whether you avoided every descriptive sentence.
The question you're actually being asked
Read the printed question twice before you start speaking. Not once, skimmed while you're already planning your opening line. Twice, properly. Candidates lose marks less often because they can't answer the question and more often because they never really registered what it was asking in the first place, and improvise something adjacent instead.
If the question asks why people might have chosen an activity, answering with what the activity involves isn't the same thing. It's related, sure, but it's not the answer. An examiner isn't going to interrupt you to redirect. They'll just note that the task wasn't fully addressed, and that shows up in the score.
Try this before your next mock: cover the photos, read only the printed question, and say out loud in one sentence what you think the answer might be, based on nothing but the theme. Then look at the photos and build your actual comparison around confirming or adjusting that instinct. It keeps the question anchored in your head instead of something you glance at once and forget by second twenty.