"Improve your writing" is bad advice, because IELTS does not give you a writing score. It gives you four scores — Task Response, Coherence and Cohesion, Lexical Resource, Grammatical Range and Accuracy — and averages them. The official scoring page is explicit:
"The criteria are weighted equally and the score on the task is the average." — ielts.org, IELTS scoring in detail
So a 6.5 is not one problem. It is usually one criterion sitting a band below the others and dragging the average down. Each criterion fails in its own way, and the fix for one does nothing for another — vocabulary drills will not repair an essay that answers the wrong question. This page is a diagnostic: four named failure modes, how to detect each in an essay you have already written, and the fix.
The failure
The mistake this page prevents is practising "writing" in general — another essay tonight, another tomorrow — without ever finding out which of the four numbers is low. That way, three quarters of your effort lands on criteria you already score well on. Diagnose first.
1. Task Response — answering the topic instead of the prompt
The current (May 2023) descriptors give Band 6 as:
"The main parts of the prompt are addressed (though some may be more fully covered than others)." — Writing band descriptors, Band 6, Task Response
and Band 7 as "The main parts of the prompt are appropriately addressed. A clear and developed position is presented." The failure mode is writing everything you know about the general topic — education, technology, cities — instead of answering the specific prompt in front of you. The prompt has parts; a prepared topic essay covers the topic and misses the parts.
Detect it: underline each instruction in the prompt — every question asked, every "discuss both views", every "to what extent". Then write the number of each part in the margin of your essay next to the paragraph that answers it. A part with no number in the margin is a main part of the prompt you did not address. Second check: could your essay be submitted, unchanged, for a different prompt on the same topic? If yes, you answered the topic, not the prompt.
Fix it: before writing, turn the prompt into questions and draft a one-sentence answer to each. Those sentences become your position — which Band 7 requires to be "clear and developed" — and each body paragraph exists to justify one of them.
2. Coherence and Cohesion — linkers glued onto a list
Band 6 here says cohesive devices "are used to some good effect but cohesion within and/or between sentences may be faulty or mechanical due to misuse, overuse or omission." Band 7 asks for something different in kind:
"Information and ideas are logically organised, and there is a clear progression throughout the response." — Writing band descriptors, Band 7, Coherence and Cohesion
The failure mode is decorating a list of unconnected points with "Furthermore" and "Moreover" — mechanical cohesion, named in the Band 6 descriptor itself. Linkers do not create progression; they only signal a relationship that must already exist between the ideas.
Detect it: circle every linking word in your essay, then delete them mentally and reread. If the paragraph still makes sense in the same order, the ideas connect and the linkers were signposts. If it becomes a heap of separate sentences, the linkers were glue over a crack. Also count them — a linker starting nearly every sentence is the "overuse" the descriptor flags.
Fix it: plan the argument as a chain — each paragraph's first sentence should be provable from the one before. Then add fewer, more varied devices, including the "reference and substitution" (this policy, such a change, one) that Band 7 wants used "flexibly".
3. Lexical Resource — memorised phrases instead of precision
The failure mode is bolting on rehearsed "academic" phrases — "in this day and age", "a double-edged sword" — that carry no meaning specific to your argument. The descriptors penalise exactly this at the low end: Band 4 Lexical Resource notes "inappropriate use of lexical chunks (e.g. memorised phrases…)", Band 3 notes "over-dependence on… memorised language", and the scale bottoms out at Band 0, reserved partly for cases "where there is proof that a candidate's answer has been totally memorised." Memorisation is not a shortcut the examiners have failed to notice — it is written into the scale.
Detect it: highlight every phrase you knew you would write before you saw the prompt. Then check each highlighted phrase: does it mention anything from this prompt? A phrase that would fit any essay is a template, not vocabulary. Second test: find your vague nouns — "things", "aspects", "factors" — and ask what, specifically, each one means here.
Fix it: replace template phrases with words tied to the prompt's actual subject matter, and practise paraphrasing the prompt's own key words rather than importing stock decoration. Precision, not rarity, is what separates the bands.
4. Grammatical Range and Accuracy — no error-free sentences
Band 6 tolerates errors that "rarely impede communication". The Band 7 line is harder and very concrete:
"Grammar and punctuation are generally well controlled, and error-free sentences are frequent." — Writing band descriptors, Band 7, Grammatical Range and Accuracy
The failure mode is chasing range — long, ambitious sentences — while almost no sentence in the essay is entirely free of error. "Error-free sentences are frequent" is a gate you can measure yourself against.
Detect it: take a marked or corrected essay and tick every sentence with zero errors — grammar and punctuation both. Count ticks, count sentences. If clean sentences are rare rather than frequent, this criterion is your ceiling, whatever your range. Note also which errors repeat: most writers have three or four recurring ones, not thirty.
Fix it: keep the "variety of complex structures" Band 7 asks for, but shorten the sentences you cannot yet control, and drill your specific recurring errors — articles, subject-verb agreement, comma splices — one at a time until they stop appearing.
The two facts that shape your hour
Task 2 counts double. Both official format pages state: "Task 2 contributes twice as much as Task 1 to the Writing score." (The scoring page says only that Task 2 "carries more weight" — the explicit 2x figure lives on the format pages.) The consequence is the official timing itself: about 20 minutes for Task 1, about 40 for Task 2. Never borrow Task 2's minutes to polish Task 1.
The under-length rule is genuinely contradictory. The format pages say: "You must write at least 250 words and will be penalised if your answer is too short." But the current May-2023 descriptors PDF contains no explicit word-count penalty rule — the only hard length rule in the scale is that "Responses of 20 words or fewer are rated at Band 1." Officialdom disagrees with itself here. The safe reading: treat 150 and 250 as real minimums, because the test-taker-facing rule says you will be penalised — but know that examiners rate against the band scale, where a short answer mostly costs you through underdeveloped Task Response.
What actually costs you
- Practising "writing" in general instead of finding your one low criterion — the average hides it.
- A prepared topic essay that never touches the main parts of this prompt.
- "Furthermore" glued between ideas that do not follow from each other.
- Memorised openers the descriptors explicitly flag — down to Band 0 for a totally memorised answer.
- Zero error-free sentences in an essay full of impressive structures.
- Spending Task 2's minutes on Task 1, when Task 2 contributes twice as much.
Try one
Diagnosis needs a marked essay — you cannot tick your own error-free sentences reliably on the first pass.
[Write a marked Task 2 →] — one full task, marked, no card required.