How Is IELTS Writing Actually Scored? The Four Criteria Explained

How is IELTS Writing scored, and what is the examiner actually marking?

Most candidates picture the Writing test as a single judgement. You write, an examiner reads, a number appears. That is not how it works, and the difference matters, because the real process is far more mechanical than the mystery people build around it.

An examiner marks your answer against four named criteria. For Task 1 they are Task Achievement, Coherence and Cohesion, Lexical Resource, and Grammatical Range and Accuracy. For Task 2 the first criterion is called Task Response instead, and the other three keep the same names. These are not internal jargon that leaked out. They are published, they are what examiners are trained against, and the full band descriptors for each are available to download from the official IELTS site.

Each criterion is scored on the same nine band scale as the rest of the test. Within a single task the four are weighted equally, so the score for that task is essentially their average. That equal weighting has an unpleasant consequence people rarely think through. One weak criterion drags the whole task down, and no amount of strength elsewhere fully compensates.

Picture a Task 2 answer with excellent ideas, a clear position, tight paragraphs and a good range of vocabulary, written by someone whose article use is inconsistent and whose complex sentences frequently break. Three criteria at 7. One at 5. The task lands near 6.5, and the candidate walks out believing the essay was strong, because in three of the four ways that count, it was.

How do two tasks turn into one Writing band?

The Writing test runs for 60 minutes in total, and nobody stops you at the twenty minute mark to move you on. Inside that hour you produce two separate pieces of writing, and each is assessed independently against the four criteria. Then the two task scores are combined, and this is where the arithmetic stops being intuitive.

Task 2 contributes twice as much as Task 1 to the Writing score, so an hour split evenly is an hour spent badly. Written as a rough formula, your Writing score behaves like the first task counting once and the second counting twice, then the total divided by three. A candidate scoring 7 on Task 1 and 6 on Task 2 does not land on 6.5. They land nearer 6.3, which reports as 6.5 only because Writing is given in whole and half bands, and can just as easily report as 6.0 depending on where the underlying scores actually sat.

Reverse the situation and the picture improves. Task 1 at 6, Task 2 at 7, and the same weighting pulls the result up toward 6.7. Same two numbers. Different order. A meaningfully different outcome.

This single fact should reorganise how you spend the hour, and for most candidates it does not, because the tasks appear on the paper in order and people treat that order as a schedule.

Read the fourth row again. A candidate who half abandons Task 1 and writes a genuinely strong Task 2 can still reach 6.5. The reverse combination, a polished Task 1 and a rushed Task 2, lands lower. That is not a loophole to exploit, and running out of time on Task 1 deliberately is a poor plan. It is a demonstration of where the marks live.

Task Achievement and Task Response: are you answering the question asked?

This criterion asks one blunt question. Did you do what the task told you to do, completely, relevantly and appropriately?

In Task 1 of the Academic test, that means describing the visual information accurately. Select the significant features rather than listing every number. Make comparisons where the data invites them. Report an overview, which is the sentence that says what the chart shows overall, and which is the single most commonly missing element in Task 1 answers. A response with no overview is capped, no matter how neatly the details are transcribed.

In General Training, Task 1 is a letter, and the equivalent failure is tone. A letter to a friend written in the register of a complaint to a bank has not met the task, even if every bullet point is covered.

Task 2 is where the criterion earns its different name. Task Response is not simply whether you wrote about the topic. It asks whether you addressed all parts of the prompt, presented a clear position throughout, and developed your ideas rather than announcing them.

Three specific ways candidates lose marks here, all of them avoidable:

• Partial coverage. The prompt asks for advantages and disadvantages, and the answer spends six hundred words on advantages. Half the task, half the credit.
• Drifting position. The introduction agrees, body paragraph two hedges, the conclusion sits on the fence. An examiner cannot credit a clear position that is not there.
• Undeveloped ideas. A paragraph that states a reason, then restates it in different words, then moves on. Development means explaining a mechanism or giving an example, not repeating yourself more elegantly.

The fix for all three is the same, and it costs about three minutes. Before writing, underline every instruction word in the prompt and write your position in one sentence. Then check that sentence against your conclusion when you finish. Our IELTS guide walks through what each Task 2 question type is asking for.

Coherence and Cohesion: the criterion almost everyone misreads

Ask a room of candidates how to improve Coherence and Cohesion and most will say linking words. That belief is why so many essays open every paragraph with a signpost and still score 6.

Coherence is whether your ideas are in an order that makes sense to a reader. Cohesion is whether the connections between them are signalled clearly. Linking words serve the second, and only when they are the right ones, used sparingly, in sentences that were already logically arranged.

Overusing them actively costs marks. The descriptors call out mechanical overuse specifically. An essay where every sentence begins with a connector reads like a form being filled in, and it tells an examiner the writer is decorating rather than organising.

What actually moves this criterion:

• One clear idea per paragraph, stated near the top, so a reader knows where they are.
• Referencing that does real work. Pronouns and phrases like this shift or such policies bind sentences together far more naturally than another however.
• Paragraphing that reflects the argument rather than a memorised shape. If two paragraphs make the same point, they are one paragraph.
• Sentence order inside the paragraph. Reason, then explanation, then example, then consequence. Shuffle those and no connector will save it.

Here is a test you can run on your own writing in thirty seconds. Delete every linking word from a paragraph. If the paragraph still reads logically, your coherence was real and the connectors were supporting it. If it collapses into a pile of unrelated sentences, the connectors were doing all the work, and an examiner has already noticed.

Lexical Resource: range, precision and accuracy together

Vocabulary is where preparation advice does the most damage. Candidates memorise lists of impressive words, deploy them, and score lower than they did before.

The criterion measures three things at once. Range, meaning how wide your vocabulary is. Precision, meaning whether you chose the word that fits the exact meaning. Accuracy, meaning spelling and word form.

A memorised word satisfies the first and fails the other two. Writing that a trend plummeted marginally shows range and shows, unmistakably, that the writer does not know what plummet means. An examiner reads that as a vocabulary error, not as ambition, and the descriptors are explicit that inappropriate word choice pulls the criterion down.

Precision beats impressiveness every time. Consider the difference between a problem that is serious, severe, acute, chronic or widespread. Those are not five ways of saying the same thing. Each carries a different meaning, and picking the right one demonstrates more control than any thesaurus substitution.

Collocation is the other half of it. Words in English travel in fixed company. You conduct research, you do not make research. You take measures, you do not do measures. Collocation errors are invisible to the writer and glaring to a reader, which makes them almost impossible to catch without feedback.

Grammatical Range and Accuracy: the usual ceiling

For most candidates preparing in Pakistan, India, Bangladesh, Nepal and Sri Lanka, this is the criterion that decides whether the score starts with a 6 or a 7.

The name contains both requirements, and they pull against each other. Range means using a variety of structures, including complex sentences, conditionals, relative clauses and passive constructions where they fit. Accuracy means those structures being correct, and it is measured across the whole answer.

Play safe with nothing but simple sentences and you are accurate but narrow, which caps you. Reach for complexity you cannot control and you have range with a scatter of errors, which caps you too. Band 7 wants frequent error free sentences alongside a genuine variety of structures, and that combination is harder than either half alone.

The errors that repeat most often in scripts from South Asia are a short list, and they are learnable:

• Articles. A, an, the and no article at all. English is unusually fussy here and most South Asian first languages give no help.
• Subject verb agreement across a long subject, where the verb ends up agreeing with the nearest noun instead of the real one.
• Tense consistency inside a paragraph, particularly drifting into present tense when describing past data.
• Prepositions after common verbs and adjectives, which are memorised rather than derived.
• Comma splices, where two full sentences are joined by a comma alone.
• Countable and uncountable nouns. Informations, advices, equipments and researches are all common and all wrong.

Notice how narrow that list is. Six patterns, not six hundred. A candidate who fixes articles and comma splices alone often gains half a band, because those two account for a disproportionate share of the errors an examiner counts. The problem is not that they are hard. The problem is that you cannot see them in your own writing, which is what feedback is for.

Why word count and timing are scoring decisions

Task 1 asks for at least 150 words, and the number is a floor rather than a target. Task 2 asks for at least 250 words, which is roughly four solid paragraphs at the pace most candidates write. These minimums are not administrative trivia. They connect directly to Task Achievement and Task Response, because an underlength answer has by definition not covered the task fully.

IELTS itself suggests spending no more than 20 minutes on Task 1 so that Task 2 keeps the other forty. Combined with the weighting, the reasoning is simple. The task worth twice as much gets twice the time.

What goes wrong in the room is predictable. Task 1 appears first, so candidates start there, get absorbed in describing every data point, and look up at the clock thirty five minutes later. The remaining twenty five minutes go to the task that carries two thirds of the score. That single misallocation costs more candidates half a band than any grammar rule.

Writing far beyond the minimum is its own trap. Length is not a criterion. A 400 word Task 2 usually contains repetition, drift away from the position, and simply more sentences in which a grammar error can occur. Somewhere near 270 to 300 words for Task 2, and 170 to 190 for Task 1, gives you room to develop ideas without handing over extra opportunities to lose marks.

A workable split of the hour looks like this. Two minutes reading Task 1 and deciding what the overview sentence says. Eighteen minutes writing it. Five minutes planning Task 2, which feels wasteful and is not. Thirty minutes writing. Five minutes checking, and check for your own known error patterns rather than reading for general quality, because reading for general quality finds nothing.

What separates a band 6 answer from a band 7?

Put a 6.0 and a 7.0 Task 2 side by side and the difference is rarely what candidates expect. The 7.0 is not longer. It usually does not contain rarer vocabulary. It often looks plainer.

What it does differently:

• Position stated in the introduction and still standing in the conclusion, unchanged.
• Every body paragraph developing one idea properly instead of two ideas partially.
• Vocabulary that is precise rather than decorative, with collocations that sound natural.
• Frequent sentences with no grammatical error at all, mixed with complex structures that are under control.
• Cohesion that comes from the order of ideas, with connectors used only where a reader genuinely needs one.

The band 6 answer, by contrast, typically shows a relevant position that wobbles, paragraphs where the second idea arrives before the first is finished, some ambitious vocabulary used slightly wrongly, and errors that recur often enough to be noticed even though meaning is clear throughout.

Read those two descriptions again and notice something. Nothing in the gap requires more English. It requires more control of the English you already have, which is a much shorter journey than most candidates assume they are on.

How do you mark your own writing honestly?

You cannot, entirely, and that is the central difficulty of preparing for this section alone. You can check a Reading answer against a key. There is no key for an essay, and the errors you cannot see are precisely the ones costing you marks.

What you can do is get closer than most people bother to:

• Download the public band descriptors and read the rows for bands 6, 7 and 8 on one criterion. Not all four at once. One.
• Take an essay you wrote last week, not one from today, because distance makes your own errors visible.
• Mark only for that criterion, and write down which band row your answer actually matches rather than which one you hope it matches.
• Keep an error log. Every correction you receive goes in it, grouped by type. After ten essays the list of patterns will be short, repetitive and extremely useful.

The gap that remains is the one that needs another reader. That is where an AI examiner earns its place, and where its limits should be stated plainly. It marks against the same public descriptors, it quotes your sentences back with corrections, and it produces the same score for the same answer every time, which means a rising trend across ten essays is real improvement rather than a mood. It is an estimate. Your official band comes from the real test and a trained human examiner, and nobody, including us, can promise you a particular result.

If you would rather talk through your target and timeline with a person, an advisor will answer on our contact page. If you are still deciding between tests, the two minute quiz asks about your destination and route before recommending one.

A four week plan that targets one criterion

Preparation fails when it is unfocused. Here is a plan built on the arithmetic above, which means it spends most of its attention on Task 2 and on whichever criterion is your weakest.

Week one, diagnosis. Write three Task 2 essays and one Task 1, all under exam timing, with a clock visible. Get all four marked against the criteria. Do not write a fifth until you know which criterion scored lowest, because writing more of the same thing is how candidates spend three months without moving.

Week two, one criterion. Take the lowest scoring criterion and work only on it. If it is grammar, drill your specific error patterns from the log rather than grammar in general. If it is Task Response, practise writing position statements and paragraph plans without writing full essays at all. Ten plans in an hour teaches more about Task Response than one essay does.

Week three, integration. Back to full timed essays, two or three of them, with attention on the criterion you drilled. Expect the others to dip slightly. That is normal and it corrects itself.

Week four, exam conditions. Both tasks in one sitting, sixty minutes, no pausing, no looking things up. Twice in the week. This is the part almost everyone skips, and it is the part that reveals whether your timing plan survives contact with the actual test.

What happens after you have the band?

A Writing band is not the goal. It is the gate in front of the goal, and once you are through it the questions change completely: which universities will take this score, what the visa route needs on top of it, how much money has to sit in a bank account and for how long, and what to do if an application comes back refused.

That is the part BandVisa does not do, and the reason this site is free. It is built by Borders and Bridges, a Lahore consultancy that has worked with students since 2016 and is registered with SECP and FBR, accredited by ICEF, and a British Council verified agent. The practice tools cost nothing because the business earns further down the road, on the admission and the visa.

Two things are worth saying plainly about that stage. A refusal is not automatically the end of a route, and people give up on perfectly recoverable applications because a refusal letter reads like a verdict rather than a list of doubts. And nobody, this firm included, can promise you a visa or a band. What an experienced adviser can do is read the refusal notice properly, work out what the officer actually doubted, and tell you honestly whether reapplying, appealing or changing route is the stronger move. If you want that conversation, it is free and it starts on our contact page.