Your Writing band is not one impression — it is four separate scores, averaged. Examiners score every essay against four criteria, each worth exactly 25%, and each one is a dial that specific things move and specific things do not. This lesson translates the official descriptors into plain language, so you stop spending effort on things that move nothing.
Four dials, each worth 25%
| Criterion | What the examiner is really asking |
|---|---|
| Task Response (TR) | Did you answer this question, completely, with a clear position and developed ideas? |
| Coherence & Cohesion (CC) | Can I follow your argument without effort? Is each paragraph one idea? |
| Lexical Resource (LR) | Do your words say precisely what you mean, with few errors? |
| Grammatical Range & Accuracy (GRA) | Do you control a mix of sentence structures — control, not just attempt? |
Your Writing band is the average of the four, rounded down to the nearest half band. That rounding direction matters: TR 7, CC 7, LR 6, GRA 6 averages 6.5 — but TR 7, CC 6, LR 6, GRA 6 is a 6.0. One weak dial drags the whole section, and no strong dial can hide it.
Task Response: the dial you control before writing a word
| Moves it up | Moves it nothing (or down) |
|---|---|
| Answering every part of the question | Writing brilliantly about a similar question |
| A position stated clearly and kept to the end | Sophisticated vocabulary |
| Each idea explained and supported, not just listed | A memorised introduction that restates nothing specific |
| Relevant, concrete examples | Extra length |
TR is decided in the two minutes before you write: did you read the question as asked, decide a position, and choose ideas that answer it? A band 5.5 TR is usually not a language failure — it is answering "should governments fund public transport?" with a prepared essay about "technology in society."
Coherence & Cohesion: one idea per paragraph
| Moves it up | Moves it nothing (or down) |
|---|---|
| One central idea per body paragraph | Stuffing in Moreover, Furthermore, Nevertheless everywhere |
| Ideas in a logical order the reader can predict | Linking words used mechanically or wrongly |
| Referencing (this policy, such a change, it) instead of repetition | Very long paragraphs covering three ideas |
| A paragraph break exactly where the idea changes | Linking every single sentence |
Examiners explicitly mark down cohesive devices that are overused or mechanical. Band 7 CC reads as effortless to follow, not decorated with connectors. If your paragraph plan is clear, you need surprisingly few linking words.
Lexical Resource: precision beats rarity
| Moves it up | Moves it nothing (or down) |
|---|---|
| The precise common word (reduce traffic congestion) | A rare word used slightly wrongly (mitigate the vehicular multitude) |
| Natural word partnerships (heavy traffic, take responsibility) | Memorised "band 9 word lists" forced into sentences |
| Accurate word forms (economy / economic / economical) | Repeating the question's exact words all essay |
| Few spelling errors | Long words with spelling errors |
The descriptor rewards words that are appropriate and accurate, and it punishes errors. A long word used at 80% accuracy scores worse than a common word used at 100%. Build precision with words you already half-know — the glossary exists for exactly this.
Grammatical Range & Accuracy: control, not decoration
| Moves it up | Moves it nothing (or down) |
|---|---|
| A mix of simple and complex sentences, mostly error-free | Long sentences that collapse halfway |
| Complex structures you actually control (one if-clause, one relative clause) | Five subordinate clauses in one sentence |
| Correct articles, tenses, agreement — the boring accuracy | "Advanced" structures copied from templates, with errors |
| Punctuation that separates sentences properly | Comma splices everywhere |
The word in the descriptors is error-free sentences. Band 7 requires that many of your sentences contain no errors at all. Range means variety you control — two clause types used correctly beat six used wrongly.
Three myths, killed
Myth 1: long words raise your band. LR rewards precision and accuracy. An essay full of misused rare words scores lower than the same argument in plain, exact English, because errors are counted.
Myth 2: a memorised template raises your band. Examiners are trained to recognise memorised language, and the descriptors instruct them not to credit it. A template can carry the two or three sentences it fits — the rest of your essay is still you, and TR still depends on ideas the template cannot supply. (Lesson 3 shows exactly where the template ceiling sits.)
Myth 3: longer essays score higher. Below 250 words you are penalised. Above roughly 280, every extra sentence is extra error risk and thinner development for zero reward. The examiner is scoring quality of development, not volume. 260–280 focused words is the professional target.
Your drill (15 minutes)
- Take one essay you have already written. No essay yet? Write one now from a real current question — pick a Task 2 from this season's reported writing questions.
- Score it against the four tables above, one dial at a time. For each dial write one line: what moved it, what wasted effort.
- Count your error-free sentences. Divide by total sentences. Under 50%? Your fastest gains for the next month are GRA accuracy, not vocabulary.
- Check the recall wall to see what the test is currently asking — the next three lessons build the method, and they train best on live questions.