How to Prepare for Cambridge B2 First Speaking: A Complete Guide

The Cambridge B2 First speaking test lasts just 14 minutes, but candidates spend months worrying about it. The reason is simple: it is the one part of the exam where you cannot revise the night before. Speaking is a live performance, and the examiner hears your habits, your gaps and your nerves in real time.

Here is the good news. The test is extremely predictable. It follows the same four-part structure every single time, and it is scored on four criteria that Cambridge publishes openly. Once you know what is being measured, learning how to prepare for Cambridge B2 First speaking stops feeling vague and becomes a checklist.

How to prepare for Cambridge B2 First speaking: What is in the test?

You sit the test with one or two other candidates, and there are two examiners in the room. One talks to you and runs the test. The other sits silently and does all the scoring. This matters, because it means the person chatting with you is not the one marking you, so do not perform for the wrong person.

The test has four parts, and they get progressively harder. Part 1 is a short interview where each candidate answers questions about themselves: home, work, studies, hobbies. It lasts about two minutes and is designed to settle you down. Part 2 is the long turn: each candidate gets two photographs and one minute to compare them and answer a question, followed by a 30-second response from the other candidate.

Part 3 is a discussion with your partner. You are given a written prompt and a set of ideas, and you have two minutes to talk it through together and reach a decision. Part 4 is a follow-up discussion on the same theme, this time with the examiner joining in. The whole thing is timed tightly, which is why practising against a clock matters as much as practising the language.

You can find the official format and sample papers on the Cambridge English B2 First exam page, which is worth reading before you do anything else.

What do examiners actually score?

There are four criteria, each worth 25% of your speaking mark. Grammar and Vocabulary looks at whether you use a range of structures and words, and how accurately. Discourse Management is about how long you can speak, how you organise your ideas and whether you can link them logically. Pronunciation covers individual sounds, word stress and intonation. Interactive Communication measures how well you take part in the conversation, including with your partner.

Notice what is not on that list: your accent, the size of your technical vocabulary, or whether you sound British or American. Candidates lose marks by chasing the wrong things. A candidate who speaks slowly but holds a two-minute turn with clear links will outscore one who fires off fancy words but dries up after twenty seconds.

The most important insight is that half your mark comes from skills you can train mechanically: discourse management and interactive communication are habits, not talent. You can read the full assessment criteria published by Cambridge to see the exact descriptors for each band.

How do you master Part 2, the long turn?

Part 2 is where most candidates feel the most exposed, because you speak alone for a full minute while everyone listens. The task is always the same shape: compare two photos, and answer the question printed above them. The trap is describing the photos instead of comparing them.

A candidate who says "In the first photo there is a woman cooking, and in the second photo there is a man eating" is describing. A candidate who says "Both photos show people involved with food, but while the first is about preparing a meal, the second is about enjoying one" is comparing. The difference is the word "both", and it is the single highest-value word in Part 2.

Build a reusable frame. Open with a comparison, answer the question directly, speculate once, and close. For example: "Both pictures show people working with their hands. In the first, the woman is concentrating on a precise task, whereas the man in the second seems to be doing something more physical. I think the first job requires more patience, because one small mistake ruins the result." That is roughly 45 seconds of solid, organised speech.

Practice this frame until it is automatic, because under pressure you will fall back on whatever you have rehearsed, not whatever you planned. One effective way to rehearse is to describe photos out loud to a partner daily. If you do not have a partner, an AI English tutor can hold the role: it shows you images, listens to your turn and scores your pronunciation word by word, which is exactly the feedback you need on a timed long turn.

How do you handle Part 3, the discussion?

Part 3 is not a debate and it is not a race. It is a collaboration. The examiner wants to see you invite your partner in, respond to what they say and build on it, not deliver two separate monologues. Many candidates treat it as a turn-taking competition and lose marks on Interactive Communication without realising.

The practical skills are turn-inviting and turn-building. Phrases like "What do you think?", "I partly agree, but have you considered..." and "That's a good point, and it connects to..." do real work. They signal that you are listening, which is half the criterion. The other half is keeping the conversation moving, so have a fallback question ready for any silence.

You also need to manage the decision. Part 3 ends with you agreeing on something, and candidates who never reach a decision lose structure marks. Around the one-minute mark, steer toward a conclusion: "So if we had to choose, which of these matters most to you?" This is a skill you can only build by doing it, which is why timed practice with a real interlocutor, human or AI, is worth more than any amount of reading.

Tutoro runs speaking practice as a conversation inside a lesson, not a separate exercise, so the long turn and the discussion are trained the way they actually happen in the exam. It scores your pronunciation word by word as you speak and brings weak vocabulary back on its own, which suits a B2 candidate who knows the format but needs structured, daily output. It does not suit anyone who wants a human examiner's subjective judgement on their answers in the final week before the test.

What does a week-by-week plan look like?

If you have eight weeks, the plan writes itself. Weeks one and two are about learning the format cold: watch sample videos, read the criteria, and build your Part 2 frame and your Part 3 phrases. Weeks three to five are volume weeks: one full speaking session a day, at least, mixing all four parts, always timed.

Weeks six and seven are about weakness. Record yourself, listen back, and mark yourself against the four criteria. Most candidates find their discourse management collapses after 40 seconds, or their pronunciation is fine on single words but flat across a sentence. Fix one thing at a time. Week eight is simulation: two or three complete run-throughs under exam conditions, with a partner and a timer.

The key variable is not hours, it is frequency. Twenty minutes a day beats three hours on Sunday, because speaking is a motor skill and motor skills consolidate with daily repetition. If your only speaking outlet is a weekly class, you will arrive under-prepared no matter how good the class is. This is where a tool you can open every evening, like Tutoro, changes the arithmetic: the daily session becomes the default rather than the exception.

What are the mistakes that cost marks?

The most expensive mistake is silence. An answer that stops at 20 seconds when the task asks for 60 loses marks on discourse management and interactive communication at once. The fix is a stock of filler that buys thinking time: "That's an interesting question, let me think for a moment." It feels awkward to say, but it is exactly what a fluent speaker does.

The second mistake is ignoring your partner. In Parts 3 and 4, a candidate who answers only when asked and never reacts to the other person is handing away marks. The third is overcomplicating vocabulary under pressure and producing errors where simple accurate language would score better. Accuracy beats ambition when you are nervous.

The fourth is preparing only content and never timing. Candidates who have never spoken for a full minute before the exam discover in the room that 60 seconds is longer than it feels. Time every practice turn. If you run short, add a speculation or a comparison; if you overrun, learn to close.

How long does it take to feel ready?

For a learner already at B2, four to eight weeks of daily speaking is enough to walk in confident. For a learner at B1, the honest answer is that the exam itself is the wrong target until the level is closer, because the speaking test measures B2 output, not B2 input. Six to twelve weeks of structured practice is realistic for most people.

The variable that matters most is feedback. Speaking into the air builds fluency but not accuracy, because you cannot hear your own errors. You need someone or something that responds and corrects, which is why a conversation partner, a tutor, or a tool that scores your pronunciation and recycles weak words is the difference between practising and improving.

Start today, not next week. The test rewards the candidate who has spoken a little every day for two months, not the one who crammed for two weekends. The format is fixed, the criteria are published, and every minute of timed speaking you do now is a mark you bank for exam day.