How to Prepare for the CAE Speaking Exam: A Part-by-Part Strategy

Cambridge reports that the C1 Advanced (CAE) speaking paper lasts only 15 minutes, yet it is the part candidates feel least prepared for. That is not surprising: most study time goes into Use of English and writing, while speaking gets whatever is left over. The good news is that 15 minutes is short enough to prepare for completely, and the exam follows a structure you can learn inside out.

This guide breaks down exactly how to prepare for the CAE speaking exam, part by part and criterion by criterion. By the end, you will have a daily routine that targets what examiners actually score, not vague "speak more" advice.

What is the CAE speaking exam format?

The C1 Advanced speaking paper has four parts, and you take it with one other candidate and two examiners. One examiner talks to you; the other sits silently and scores. Knowing what happens in each part removes half the anxiety before you ever open your mouth.

Part 1 is a two-minute interview: questions about your work, studies, travel, and interests. Part 2 gives you three pictures and asks you to compare two of them for one minute, then your partner answers a short follow-up question. Part 3 is a two-minute discussion with your partner about a prompt, followed by a one-minute decision-making task. Part 4 extends that discussion with broader questions from the examiner.

The key detail most candidates miss is that every part tests the same four skills, just dressed differently. A comparison of pictures in Part 2 and a discussion in Part 3 both score your grammar, vocabulary, coherence, and pronunciation. This means your preparation can be systematic rather than part-specific.

What do CAE examiners actually score?

The speaking paper is marked on four criteria, each weighted equally: grammatical resource, lexical resource, discourse management, and pronunciation. Understanding these is the single biggest lever in your preparation, because it tells you what to practise instead of just "talking a lot". Tutoro tracks exactly these gaps as you speak, so weak grammar and vocabulary resurface on their own rather than waiting for you to notice them.

Grammatical resource means using a range of structures accurately, including complex ones like conditionals, passives, and reported speech. Lexical resource means a wide, precise vocabulary with natural collocations, not just fancy words dropped in randomly. Discourse management is how you organise ideas: linking words, coherent paragraphs of speech, and keeping answers relevant. Pronunciation covers individual sounds, word stress, and intonation, not accent.

Cambridge publishes the official C1 Advanced speaking assessment scales so candidates can see exactly what each band requires. Reading them is worth an hour of your time: they turn a vague fear into a checklist you can self-assess against. The Cambridge English assessment criteria for speaking break each band into observable behaviours, which is what makes them useful for self-assessment.

Examiners are not looking for perfection. They are looking for range, control, and whether you can keep a conversation moving at C1 level, not whether you sound like a native speaker.

How do you handle Part 2, the long turn?

Part 2 is where many strong candidates underperform, because speaking alone for a full minute feels unnatural. The fix is a structure you can reuse for any set of pictures, so the minute becomes a matter of filling a familiar frame rather than improvising under pressure.

Start by comparing the two pictures directly: "Both show people working, but the first is outdoors and the second is in an office." Then speculate about one or two of the people: "The man might be a photographer, given the equipment he's holding." Then give a brief opinion or relate it to experience. This compare, speculate, conclude pattern gives you a reliable skeleton.

Practise this with a timer, because one minute is shorter than it feels. Record yourself doing three Part 2 turns in a row, then listen back and count how many times you repeated "I think". If you want a partner who will run this drill without getting bored, an AI English tutor can present picture prompts and score your pronunciation word by word while you speak.

How do you handle Part 3, the collaborative task?

Part 3 is scored on interaction, and this is where candidates who practised alone get exposed. The examiner wants to see you invite your partner in, respond to what they say, and build on it, not deliver two parallel monologues.

Use phrases that hand the conversation over: "What do you think about that?", "Would you agree?", "That's a good point, and it connects to...". Disagreeing politely is also a skill examiners reward: "I see it slightly differently because...". The decision-making minute at the end is not about finding the "right" answer; it is about negotiating toward any answer together.

The challenge is that you cannot practise collaboration alone. A human partner is ideal, but scheduling one daily is hard. A conversation partner that never gets tired and can run the whole Part 3 prompt with you is the gap an AI English conversation practice tool fills, giving you repetition most learners never get before exam day.

How do you build a daily speaking routine?

Fifteen minutes of speaking every day beats two hours once a week, because fluency is a motor skill that degrades without regular use. The routine below targets the four criteria deliberately, so no session is just "chatting aimlessly".

• Warm-up (2 minutes): describe your day using two conditional sentences and one passive, to force grammatical range.
• Part 2 drill (5 minutes): two one-minute long turns with a timer, using the compare-speculate-conclude frame.
• Part 3 drill (5 minutes): a discussion prompt where you consciously hand over and build on ideas.
• Shadowing (3 minutes): repeat short audio clips to work on stress and intonation.

Recording yourself is non-negotiable. You cannot hear your own filler words or flat intonation while speaking; you can on playback. Listen for one thing per session: this week, count your linking words; next week, check your question intonation.

Tutoro runs a whole CAE speaking lesson inside a conversation, which suits candidates who need daily reps without a human partner on call. The tutor presents Part 2 picture prompts and Part 3 discussion tasks as messages in the thread, then scores your pronunciation word by word as you speak, so the drills from this guide become something you can run every day and review against the four marking criteria.

It is less useful if you already have a structured class with regular speaking feedback, in which case it works better as a supplement than a replacement.

What mistakes cost CAE candidates marks?

Some errors are invisible to the candidate and obvious to the examiner. The most common is the filler-word habit: "I think", "actually", and "you know" repeated until they become punctuation. One or two are natural; ten per answer signals a vocabulary ceiling.

Another is answering the question you wish had been asked rather than the one asked. Examiners hear this constantly in Part 1, and it costs discourse management marks for relevance. Listen to the question, answer it directly, then extend. A third is flat intonation, which makes strong content sound uncertain.

Finally, candidates lose marks by treating Part 2 as a description contest. Describing what you see is A2-level; C1 requires comparison, speculation, and opinion. Every picture turn should contain at least one speculative phrase like "might be", "could be", or "seems to be".

How long should you prepare for the CAE speaking exam?

Most candidates need six to eight weeks of consistent speaking practice before the CAE speaking exam, assuming they are already at a solid B2-C1 level. The speaking paper rewards habits built over time, not cramming, because fluency and pronunciation change on a timescale of weeks, not days.

If your exam is sooner, prioritise the criteria you can move fastest: discourse management improves quickly once you learn linking phrases and the Part 2 frame, while pronunciation of individual sounds takes longer. Do not spend your last week learning new vocabulary; spend it running timed mock parts so the format feels automatic.

On exam day, the single best thing you can do is treat the examiner as a conversation partner rather than a judge. Candidates who relax into the interaction consistently score higher, because the four criteria all reward natural, connected speech over stiff, rehearsed answers.

Frequently asked questions

Here are the questions candidates ask most often once they start preparing.