How IELTS writing is scored should not be a mystery, because the marking criteria are public documents that anyone can download from IELTS.org — yet most candidates have never read them, and prepare instead on folklore: longer essays score higher, big words impress examiners, more linking words mean more cohesion.
All three beliefs are wrong, and the descriptors say so in writing.
This guide decodes the four criteria your examiner actually uses, explains in plain language what separates Band 6 from Band 7 from Band 8 on each one, exposes the hard caps that catch candidates by surprise, and shows how to turn the descriptors from an official PDF into a personal diagnostic tool.
It is the sister piece to our breakdown of how IELTS Speaking is scored, and the two systems share the same architecture — read them together and the whole productive side of the test stops being a black box.
The four criteria, each worth 25 per cent
Every IELTS Writing task is marked on four criteria, and each contributes exactly a quarter of the task score. There is no secret fifth criterion, no bonus for handwriting or ambition, and no criterion that matters more than the others.
The first criterion changes name by task: for Task 1 it is Task Achievement — did you do the specific job the task set, covering every required element accurately — and for Task 2 it is Task Response — did you answer the actual question with a clear position and developed ideas.
The second is Coherence and Cohesion: whether your ideas are logically organised, progress clearly, and connect through appropriate linking, referencing and paragraphing. The third is Lexical Resource: the range, precision and accuracy of your vocabulary, including collocation and spelling.
The fourth is Grammatical Range and Accuracy: the variety of sentence structures you attempt and the density of errors in them.
The examiner awards a whole band for each criterion — there are no half bands at criterion level — by matching your writing against the published descriptor for each band. This is worth internalising because it changes what "improving your writing" means.
You do not have one writing ability; you have four, and they are scored independently. A candidate can hold Band 7.5 ideas in Band 6 grammar, or produce nearly flawless sentences that never quite answer the question.
Knowing which of your four profiles is weakest is the single highest-value piece of information in Writing preparation, and the rest of this article is designed to help you find it.
How your two tasks combine into one Writing band
Each task is assessed separately: four criterion bands for Task 1, four for Task 2. The criterion scores are equally weighted within each task, giving each task its own band.
The two task bands are then combined with Task 2 counting double — it contributes twice the weight of Task 1 to your final Writing band, which is then reported on the familiar scale that includes half bands.
The exact arithmetic examiners apply is internal, but the weighting itself is official and public, and its practical consequences are blunt.
First, the time allocation of 20 minutes for Task 1 and 40 minutes for Task 2 is not a suggestion; it matches the marks. Spending 35 minutes perfecting a Task 1 letter or report while rushing the essay is trading a heavy criterion set for a light one.
Second, any systematic weakness hurts you eight times, because the same four criteria are applied to both tasks — a grammar problem does not stay in one essay. Third, a disaster on Task 2 cannot be rescued by a brilliant Task 1, but the reverse is partially survivable.
Every preparation decision should flow from this weighting, which is why our guide to IELTS Writing Task 2 structure treats the essay as the centre of gravity of the whole Writing test.
Task Achievement and Task Response between Bands 6, 7 and 8
On Task Response, the official IELTS.org Writing Task 2 band descriptors draw the Band 6 to 7 line at completeness and clarity of position.
A Band 6 essay addresses all parts of the question but unevenly — one part gets a full paragraph, another a passing sentence — and while it takes a relevant position, the conclusions may become unclear or repetitive. Its main ideas are relevant but some are underdeveloped.
A Band 7 essay addresses all parts properly, holds a clear position from introduction to conclusion, and presents ideas that are extended and supported, though it may still over-generalise or lose focus in places.
Band 8 is the same architecture executed fully: a well-developed response whose ideas are relevant, extended and well supported throughout.
On Task Achievement for Task 1, the equivalent ladder in the IELTS.org Writing Task 1 band descriptors runs on coverage and selection.
Band 6 addresses the requirements and presents the key features or bullet points, but details may be irrelevant, inappropriate or inaccurate, and in General Training letters the tone may be inconsistent.
Band 7 covers the requirements with a clear purpose and consistent tone, presenting key features clearly but not always extending them. Band 8 covers every requirement sufficiently, with key features highlighted and illustrated appropriately.
The pattern across both tasks is identical, and it is not about language at all. The first criterion measures whether you did the assigned job completely and proportionately.
This is why misreading the question is the most expensive mistake in IELTS Writing: a beautifully written essay on a slightly different question has already lost a quarter of its marks before a single sentence is judged.
Coherence and Cohesion between Bands 6, 7 and 8
Coherence and Cohesion is the criterion candidates most consistently misunderstand, because they equate it with linking words. The descriptors say otherwise.
A Band 6 answer arranges ideas coherently with a clear overall progression, but its cohesion is described as potentially "faulty or mechanical" — connectors bolted onto sentences whether needed or not — and its referencing and paragraphing may be unclear or illogical.
A Band 7 answer organises ideas logically with clear progression throughout, uses a range of cohesive devices appropriately even if some are over- or under-used, and gives each paragraph a clear central topic.
Band 8 sequences ideas logically, manages all aspects of cohesion well, and paragraphs sufficiently and appropriately — at which point the cohesion has become invisible.
Read that ladder carefully and the operative word is mechanical. Stacking "Moreover", "Furthermore" and "In addition" at the start of consecutive sentences is not climbing towards Band 7; it is exhibiting the named Band 6 trait.
Progression, paragraph discipline and precise referencing do more for this criterion than any list of connectors. We unpack the whole topic — including a before-and-after paragraph showing over-linked writing repaired — in our guide to IELTS linking words and cohesive devices.
Lexical Resource between Bands 6, 7 and 8
Lexical Resource rewards range, but always range in the service of precision. A Band 6 answer uses an adequate range of vocabulary for the task and attempts less common words, but with some inaccuracy, plus spelling or word-formation errors that do not block understanding.
A Band 7 answer commands enough vocabulary to show flexibility and precision, uses less common items with some awareness of style and collocation, and slips only occasionally.
A Band 8 answer uses a wide range fluently and flexibly to convey precise meanings, handles uncommon items skilfully, and makes only rare errors.
Two implications follow. First, the ladder from 6 to 7 is not "use rarer words" — it is "use less common words correctly, in the right style, with the right partners".
The word collocation enters the descriptors at Band 7 precisely because word partnerships are where memorised vocabulary fails: a candidate who writes "make a research" or "big rain" has demonstrated reach without control, which is the Band 6 profile.
Second, spelling genuinely counts here, which surprises candidates who think of it as a formality. Systematic spelling errors sit inside this 25 per cent, not outside the marking.
Grammatical Range and Accuracy between Bands 6, 7 and 8
Grammatical Range and Accuracy has the clearest band boundary in the whole rubric, and it is worth memorising. A Band 6 answer uses a mix of simple and complex sentence forms and makes some errors in grammar and punctuation, but they rarely stop the reader understanding.
A Band 7 answer uses a variety of complex structures and — this is the hinge — produces frequent error-free sentences, with good control and only a few persistent errors.
A Band 8 answer uses a wide range of structures with the majority of sentences error-free, and slips only very occasionally. In other words, the question the examiner is implicitly asking as they read is brutally countable: how many of your sentences contain no error at all?
At Band 6 the errors are frequent but tolerable; at Band 7 clean sentences are frequent; at Band 8 clean sentences are the norm. The table below condenses all four criteria into one reference you can check your own writing against.
| Criterion | Band 6 in essence | Band 7 in essence | Band 8 in essence |
|---|---|---|---|
| Task Achievement / Response | All parts touched but unevenly; position or details can blur or repeat | All parts covered; clear position or purpose held throughout; ideas extended | Well-developed response; every requirement covered, supported and illustrated |
| Coherence and Cohesion | Clear overall shape but linking is faulty or mechanical; paragraphing wobbles | Clear progression throughout; devices used appropriately; one clear topic per paragraph | Logical sequencing; all aspects of cohesion managed well; paragraphing apt |
| Lexical Resource | Adequate range; less common words attempted with some inaccuracy | Flexibility and precision; some awareness of style and collocation; occasional slips | Wide range used fluently; uncommon items handled skilfully; rare errors |
| Grammatical Range and Accuracy | Mix of simple and complex forms; errors present but rarely block meaning | Variety of complex structures; frequent error-free sentences | Wide range of structures; majority of sentences error-free |
The hard caps candidates never hear about
Beyond the four ladders, the marking system contains hard limits that operate regardless of language quality, and they cause the most painful surprises on results day.
The first is length: Task 1 requires at least 150 words and Task 2 at least 250, and under-length answers are penalised under Task Achievement or Task Response. No skill elsewhere compensates, because the penalty lands before range or accuracy are weighed.
The second is topic: an essay that answers a different question from the one asked — usually through misreading a single word in the prompt — cannot score well on Task Response, since that criterion begins with whether the actual task was addressed.
The third cap concerns memorised content. Examiners are trained to recognise rehearsed, formulaic passages, and memorised material is penalised because it does not demonstrate your ability to respond to the specific question in front of you.
Related to this, wording lifted directly from the question paper does not showcase your own language — copying the prompt into your introduction demonstrates transcription, not English. The safe path is genuine paraphrase, which is a trainable skill in its own right.
Finally, answers written in note form or bullet points suffer, because paragraphing and organisation are explicitly assessed under Coherence and Cohesion; the test wants continuous prose.
None of these caps punishes weak English. They punish process errors — misjudged time, misread prompts, borrowed language — which is precisely why they are worth knowing about in advance: every one of them is preventable by habit rather than by years of study.
How to use the descriptors diagnostically
Here is the practical payoff. Because the four criteria are scored independently, your Writing band is not one number but a profile, and profiles are rarely flat.
In practice most candidates have a signature: ideas that outrun their grammar, or clean sentences arranged in shapeless paragraphs, or accurate but narrow vocabulary that caps Lexical Resource at 6 while everything else sits at 7. Averaging hides the signature; the descriptors reveal it.
If you can name your weakest criterion, you can train it specifically — an error log and proofreading drills for Grammatical Range and Accuracy, planning and paragraph discipline for Coherence and Cohesion, collocation-focused vocabulary work for Lexical Resource, idea development and question analysis for Task Response.
The obstacle is measurement: you cannot see your own profile by rereading your essays, because you wrote them and your brain fills the gaps.
This is exactly the problem the IELTSbiz AI writing checker is built for — it returns a band estimate with feedback on each of the four criteria for every essay you submit, so patterns emerge across attempts instead of staying invisible.
One flagged weakness is a data point; the same criterion flagged across five essays is a diagnosis.
From there, our guide on how to improve IELTS Writing lays out targeted work for each criterion, and the British Council's overview of understanding your IELTS scores is a useful official reference for how the reported bands fit together across the four skills.
Work this way and the descriptors change role: instead of a rubric that judges you after the fact, they become a checklist you write towards — all parts of the question, one clear topic per paragraph, precise words in correct partnerships, and as many error-free sentences as you can honestly produce.
Conclusion
How IELTS writing is scored is public, precise and stable: four criteria at 25 per cent each, applied to both tasks, with Task 2 counting double and whole bands awarded per criterion.
Band 7 is not Band 6 plus decoration — on every criterion it names a specific, observable shift: complete and even coverage, clear progression without mechanical linking, precision and collocation awareness, frequent error-free sentences.
Add the hard caps — length, topic, memorisation, note form — and you have the entire system on one page. Candidates who read the descriptors stop guessing what examiners want and start practising against the actual standard.
Find your weakest criterion, train it deliberately, measure again, and your Writing band stops being a lottery and becomes an engineering problem.