Writing in the Final Week
In the final week the goal is not improvement, it is reliability. Two timed tasks a day and a disciplined way of reading the feedback will do more than new material.
What the last week can and cannot change
Seven days is not enough to raise your grammar. It is enough to change three things that decide bands: whether you cover every bullet, whether you commit on Task 2, and whether you finish with three minutes to check.
That is a useful narrowing. Those three are decisions, they are all inside Task Fulfilment and Content and Coherence, and together those criteria are 55 per cent of the score. Working on anything else this week has a worse return.
So the week's target is reliability rather than improvement. You are trying to make your best performance repeatable under a timer, not to reach a level you have not reached before. The most common final-week mistake is starting new material, which produces uncertainty rather than fluency.
The daily session: two tasks, timed
One Task 1 and one Task 2 each day, written on the clock, at 27 and 26 minutes. Fifty-three minutes of writing, and about twenty minutes afterwards for review. Not more: a two-hour session produces one tired answer and a reluctance to sit down tomorrow.
Time both tasks properly, including the internal deadlines. Stop writing Task 1 at minute twenty-four and check for three. Stop Task 2 at minute twenty-three. The habit you are building is the stop, because on the day the timer will not ask you whether the sentence is finished.
Type into a plain box with no spellchecker, on the keyboard you will actually use. This matters more than it sounds. The words you habitually mistype are invisible to you while autocorrect exists, and they become visible only when it does not.
Vary the prompt type across the week so you are not rehearsing one scenario. Alternate formal and informal Task 1 prompts, and take Task 2 surveys on different subjects, because a memorised topic is the fastest way to write something that does not answer the question.
Reading the feedback properly
Feedback is only useful if it changes the next attempt, and the way to guarantee that is to convert it into one instruction before you write again.
Start with the criterion scores rather than the overall band. The overall number tells you where you are; the four separate numbers tell you what to do. If Task Fulfilment is your lowest, the fix is coverage and commitment, and no amount of vocabulary work will move it.
Then read the weaknesses and find the one that appears most often across your last three attempts. A weakness that recurs is a habit, and habits are what a week can change. A weakness that appeared once is noise.
Convert that recurring weakness into a single sentence you write at the top of tomorrow's page. Give the administrative bullet a number, a date and an amount. Name my option in the last sentence. Stop at minute twenty-four. One instruction, not five, because five instructions are none.
Where feedback includes a rewrite of your own text, compare it to yours sentence by sentence rather than reading it as a model. The useful question is not is this better, it is what specifically changed: a category word made into an event, an intensifier made into a consequence, a missing fact supplied.
A week that works
Days one and two. Two timed tasks each day, full review, and identify your weakest criterion from the scores. Do not try to fix anything yet; you are gathering evidence.
Days three and four. Two timed tasks each day with a single written instruction at the top of the page, drawn from the recurring weakness. Review only against that instruction and the criterion scores.
Day five. Two timed tasks, then a rewrite session: take your weakest answer of the week and rewrite one paragraph at the next band, by supplying specifics and cutting padding. Rewriting is where the learning consolidates.
Day six. One full writing section, both tasks back to back, exactly as on the day, with no pauses between them. This is the only session that tests whether your Task 2 is still good after Task 1 has used your concentration.
Day seven. No writing. Re-read your own best two answers and your one instruction. Confirm the mechanical things you will do on the day: read the scenario twice, plan for four minutes, stop at minute twenty-four, run the four checking passes.
Write one instruction at the top of the page each day, drawn from the weakness that recurs across your last three attempts. One instruction changes behaviour; five change nothing.
What to stop doing this week
Stop learning new vocabulary. A word learned on Tuesday will be used incorrectly on Saturday, and advanced vocabulary used incorrectly does not score high. The words you already own and can place confidently are the ones that will earn marks.
Stop reading model answers passively. Reading a good answer feels productive and changes nothing. If you read one, write one immediately afterwards on a different prompt.
Stop untimed practice. An answer written in forty minutes tells you nothing about an answer written in twenty-seven, and it can build a pace you cannot sustain.
Stop collecting templates. Adding a new stored opening in the final week is how a memorised sentence ends up in a prompt it does not fit.
Stop practising after the evening before. A last-minute session the night before does not consolidate, and it produces the tiredness that costs you the three checking minutes.
On the day itself
Read the scenario twice before you plan. Almost every fidelity error comes from reading once and writing from memory, and the second read costs twenty seconds.
Write your four-minute plan even though it feels like time you do not have. Register, three bullet labels, every invented fact. On Task 2: option, two reasons from different stakeholders, one example, one concession.
Give Task 2 the same effort as Task 1. It is the second task, you will be tired, and the temptation is to treat it as the shorter one. It is worth the same.
Stop writing with three minutes left and run the four passes in order: task checks, verbs, sentence boundaries, surface. Do not upgrade vocabulary in those three minutes.
If a task goes badly, let it go completely before the next one starts. The two tasks are scored on the same criteria but they are separate pieces of writing, and carrying a bad Task 1 into Task 2 turns one weak response into two.
Practise on a plain box with no spellchecker, on the keyboard you will use. The words you habitually mistype are invisible while autocorrect exists, and the exam box does not have it.
Practise it
Five questions on how to spend the last seven days, and on what to stop doing.