How IELTS Writing is really scored: a complete guide from Band 6 to Band 8

IELTS Writing is the most misunderstood section of the most widely taken English test in the world. Thousands of candidates who should be scoring 7 or above are stuck at 6 or 6.5, not because their English is weak, but because they have never been shown precisely how their scripts are marked. Two candidates with essentially the same English level can score a full band apart on the same task, purely because one understands the descriptors and the other does not.

This post walks through exactly how IELTS Writing is scored, what each of the four criteria actually measures, and where candidates most commonly lose the half-bands that stand between them and their target. We will use concrete worked examples throughout, with sample sentences at Band 6, 7, and 8, so you can see for yourself what actually separates the bands. At Shoreline Languages, this is the diagnostic framework we use with every IELTS candidate in their first lesson, and it is the fastest way we know to turn effort into measurable band gains.

The four criteria and the score you actually receive

Your Writing band for each task is the average of four equally weighted criteria: Task Achievement (Task Response for Task 2), Coherence and Cohesion, Lexical Resource, and Grammatical Range and Accuracy. Each is scored on a nine-band scale, and your final Writing band is calculated from the average of Task 1 (weighted one third) and Task 2 (weighted two thirds), rounded to the nearest half-band.

This weighting matters enormously. A Band 7 on Task 1 with a Band 6 on Task 2 gives you an overall 6.5, not 6.5 rounded up to 7. A Band 6 on Task 1 with a Band 7 on Task 2 gives you a 7, because Task 2 counts twice. Most candidates under-prepare for Task 2 relative to its weight, and they pay for it on the day.

Weak performance in any single criterion also drags down the average even if the other three are strong. A script with Band 7.5 vocabulary, Band 7.5 grammar, and Band 7.5 coherence but a Task 2 response that misreads the question and fails to fully address both parts can score Band 6 for Task Response, and the overall Task 2 score collapses to Band 7 instead of the Band 7.5 the language deserves. Understanding this is the first step in targeted preparation: your weakest criterion sets your ceiling.

Task Achievement / Task Response: the criterion most candidates misread

Task Achievement (for Task 1) and Task Response (for Task 2) measure whether you have done what the question actually asked. For Task 1, this means accurately describing the key features of the graph, chart, or process without missing major trends or inventing details the data does not support. For Task 2, this means fully addressing every part of the prompt, developing a clear position, and supporting it with relevant reasons and examples.

To see what "Task Response" means in practice, consider this classic Task 2 prompt:

Some people believe that governments should make university education free for all students. Others believe students should pay for their own education. Discuss both views and give your own opinion.

This prompt has three distinct requirements baked into it: discuss view one, discuss view two, and give your own opinion. A script that discusses both views without making the author's position clear is capped at Band 6 for Task Response. A script that gives only an opinion without discussing both views is capped at Band 5 or Band 6. A Band 7 script must do all three and do them with developed, extended argument.

Here is the same body paragraph discussing the "free university" side, written at three different bands. Notice the structural difference, not just the language.

Band 6 version:

Free university is good because it helps poor students. Students from poor families cannot afford to study, so free education gives them a chance. Also, the country benefits because there are more educated people. Therefore, governments should pay for university.

This is a list of related points. It makes three assertions in three sentences, supports none of them beyond the assertion itself, and closes with a restatement. The connectors ("also," "therefore") are bolted on. Task Response here would likely be 6 because the argument is not extended.

Band 7 version:

One compelling argument for government-funded tertiary education is the barrier that tuition fees create for students from lower-income families. When a university degree costs tens of thousands of dollars in fees alone, even students who meet academic entry requirements may be unable to enrol, and the country loses the contribution these individuals could have made. By removing this financial barrier, governments can unlock a pool of talent that would otherwise be wasted, and the long-term economic return from a better-educated workforce, through higher tax revenues and more productive industries, often exceeds the upfront cost of the tuition subsidy.

This is the same idea, but it is developed. It identifies a specific mechanism (fees as barrier), gives a specific example (tens of thousands in fees), considers consequences (loss of contribution), and closes with a cost-benefit argument. Task Response at 7 or above rewards this kind of extension.

Band 8 version:

The most compelling argument for publicly funded tertiary education is that tuition fees systematically exclude students whose academic ability would otherwise qualify them, and in doing so they reduce the efficiency of the entire educational system. When a degree costs upwards of fifty thousand dollars in Australia, and twice that in the United States, even students admitted to top universities may decline their offers on affordability grounds. The country therefore loses not the marginal student but in many cases exactly the student whose admission the selection process was designed to identify. By contrast, countries that fund tuition from general taxation, Germany and the Nordic states being the clearest examples, retain a significantly higher proportion of their admitted students, and the long-term fiscal return from higher lifetime earnings and the broader productivity gains of a more educated workforce consistently exceeds the annual subsidy, particularly when measured over a working life of forty or more years.

This is Band 8 language but, more importantly, Band 8 argumentation: a claim, a mechanism, specific evidence (dollar figures, country examples), an address of a counter-intuitive implication (it is the best students who are excluded), and a quantitatively qualified conclusion. Examiners are not looking for rhetorical flourish. They are looking for this specific kind of reasoning depth.

The two most common Task Response traps are underdevelopment (the Band 6 paragraph above) and answering the wrong question (discussing university education in general rather than the free-versus-paid debate). Candidates who have memorised essay templates for generic topics frequently fall into both. At Shoreline Languages, the single most common piece of feedback we give our IELTS students in their first Task 2 is: "develop one paragraph properly rather than writing three paragraphs at the surface level." That shift alone is usually worth half a band.

Coherence and Cohesion: the paragraph-level criterion

Coherence and Cohesion measures how clearly your ideas are organised and connected, both within paragraphs and across the whole essay. This is the criterion where many candidates lose marks without realising they are doing anything wrong, because they have been taught that cohesion means inserting connector words.

At Band 6, cohesion is marked by inaccurate or over-use of cohesive devices. A candidate who begins every sentence with "moreover," "furthermore," "in addition," or "on the other hand" is demonstrating that they have memorised a list of connectors but not a natural sense of how ideas flow. Band 7 and above requires cohesion that is built into the structure of the writing, not bolted on.

Consider this Band 6 paragraph:

Firstly, technology has changed education. Moreover, students can learn online. Additionally, teachers can use new tools. Furthermore, students have access to many resources. In conclusion, technology has brought many changes to education.

Every sentence begins with a connector. The connectors do not actually link the ideas; they are labels on sentences that are functionally a list. This paragraph will score Band 6 for Coherence and Cohesion no matter how good the vocabulary is, because the structure is not coherent.

Here is a Band 7 rewrite of the same ideas:

Technology has reshaped education in several overlapping ways. The most visible change is the shift toward online learning, which allows students to access courses from anywhere in the world and attend lectures that would previously have required a move to another city. This flexibility in turn changes the role of classroom teachers, who now curate and deliver material alongside digital resources rather than serving as the sole source of information, and students themselves have access to a far wider range of supplementary materials, from academic journals to specialist online communities, than any previous generation.

The sentences connect through their content, not through inserted connectors. "The most visible change" introduces the topic; "this flexibility in turn changes" builds on it; "students themselves have access" parallels and extends it. A reader follows the line of reasoning because the ideas build on each other, which is what real cohesion looks like.

The most effective way to build real cohesion is to think at the paragraph level rather than the sentence level. Each paragraph should have a clear topic sentence, a developed explanation, an example or specific evidence, and a link to the overall argument. A Band 7 script reads as a coherent line of reasoning; a Band 6 script reads as a list of related but unconnected points with connectors inserted between them.

Paragraphing itself is another common issue. Candidates who present a single long paragraph, or who divide their writing into three-sentence mini-paragraphs, are penalised. A clear introduction (two or three sentences), two or three developed body paragraphs (typically five to eight sentences each), and a conclusion (two or three sentences) is the structure examiners expect. Deviation costs marks, regardless of the content quality.

Lexical Resource: range is more important than complexity

Lexical Resource measures your range of vocabulary and your ability to use it accurately and appropriately. This is the criterion most candidates approach incorrectly, because they believe that Band 7 and above requires obscure, advanced vocabulary.

It does not. What Band 7 and above requires is flexibility and precision: a wide enough range of vocabulary to avoid repetition, collocations that native speakers actually use, and word choices that match the register of the task. Candidates who try to force in unusual words often make collocation errors (using words in combinations that native speakers do not use), register errors (using informal language in a formal task), or meaning errors (using a word whose precise sense is wrong for the context).

Consider how the same idea, "pollution is getting worse," can be expressed across bands:

Band 6: Pollution is getting worse every year, and it is bad for people and the environment.

Band 7: Air pollution levels continue to rise year on year, with serious consequences for public health and for the natural environment.

Band 8: Ambient air pollution is worsening measurably across most major urban centres, and its health impacts, from respiratory illness to premature mortality, now represent one of the most pressing public health challenges in the developed world.

What changes between Band 6 and Band 7 is not the complexity of individual words but the specificity of the collocations: "air pollution levels" instead of "pollution," "year on year" instead of "every year," "serious consequences for public health" instead of "bad for people." Each of these is a high-frequency collocation that native speakers use automatically. None of them are obscure. What makes them Band 7 is precision.

A candidate who tries to inject "advanced" vocabulary without collocational awareness typically produces something like:

Pollution is aggravating annually, and it is detrimental for human beings and the natural environment.

"Aggravating" and "detrimental" are both known words, but "pollution is aggravating" is not a natural English collocation (pollution can worsen or deteriorate; situations can aggravate), and "detrimental for" is less natural than "harmful to" or "damaging to." This script may use more "sophisticated" individual words, but the collocational errors reduce the Lexical Resource score, not raise it.

The highest-leverage investment for Lexical Resource is not expanding your vocabulary upwards; it is building collocational awareness for vocabulary you already know. Knowing that you "make a decision" but "take action," that you "pay attention" but "show interest," that you "curb pollution" but "tackle poverty," is what separates Band 6.5 from 7.5.

At Shoreline Languages, we maintain topic-specific collocation banks across the most common IELTS Task 2 themes: environment, education, technology, health, work, urbanisation, globalisation, crime. In the first few weeks of preparation, our candidates typically build up a hundred or more high-frequency collocations per theme, which is often the single highest-leverage move for Lexical Resource across an eight-week preparation programme.

Grammatical Range and Accuracy: the quiet criterion

Grammatical Range and Accuracy measures two things together: the variety of grammatical structures you can use, and how accurately you use them. To score Band 7 or above, your script must contain a variety of complex structures (not just simple sentences), and the majority of your sentences must be error-free.

The trap here is the mirror image of Lexical Resource. Candidates often use either simple sentences throughout (which caps their score at 6) or try to use complex structures and make errors that drag their score down. The fix is to write shorter, accurate complex sentences rather than long, error-filled ones.

Common high-value structures that reliably lift Grammatical Range include:

Relative clauses that integrate additional information without starting a new sentence:

People who work from home report higher job satisfaction, though they often report lower rates of promotion over time.

Conditional structures that let you argue about consequences and hypotheticals:

If governments were to invest more heavily in public transport, the demand for private vehicles would fall, and urban air quality would improve as a direct result.

Noun phrases with embedded modifiers that compress information efficiently:

A growing body of research published in the last decade suggests that screen time among adolescents correlates with reduced sleep quality.

Participle clauses that avoid repetition and read more naturally than a series of short sentences:

Recognising the scale of the problem, most developed countries have introduced stricter emissions targets, though few have yet met them.

Inversion and cleft structures for emphasis in the top bands:

Not only do these policies reduce emissions, but they also generate significant economic activity in the renewable energy sector.

A Band 7 script typically contains three or four of these structures used accurately, not fifteen of them crammed into every paragraph. Quality and accuracy matter more than quantity. Candidates who reach for structures they cannot control make grammar errors that pull their overall band down, while candidates who use a small number of complex structures with precision score higher with less effort.

The single most common grammar error we see at Shoreline Languages is the missing article on abstract nouns ("Education is important" is correct; "Government should invest in education" is usually wrong because it should be "The government" or "Governments"). The second most common is subject-verb disagreement on long sentences where the subject and verb have drifted apart. Fixing just these two patterns typically lifts Grammatical Range and Accuracy by half a band.

The moves that reliably shift candidates from 6.5 to 7.5

After reading thousands of IELTS scripts, the pattern of what moves candidates from 6.5 to 7.5 is remarkably consistent.

Plan before writing. Five minutes of planning saves fifteen minutes of rewriting and produces a more coherent script. The plan does not need to be elaborate: a thesis statement, three or four main points with their supporting examples, and a conclusion. Candidates who plan consistently outperform candidates of similar English level who do not.

Write two fewer sentences per paragraph, and make each one count. Over-writing produces redundant sentences, unnecessary connectors, and language errors. Under-writing, within reason, produces tighter argumentation and fewer errors.

Learn collocations deliberately. Spend thirty minutes a week building lists of collocations for common IELTS topics. The gap between "pollution" as a word you know and "to curb air pollution" as a collocation you can use is directly worth half a band.

Master four or five complex structures and use them accurately. Relative clauses, conditionals, and participle clauses, used correctly, are sufficient for Band 7 Grammatical Range. You do not need to use every structure in the English language.

Submit timed practice essays for detailed feedback. Writing in untimed conditions, or writing without detailed feedback, is how most candidates study for IELTS and is why so many plateau. One well-analysed essay per week is more valuable than five self-assessed essays.

How Shoreline Languages prepares you for the Writing band you actually need

At Shoreline Languages, IELTS Writing is the single most common skill our candidates come to us to improve, and the one where we see the most dramatic gains. Our approach is built around the four criteria, not around generic writing advice.

Every candidate begins with a full diagnostic essay, scored against the public band descriptors, with a detailed breakdown of where each criterion sits and which specific features are holding the score back. From there, we build a programme that targets the two criteria where you have the most room to improve, typically Task Response and Lexical Resource, because those are where the highest-leverage gains sit for most candidates.

Every week, you write one Task 1 and one Task 2 under timed conditions. Every script comes back with script-level corrections, a predicted band for each criterion, and a specific focus for the following week. Our teachers are trained in the exact descriptors examiners use, and we do not give vague feedback like "try to use more varied vocabulary." We tell you which four collocations to add to your active vocabulary this week, which grammatical structure to practise, and which aspect of Task Response to deepen. We show you Band 7 and Band 8 rewrites of your own paragraphs so you can see the upgrade in your own writing, not in textbook examples.

Candidates who work with us for eight to twelve weeks typically move up a full band, often more. We are not the cheapest IELTS preparation in Sydney, but we are the most precise, and for candidates who need a specific score by a specific date, that precision is what matters. Our candidates routinely hit their target bands on their next attempt, and many exceed them.

If you are preparing for IELTS, here is what to do next

If you are scoring 6 or 6.5 in Writing and need 7, the first step is to diagnose exactly which criteria are holding you back. We offer a free trial lesson that includes a full diagnostic essay, scored against the band descriptors, with a clear plan for how to close the gap. If you book one, you will walk away with a concrete understanding of what stands between you and your target band, regardless of whether you continue with us afterwards.

The candidates who hit their target IELTS band are almost never the ones who study the hardest. They are the ones who study the most precisely, targeting the specific features their scores are losing marks on. If you would like that precision, we would love to help.