Skip to main content
Writing

How to Improve IELTS Writing: What Examiners Actually Reward

Stuck below band 7 in writing? Learn the exact examiner criteria holding your score back and the targeted moves that lift both Task 1 and Task 2.

25 June 2026 9 min read By BandNine Editorial

Most candidates who stall at band 6 in IELTS Writing are not making random mistakes — they are making the same predictable mistakes, over and over, because nobody has explained what the examiner is actually looking for. This guide fixes that.

#The four criteria, and why most candidates only half-understand them

Every IELTS Writing Task 2 script is marked against four equally weighted criteria: Task Achievement, Coherence and Cohesion, Lexical Resource, and Grammatical Range and Accuracy. Each criterion contributes equally to your final writing score. That last part is the bit candidates keep forgetting.

The practical consequence is this: obsessing over vocabulary while ignoring argument structure will cap your score. A script with sophisticated word choices but a wandering, under-supported argument will not reach band 7, because Task Achievement alone will drag the average down. You need to perform reasonably well across all four, not brilliantly on one.

#What band 7 actually requires, descriptor by descriptor

The public band descriptors — freely available from the British Council and IDP — are specific, and candidates who have read them carefully have a real advantage over those who have not. Here is what band 7 asks for in plain language:

  • Task Achievement: The response addresses all parts of the task. The position is clear throughout, and main ideas are extended and supported, even if this is not always done with complete precision.
  • Coherence and Cohesion: Information and ideas are logically organised. There is clear progression throughout. A range of cohesive devices is used flexibly, though there may be occasional lapses.
  • Lexical Resource: A sufficient range of vocabulary is used with flexibility and precision. Less common lexical items are used with awareness of style and collocation, with only occasional inaccuracies.
  • Grammatical Range and Accuracy: A variety of complex structures is used with a good degree of control and flexibility. Errors occur but rarely impede communication.

Notice what none of these descriptors say: they do not require perfect grammar, zero errors, or a certain number of complex sentences per paragraph. They reward control and flexibility, not mechanical perfection. That is an important distinction.

#Task Achievement: the criterion candidates underestimate most

Task Achievement is the examiner's first question: did this person actually do what the question asked? It sounds obvious. It trips up an enormous number of candidates.

The two most common failures are partial task coverage and an unclear or shifting position.

#Partial task coverage

Many questions are compound. "Discuss both views and give your own opinion" is not one instruction — it is three. Candidates routinely discuss both views at some length but then tack on a single vague sentence as their opinion, or they spend so much time on their preferred side that the other view gets a cursory mention. The examiner reads the whole script with the question in hand. Imbalance is immediately visible.

The fix is structural and takes roughly thirty seconds of planning. Before you write, identify every distinct sub-task in the prompt, then allocate at least one developed paragraph to each.

#A clear, sustained position

For opinion essays specifically, a near-universal weak point is the wavering thesis. A candidate writes "I partially agree" in the introduction, then argues strongly for one side throughout the body, then tries to balance things out again in the conclusion. The examiner sees a position that shifts, which reads as confused thinking rather than nuanced thinking.

Nuance is good. Contradiction is not. If you partially agree, your body paragraphs should actually reflect that — one paragraph supporting, one qualifying, with explicit signposting of the distinction.

#Coherence and Cohesion: where band several essays visibly plateau

Coherence is about the logic of your argument. Cohesion is about the language tools — pronouns, connectives, referencing — that hold sentences together. Candidates conflate them and, in doing so, treat both as simple vocabulary problems.

The actual problem at band 6 is almost always mechanical cohesion paired with weak coherence. The essay is studded with connectives — "Furthermore," "Moreover," "In addition," "Nevertheless," — but the ideas themselves do not build on each other. Adding a connective to a sentence that does not logically follow from the previous one does not create coherence. It creates the appearance of structure over the absence of it.

#The over-reliance on listing connectives

The band-7 Coherence and Cohesion descriptor calls for cohesive devices that are used flexibly, not mechanically. A script that opens every body sentence with a listing adverbial — "Firstly," "Secondly," "Additionally," "Finally," — will not demonstrate flexibility. It demonstrates a formula.

The fix is to vary your cohesive tools: use pronoun referencing ("This argument rests on…"), lexical chains (referring back to a key concept with a synonym or superordinate term), and clause-internal connectors ("although," "whereas," "given that") as well as sentence-initial adverbials.

#Before and after: paragraph-level coherence

Before (band 6 level): "Social media has many advantages. Firstly, it allows people to communicate easily. Furthermore, it helps businesses to reach customers. Moreover, young people use it for entertainment. However, there are also disadvantages."

After (band 7 level): "One of the most significant advantages of social media is the way it collapses distance in communication — a small business in a provincial town can now reach customers nationally with no advertising budget to speak of. This accessibility, however, comes with a structural trade-off: the same openness that benefits legitimate commerce also provides infrastructure for misinformation to travel at scale."

The second version uses fewer connectives but demonstrates far stronger coherence. Each sentence develops from the previous one. The contrast is introduced through the logic of the argument, not announced with "However."

#Lexical Resource: the vocabulary mistakes that actually cost marks

The instinct of most candidates preparing to improve their writing score is to memorise vocabulary lists. This is not wrong, but it targets the wrong problem. The Lexical Resource criterion penalises inaccurate use far more than it rewards ambitious use. An incorrect collocation or a misapplied formal synonym does active damage to your score.

#Collocation errors are among the costliest mistakes

Collocation — the way words naturally pair with each other — is what separates a fluent writer from a competent one. Candidates who have learned that "considerable" is a good formal adjective will write "considerable damage," which is natural, but also "a considerable idea," which is not. The band-7 descriptor specifically calls for "awareness of style and collocation." Mistakes here are noticed.

The most practical habit you can build is not memorising isolated words but memorising words in their natural chunks: "raise awareness," not just "awareness"; "reach a conclusion," not just "conclusion"; "impose restrictions on," not just "restrictions."

#Paraphrasing the question prompt: do it, but do it carefully

Copying the exact wording of the question prompt into your introduction is penalised under Lexical Resource. Paraphrasing it is expected. The problem is that candidates often paraphrase inaccurately — changing words in a way that subtly shifts the meaning of the prompt. This creates a mismatch between the question being asked and the one being answered, which then also damages Task Achievement.

Paraphrase for vocabulary variation, but re-read your paraphrase against the original to make sure the core claim is intact.

#Grammatical Range and Accuracy: what "range" actually means

Grammatical Range and Accuracy is the criterion most candidates feel most anxious about, and that anxiety is, to some extent, misplaced. The band-7 descriptor says errors occur but "rarely impede communication" — it does not say errors are absent. The more important word in the descriptor is range.

Range means that your script should demonstrate varied sentence structures: relative clauses, conditional structures, passive constructions where appropriate, nominal clauses, participle phrases. A script written entirely in short, simple, accurate sentences will demonstrate accuracy but not range, and will not satisfy the descriptor.

#The common trap: accuracy at the expense of complexity

Candidates who have been burned by grammar errors sometimes retreat to short, simple sentences because they feel safer. The result is a script that reads mechanically and fails to demonstrate the structural variety the examiner is looking for. The band descriptor for range explicitly calls for "a variety of complex structures."

The better approach is to use complex structures deliberately and check them. Write the sentence, then reread it. If the relative clause or conditional you have used is awkward, simplify it. But do not avoid complexity across the board — that trades one problem for another.

#Before and after: simple versus ranged grammar

Before: "Cities are growing fast. This causes many problems. Governments need to find solutions. Some cities have built new transport systems. This has helped reduce traffic."

After: "The rapid growth of urban populations has generated a set of infrastructural pressures that many city governments were not designed to handle. Where authorities have invested in integrated public transport networks, congestion has eased — though rarely to the degree that initial projections promised."

The second version uses a noun clause, a relative clause, a conditional-style adverbial clause, and a concessive structure, all in two sentences. This is what grammatical range looks like in practice.

#The planning habit that separates band 6 from band 7

IELTS Writing Task 2 gives you approximately forty minutes. Candidates who use none of that time planning are almost always the ones whose essays drift off-topic, lose their argument mid-essay, or fail to address all parts of the question. Five minutes of focused planning is not lost writing time — it is the investment that makes the writing time more efficient.

A useful planning framework involves three steps: identify every sub-task in the question, decide your position or approach, and then map one idea per body paragraph with a supporting example or piece of reasoning. You do not need to plan every sentence. You need to know, before you start writing, what each paragraph is for and how it connects to your overall answer.

Candidates who plan in this way tend to produce essays with clearer coherence, fuller task coverage, and fewer half-abandoned ideas — which directly addresses three of the four marking criteria simultaneously.

#How to practise in a way that actually improves your score

Writing essays without feedback is the most common and most wasteful form of IELTS Writing practice. If you repeat the same structural habits in every essay, you will consolidate those habits regardless of whether they are good or bad ones. What changes scores is targeted feedback against the band descriptors, followed by deliberate revision.

The most effective practice loop is: write a timed essay under exam conditions, receive criterion-by-criterion feedback, identify the one or two specific weaknesses flagged most often, work on those in isolation (for instance, writing a single paragraph and checking its cohesive devices), then write another full essay. This is slower than simply producing volume, but it is considerably more effective.

Reading examiner reports published by the British Council and IDP is also underutilised. These reports identify — at a general level — the patterns that distinguish higher-scoring scripts from lower-scoring ones in each sitting. They are not specific to your script, but they are based on actual marking data, which makes them more reliable than generic advice.

If you want to put this into practice immediately, BandNine's AI writing tool gives you criterion-by-criterion feedback on every Task 2 essay you submit, mapped directly to the band descriptors covered in this guide. You can take the free diagnostic to see where you currently stand across all four criteria and get a personalised starting point for your preparation.

Found this useful?

B9

BandNine Editorial

Written and reviewed by the BandNine team — IELTS practitioners and language-assessment researchers building the AI examiner. Our guidance is grounded in the official public IELTS band descriptors and the real mistakes we see in candidates' work.

Stop reading about IELTS. Start practising it.

Get AI-scored feedback on your writing and speaking in 30 seconds.

Start free

Keep reading

All articles →