IELTS Writing Task 2: What the Examiner Is Actually Marking

Here is the strange thing about IELTS writing preparation: the marking criteria are public — published by the test's own partners as band descriptors anyone can download — and yet most candidates prepare as if the score were a mystery, collecting "high-band phrases" and template introductions from videos of uneven quality. If you read what examiners are actually instructed to reward, most of that folklore collapses, and preparation gets simpler and more effective at the same time.

Four criteria, equally weighted

Task 2 is marked on four criteria, each worth a quarter: task response (did you answer the whole question, with a clear position and developed ideas), coherence and cohesion (does the essay progress logically, with paragraphs that each do one job), lexical resource (range and precision of vocabulary), and grammatical range and accuracy. The practical consequence is immediate: a candidate who writes memorised "advanced" vocabulary into an essay that only half-answers the question is trading a quarter of the marks for a fraction of another quarter. The question, not the thesaurus, is the main event.

Task response is where bands are lost

Read prompts forensically, because most contain two instructions: "discuss both views and give your opinion" is three tasks, not one; "to what extent do you agree" requires a stated extent, sustained through the essay, not discovered in the conclusion. The most common ceiling on scores is an essay that addresses part of the prompt well and ignores the rest. Before writing, spend three minutes planning: position, two or three main ideas, one line of support each. An examiner can see a planned essay in its paragraph structure within seconds.

Cohesion is structure, not connectors

Folklore says sprinkle "moreover" and "furthermore" generously. The descriptors say the opposite: mechanical overuse of linking words is explicitly a feature of the middle bands, while the higher bands manage progression so naturally that cohesion barely shows. One clear idea per paragraph, a first sentence that states it, and sentences that each follow from the last will do more for this criterion than any connector list. The same logic applies to vocabulary: precision beats rarity. The descriptors reward words used accurately and naturally, including less common ones — not rare words used almost correctly, which examiners are specifically trained to notice.

Grammar: range you control

The grammar criterion rewards a mix of simple and complex sentences with high accuracy — not maximal complexity. A long sentence collapsing under three subordinate clauses costs more than two clean sentences saying the same thing. In practice, most candidates raise this score fastest by fixing their three commonest recurring errors (articles, subject-verb agreement and comma splices lead the list for Indian candidates) rather than by adding structures they cannot yet control.

So prepare like this: download the public band descriptors and read them once a week; write full essays under time; then mark your own work criterion by criterion, or have someone competent do it, and fix the specific criterion that is lagging. Two essays a week marked this way beat ten written blind. The examiner is not looking for brilliance — they are matching your writing to published descriptions. You are allowed to read the descriptions. Start there.