The writing paper has two tasks and they are not equal. Task 2 carries two thirds of the mark for the paper and Task 1 carries one third, and the published timings match that: twenty minutes for the first and forty for the second.
Most candidates spend roughly equal time on the two, because Task 1 comes first and because describing a chart feels finishable in a way that an argument does not. That single misallocation costs more band than any vocabulary gap, and correcting it takes one decision rather than any study at all.
Eight facts about the format
- bands from 1 to 9, reported in half bands, with a separate score for each paper and an overall average.
- Academic and General Training. The first is for university, the second for immigration and work, and the reading and writing papers differ between them.
- 6.5 is a common university requirement, often with a minimum in each paper, which is the condition candidates most often miss.
- four papers, one score each: listening, reading, writing and speaking.
- the speaking test is a real conversation with an examiner, face to face, which suits some candidates and unsettles others.
- part 2 is a two minute talk from a card, with one minute to prepare, and it is the part that most rewards rehearsal.
- the writing has two tasks, a description of a chart and an essay, and the second is worth twice the first.
- the listening is played once, with no repeat, which is why connected speech practice matters more here than any vocabulary work.
Four papers, and only one of them is a format
Listening, Reading, Writing and Speaking, reported on a nine-band scale in half bands. Three of the four are ordinary skills being measured under time pressure, and the practice that improves them is the practice that improves them anywhere.
The exception is Reading, which is a format as much as a skill: the question types are fixed, several of them reward scanning rather than reading, and matching headings to paragraphs or deciding between false and not given are exam techniques that transfer to nothing outside the room. Those are worth drilling specifically. The other three are not.
Two versions exist, Academic and General Training, and they differ in the Reading and Writing papers rather than throughout. Which version you need is determined by what you are applying for, and the Academic paper is not the harder one.
Speaking is three parts, and the band is decided in the third
The Speaking test is conducted by a person, in three parts: personal questions, a short talk from a prompt card, and a discussion of the abstract issues behind that topic.
The third part is where the band is decided, because it is the only part that requires you to develop an argument, qualify a position and speculate. It is also the part almost nobody prepares, because the first part is easy to rehearse: candidates arrive with polished answers about their home town and then meet a question about whether governments should fund the arts.
Rehearsing part one has a second cost. A memorised answer sounds memorised, and an examiner who suspects rehearsal moves off the script early, which is exactly the situation the rehearsal was avoiding.
Where it stops: preparing the part that is already fine
The stall is putting the preparation where the anxiety is rather than where the marks are. The anxiety sits in part one of Speaking and in Task 1 of Writing, because both come first and both are the moments where the test starts going wrong or does not.
The marks sit in part three and in Task 2. A candidate can rehearse the opening of both papers to a high polish and still lose a band in the sections that carry the weight, and nothing in the preparation will have told them, because the practice sessions all felt productive.
The document that says why an essay is a six
The public band descriptors for Writing and Speaking are published free, and they are the most valuable free document in this subject. They set out, band by band, exactly what a marker is looking for under each of four criteria, in specific terms: whether the position is clear throughout, whether paragraphing is logical, whether the vocabulary is used precisely rather than merely present.
They replaced a teacher's opinion about why an essay was a six rather than a seven, which is an opinion however experienced the teacher is, and which usually arrives as advice to use more advanced vocabulary. The descriptors say something much more specific, and they say it in the words the marker will be using.
Their limitation is that they are the public versions, which are shorter than the ones examiners work from, so they tell you the criteria and not every distinction inside a band. That is still far more than most candidates ever read.
What it looks like when it goes wrong
The recognisable failure is a Task 2 essay that runs out of time and ends mid-argument. It happens because the candidate spent thirty minutes on Task 1 making the chart description perfect, and a chart description cannot be perfect enough to be worth ten minutes of the essay.
The second is a Task 2 that answers a nearby question rather than the one asked. The prompts are precisely worded and often ask for two things, and an essay that covers one of them beautifully is being marked against a criterion it did not meet.
The skill under Task 2, once the format is stripped away, is thesis, argument and evidence, in the shape examiners expect, and the abstract discussion in the third part of Speaking is the same demand made out loud, which is speed without accuracy, accuracy without speed, and how they meet.
What none of this settles is whether this is the right test to be sitting, since the marking here is done by people and there is a machine-marked alternative that rewards a different performance: what each test scores, and how to prepare for yours.