14 Real B2 First Speaking Scores, Schools and Adult

Cambridge publishes real examiner scores for B2 First and B2 First for Schools, with commentary, as free PDFs. There are six such documents covering fourteen candidates.

We read all fourteen and put them in one table. Pronunciation is the top mark nine times out of twelve, and every single time it is not, Interactive Communication is.

SpeakShark is free for the criterion that steps in when accent is not the answer, three AI conversation sessions a day with no card.

In this guide: all 14 profiles · the substitution · what drags profiles down · schools against adult · correcting ourselves · practice · limits · method · FAQ

Key takeaways

Fourteen published B2 profiles, seven from B2 First for Schools and seven from adult B2 First.
Pronunciation is top or joint top nine times out of the twelve that are not flat.
All three exceptions have Interactive Communication as their single highest mark.
Grammar and Vocabulary is lowest or joint lowest for ten of the fourteen.
The school set spans 2 to 5. The adult set spans 3.5 to 4.5.

Every published B2 profile in one table

Five criteria are scored at B2. Bold marks each candidate's highest.

Exam	Candidate	Grammar and Vocabulary	Discourse	Pronunciation	Interaction	Global
Schools	Francesco	4.5	4.5	5	5	5
Schools	Emma	4.5	4.5	5	5	5
Schools	Lara	5	5	5	5	5
Schools	Agustina	2	2.5	3	3	3
Schools	Mateo	4.5	4	4.5	4.5	4.5
Schools	Julia	4.5	4.5	4	5	4.5
Schools	Malena	2.5	2.5	3	3.5	3
Adult	Giulia	4	4	4.5	4	4
Adult	Tabatha	3.5	3.5	3.5	4	3.5
Adult	Elena	4	4	4	4	4
Adult	Julia	3.5	4	4.5	4.5	4
Adult	Isabela	4	4	4.5	4.5	4.5
Adult	Natalia	3.5	3.5	4	4	4
Adult	Enzo	4	4	4.5	4	4

Lara and Elena have completely flat profiles, identical marks on all five criteria, so they cannot tell us which criterion stands out. We exclude them from the counts below and say so rather than letting them quietly inflate a figure.

When pronunciation is not top, interaction is

Of the twelve profiles that are not flat, pronunciation is the highest or joint highest mark in nine.

Three candidates break it, and the interesting part is that they break it the same way.

Candidate	Pronunciation	Interactive Communication	Gap
Julia, Schools	4	5	one full band
Malena, Schools	3	3.5	half a band
Tabatha, Adult	3.5	4	half a band

Those three are also the only candidates of the fourteen whose single highest mark is Interactive Communication. The set of "pronunciation is not top" and the set of "interaction is top" are the same three people.

Julia is the clearest case. Her pronunciation, at 4, is her single lowest mark, below her grammar, her discourse and her global achievement. Her Interactive Communication is a 5, a full band above it. Whatever she does with her partner is worth more to her profile than how she sounds.

This is the same pattern we found one level down. Across nine published B1 candidates, Interactive Communication is the single highest mark for four and top or joint top for eight, while pronunciation is the single highest for none. B2 is where pronunciation starts to dominate, and these three are the candidates for whom it has not taken over yet.

It also explains why Cambridge scores interaction separately at all. As we set out in the two examiner roles, Interactive Communication is marked by the assessor, the examiner who never speaks to you, on the basis of what you do with the other candidate rather than with the examiner.

Grammar and Vocabulary is the anchor

Read the first score column down and the more useful preparation message appears.

Grammar and Vocabulary is the lowest or joint lowest mark for ten of the fourteen. Agustina at 2 and Julia at 3.5 have it as their single lowest. Francesco, Emma, Malena, Giulia, Tabatha, Isabela, Natalia and Enzo have it joint lowest.

Only Mateo has it as a joint highest, and only Julia at Schools has a different criterion sitting alone at the bottom.

That inverts how most people prepare. Accent work feels productive and measurable. On this evidence, at B2, range and accuracy in grammar and vocabulary is what is actually holding profiles down, which is what our first look at the C1 and B2 scores found and what a larger sample now supports.

The school set spans the whole scale

Cambridge states that B2 First and B2 First for Schools follow the same format and use the same criteria. The difference is in the topics and the candidates, not in the standard, and we cover the format differences in B2 First against B2 First for Schools.

The published score sets differ sharply all the same.

B2 First for Schools: marks range from 2 to 5. The whole usable scale.
Adult B2 First: marks range from 3.5 to 4.5. A single band of movement.

Every adult candidate Cambridge publishes sits in a narrow competent band. The school set includes Lara scoring 5 on everything and Agustina scoring 2 on grammar and vocabulary in the same document as Mateo scoring 4.5.

This is a fact about the sample, not about the exams. These are teaching materials, chosen to illustrate marking. A wider spread in the school set most likely means Cambridge wanted teachers to see what a weak performance looks like next to a strong one in a school context. It is not evidence that school candidates vary more, and we would not use it that way.

What this changes about what we published earlier

We have to correct ourselves here, because we published a stronger claim than the evidence now supports.

Our first guide on Cambridge speaking scores reported that pronunciation was the highest or joint highest mark for all nine candidates in the set we had read. That set contained just two B2 candidates, Natalia and Enzo, plus seven from C1 Advanced.

With fourteen B2 candidates rather than two, the claim does not hold at B2. Nine of twelve is a strong tendency. It is not "every single one".

The honest summary across everything we have now read is this. Pronunciation is usually a B2 candidate's strongest criterion and rarely their weakest. Grammar and vocabulary is usually their weakest. And when pronunciation is not on top, it is interaction that has replaced it, not grammar.

We are leaving the earlier posts in place with their scope stated, rather than quietly editing the numbers, because the correction is the interesting part.

What to do with this

Do not lead your preparation with accent. At B2 it is already most candidates' strongest column.
Work on range and accuracy. Grammar and vocabulary anchors ten of fourteen profiles.
Practise talking to a partner, not to a teacher. Interaction is the criterion that lifts the candidates whose pronunciation is not carrying them.
Treat a flat profile as a signal. Lara and Elena are level across all five, which usually means nothing is holding them back and nothing is standing out either.

The interaction parts are where this is decided. B2 First Part 3 and Part 4 put you with the other candidate, while Part 1 and Part 2 give you far less room to initiate. Our B2 First speaking overview covers the shape of the whole test, and B2 First against C1 Advanced covers what changes if you go up.

Interaction is a conversation problem rather than a study problem, which is why role play scenarios and a daily speaking partner move it faster than a grammar book. Test your speaking level free first, and CEFR levels for speaking explains what B2 describes.

What we could not verify

This is a curated sample, not a random one. Fourteen candidates chosen by Cambridge to illustrate marking. Proportions here describe the published set and nothing wider.

Two profiles are flat and carry no signal. Lara and Elena score identically on all five criteria. We excluded them from the counts and have said so at every point.

We did not read every commentary in full. We extracted and verified every score table. The surrounding examiner prose is quoted only where we read it directly.

We did not check whether these are current editions. The documents carry a 2024 copyright line.

We are not claiming either exam is easier. The spread difference is about which candidates were published.

We did not re-examine C1 Advanced. Our C1 figures come from the earlier set of seven and were not revisited here, so the C1 part of the original claim stands untested by this work.

How we researched this guide

We opened Cambridge's B2 First preparation page in a browser and read the actual link list rather than working from a note. That is where the sample grew: our queue recorded three B2 First for Schools documents, and the page publishes six B2 documents in total, three for Schools and three for the adult exam.

Five of the fourteen candidates had never been read by us before, and they include two of the three exceptions the post is built on. Had we written from the queue's list, the finding would have been weaker and the correction to our earlier claim would not have surfaced.

We downloaded all six PDFs and extracted the score tables programmatically rather than transcribing them, then ran the counts in code rather than by eye, because "nine of twelve" and "ten of fourteen" are exactly the kind of figures a manual count gets wrong.

The counting rules were fixed before looking at the result. Flat profiles were excluded from any "highest" or "lowest" count, since a candidate scoring the same on everything is simultaneously top and bottom on all of it. That decision removed Lara and Elena and is why the denominators differ between claims.

Cambridge serves these PDFs normally to a browser. We fetched them from within a page on the same origin, which is our default for this site.

Practise speaking, from SpeakShark

SpeakShark is an AI English speaking practice app, and this data set is a reasonable argument for it. The criterion that rescues candidates whose pronunciation is not carrying them is interaction, and interaction cannot be drilled alone.

You talk, the AI answers what you actually said, and you get speaking feedback while the conversation is still running. The free tier gives basic feedback; the detailed pronunciation and grammar breakdown is on Premium.

Being straight about the limits: a free session runs five minutes with four turns, and it is one on one. A B2 First speaking test pairs you with another candidate, which we do not reproduce.

Start free with three sessions a day and no card. Paid sessions run ten minutes with unlimited turns. Limits are on the pricing page, and how it works walks through a session. Sign up here if the interaction column is the one you want to move.

We are a speaking improvement tool. We are not an exam preparation provider, and we are not affiliated with Cambridge English, the IELTS partners, Pearson or any exam board. For test format, booking and official practice material, go to the exam body directly. Use SpeakShark to make your spoken English stronger, and use official material to learn the test.

Site: speakshark.com
Support: speaksharksupport@gmail.com
Free tier: 3 AI conversation sessions per day, no card

Sources

Cambridge English, B2 First preparation: the page publishing all six examiner commentary documents for B2 First and B2 First for Schools
B2 First for Schools examiner comments, Francesco and Emma, Lara, Agustina and Mateo and Julia and Malena: the seven school score tables
B2 First examiner comments, Giulia, Tabatha and Elena, Julia and Isabela and Natalia and Enzo: the seven adult score tables
Council of Europe, CEFR Table 1, common reference levels global scale: the B2 descriptor both exams target

FAQ

How many B2 First speaking profiles does Cambridge publish?

Fourteen across six documents. Three cover B2 First for Schools and account for seven candidates: Francesco and Emma, Lara, Agustina and Mateo, and Julia and Malena. Three cover the adult B2 First and account for seven more: Giulia, Tabatha and Elena, Julia and Isabela, and Natalia and Enzo. Each document gives criterion by criterion commentary and a score table.

Is pronunciation always the strongest criterion at B2 First?

No. Across the twelve profiles that are not flat, pronunciation is the highest or joint highest mark nine times, so it is common but not a rule. Three candidates break it: Julia and Malena at B2 First for Schools, and Tabatha at adult B2 First. For Julia, pronunciation is her single lowest mark while her Interactive Communication is her highest.

What replaces pronunciation when it is not the top mark?

Interactive Communication, in all three cases. Julia scores 5 for Interactive Communication against 4 for pronunciation. Malena scores 3.5 against 3. Tabatha scores 4 against 3.5. Those are the only three candidates of the fourteen whose single highest mark is Interactive Communication, and they are exactly the three where pronunciation is not top.

What usually drags a B2 First profile down?

Grammar and Vocabulary. It is the lowest or joint lowest mark for ten of the fourteen published candidates. That is consistent with what Cambridge's C1 Advanced sets show, and it is the opposite of where most candidates spend their preparation time, which tends to go on accent work rather than on range and accuracy.

Is B2 First for Schools easier than B2 First?

Cambridge states the two follow the same format and the same criteria, and nothing in the published scores suggests a different standard. What the published sets do show is a much wider spread among the school candidates, from 2 to 5, against 3.5 to 4.5 among the adults. That is a fact about which candidates Cambridge chose to publish, not evidence that either exam is easier.

How many criteria are scored in B2 First speaking?

Five: Grammar and Vocabulary, Discourse Management, Pronunciation, Interactive Communication and Global Achievement. Four of them come from the assessor, the examiner who listens without speaking to you. Global Achievement comes from the interlocutor, the examiner who runs the conversation. At C1 and C2 the grammar and vocabulary criterion splits in two, making six.