IELTS 6.5 to 7: The Small Changes That Close the Gap

Band 6.5 to 7 is the most crowded gap in IELTS. It is where Express Entry draws its line for the top language points, where most professional registers draw theirs, and where a large share of university offers sit. It is also the gap where the usual advice is least useful, because "improve your English" is not a plan and nobody at 6.5 has a general English problem.

Here is what actually decides it. A 6.5 is not one thing. It is at least four different situations that produce the same number on a report form, and they need four different projects. The first job is not studying. It is working out which of the four you are in, because one of them is a single half band away and another one is four half bands away, and they feel exactly the same from the inside.

#Do the arithmetic before you study anything

Your overall band is the average of your four module bands, rounded to the nearest half. Our own scorer does it in one line, and it is the same rule the real test uses.

Work with the total rather than the average, because the total is easier to think about. Add your four module bands together. To be reported as a 7.0 overall, that total has to reach 27.0. A 6.5 overall means your total is somewhere from 25.0 to 26.5. So the distance to 7 is anywhere between one half band and four half bands, depending entirely on where in that range you are sitting.

Every row in that table is a 6.5. The top row is one good week of work away from a 7. The bottom row is a different exam and probably two months. If you have been told, or have told yourself, that "6.5 to 7 is a small jump", check which row you are on before you believe it.

There is a second question that matters more than the first, and it is the one people get wrong most often.

#Do you need 7 overall, or 7 in every module?

These are completely different targets and they lead to opposite study plans.

• 7 overall lets your strong modules carry your weak ones. Listening 8.0 can pay for Writing 6.0. The right plan is to buy the cheapest half bands available, wherever they are.
• 7 in every module means your strongest modules are worth nothing extra. Only your weakest module matters, and every hour spent on anything else is wasted. Australia's Proficient English level works this way, and so do most nursing, medical and teaching registers.
• A per skill profile is a third case. Canada's CLB 9 wants Listening 8.0 with Reading, Writing and Speaking at 7.0. That is not "7 in everything" and planning as if it were will leave you half a band short in Listening on results day.

Find the sentence on the receiving organisation's own page that names the requirement, and read whether it says overall, or each, or lists a number per skill. Then put your current bands into the free band calculator, which converts them into an overall band, a CLB level and immigration points, so you can see which half band is worth something and which is worth nothing.

#Where a half band is cheapest, in actual marks

Two of your four modules are counted. Two are judged. That difference should drive how you spend your time, and almost nobody uses it.

Listening and Reading convert a raw score out of 40 into a band. These are the conversion points our practice scoring uses, taken from calculateListeningBand and calculateReadingBand in our scoring library:

Read the 6.5 and 7.0 rows together and the size of the job becomes concrete:

• Listening 6.5 to 7.0 is four more correct answers.
• Academic Reading 6.5 to 7.0 is three more correct answers.
• General Training Reading 6.5 to 7.0 is two more correct answers.

Three more right answers is a job you can finish. It has an end. You can count your way to it, and you will know on the day you have done it. Half a band in Writing has no such shape until you know which of the four criteria is holding you down, which is exactly why people at 6.5 spend three months on Writing and get nowhere while three cheap marks sit unclaimed in Reading.

An honest caveat about the table. These are the conversion points we use in practice mode and they follow the tables commonly published for IELTS. The exam board does not publish a per paper table and the real boundaries are adjusted slightly for each version of the test. Treat the numbers as the size of the job, which is reliable, rather than as a promise about the exact paper you sit.

#The Task 2 weighting, which decides more Writing bands than grammar does

Writing Task 2 is marked twice as heavily as Task 1. Most candidates know this and almost none of them act on it. Here is what it does, using the same weighting our calculateWritingBand function applies:

• Task 1 band 6.5, Task 2 band 7.0 gives a Writing band of 7.0.
• Task 1 band 7.0, Task 2 band 6.5 gives a Writing band of 6.5.

Same two numbers, swapped between the tasks, and the reported band moves by half a band. If your Writing is stuck at 6.5 and you have forty minutes a week for it, the forty minutes belong to Task 2. And if you routinely run out of time and rush Task 2, that habit alone can be your whole gap.

#Writing: what the descriptors change between 6 and 7

Your Writing band is built from four criteria, each marked separately and then combined. That means one criterion at band 6 with three at band 7 still reports as a 7, and two at 6 does not. Finding the one weak criterion is the entire task. Guessing at it is why people rewrite essays for months without moving.

Here is what each criterion asks for at 6 and at 7, quoted from the official public band descriptors, followed by what actually changes.

#Task Response

Band 6 "addresses all parts of the task although some parts may be more fully covered than others" and "presents relevant main ideas but some may be inadequately developed/unclear". Band 7 "addresses all parts of the task", "presents a clear position throughout the response" and "presents, extends and supports main ideas".

Two words carry the whole difference: throughout and extends.

Throughout means your position cannot drift. A very common 6 pattern is an introduction that says one thing, a second body paragraph that quietly argues the other side without saying it is doing so, and a conclusion that lands somewhere in between. Every sentence of that essay can be correct English and it is still a 6, because the reader cannot say what you think.

Extends means one idea, taken further. Not three ideas mentioned. Take the prompt from our own bank, WA2-05: "Working from home has become much more common in recent years. Do the advantages of working from home outweigh the disadvantages?"

A band 6 body paragraph on that prompt usually looks like this:

Firstly, working from home has many advantages. Employees save time because they do not travel to the office. They also save money on transport. Moreover, they can spend more time with their family, which is good for their mental health. In addition, companies can reduce their office costs.

Four ideas, none of them developed. The examiner sees a list. Now the same paragraph rebuilt as one idea with a reason and a consequence:

The clearest advantage is the return of commuting time. A worker who spends ninety minutes a day travelling recovers more than seven hours a week, which is close to a full working day. What matters is not the hours themselves but where they go: people who stop commuting tend to sleep longer and exercise more often, and both of those show up later as fewer sick days. The saving is therefore not only personal. An employer who never sees the commute on a balance sheet still pays for it in absence.

The second version is shorter in idea count and longer in thinking. That is what "extends and supports" is describing. Note also that it contains no impressive vocabulary at all. Nothing in it would be out of place in a newspaper.

One more Task Response trap, and it is the most expensive single mistake at this level. Take WA2-03 from our bank: "In many countries, young people are finding it increasingly difficult to get jobs. What are the main causes of this problem? What solutions can you suggest?" That is two questions. An essay that explains four causes brilliantly and offers one thin solution in the conclusion has not addressed all parts of the task, and Task Response is capped around band 5 whatever the rest of the writing looks like. Before you write anything, count the questions in the prompt and write that number in the margin.

#Coherence and Cohesion

Band 6 "uses cohesive devices effectively, but cohesion within and/or between sentences may be faulty or mechanical". Band 7 "logically organises information and ideas; there is clear progression throughout" and "presents a clear central topic within each paragraph".

The word to look at is mechanical. Firstly, secondly, moreover, in addition, in conclusion, one per paragraph, bolted to the front of the sentence. That pattern is the single most reliable marker of a band 6 essay, and candidates use it because every template teaches it.

Band 7 cohesion mostly comes from something else: each sentence picks up a word or an idea from the one before it. Look again at the rebuilt paragraph above. "Commuting time" becomes "a worker who spends ninety minutes travelling", which becomes "the hours themselves", which becomes "where they go". There is exactly one linking word in the whole paragraph, and it is "therefore". The paragraph holds together because the ideas are in a chain, not because signposts have been stapled on.

The practical test: delete every connector from one of your paragraphs. If it still reads in order, your cohesion is real and you can put a few connectors back. If it falls apart into unrelated sentences, the connectors were doing work the sentences should have been doing.

#Lexical Resource

Band 6 "attempts to use less common vocabulary but with some inaccuracy". Band 7 "uses less common lexical items with some awareness of style and collocation" and "may produce occasional errors in word choice, spelling and/or word formation".

Read that band 7 line twice. It allows occasional errors. What it does not allow is a text with no ambition in it at all. The safe essay, written entirely in words the candidate is certain about, is a 6 by design.

The other half is collocation, which is where synonym lists do their damage. Swapping "important" for "paramount" is not range if the sentence becomes "a paramount number of people". The word is rarer and the sentence is worse. Collocation is about which words live together: you make a decision, you take a risk, you raise a concern, unemployment is widespread rather than big. Learning ten of those pairings is worth more than learning fifty single words.

#Grammatical Range and Accuracy

Band 6 "uses a mix of simple and complex sentence forms" and "makes some errors in grammar and punctuation". Band 7 "uses a variety of complex structures" and "produces frequent error-free sentences".

Frequent error-free sentences is the line, and it is countable, which makes it the easiest criterion to self assess honestly. Print one of your essays. Mark every sentence that has no error at all in it. If fewer than half your sentences survive, you are at 6 on this criterion no matter how ambitious the structures are.

The pattern that keeps people at 6 is not simple writing. It is complex attempts that break. A candidate writes a three clause sentence with a relative clause and a conditional in it, and drops an article, a plural and a preposition on the way through. The fix is not to stop trying. It is to keep the ambition and shorten the span: one complex idea per sentence, then a full stop. Two clean complex sentences score better than one broken long one, every time.

#Speaking: the two criteria that move

Speaking at 6.5 is usually held there by Fluency and Coherence and by Grammatical Range and Accuracy. The other two rarely move on their own at this level.

Band 6 fluency "is willing to speak at length, though may lose coherence at times due to occasional repetition, self-correction or hesitation". Band 7 "speaks at length without noticeable effort or loss of coherence" and "may demonstrate language-related hesitation at times".

Notice what band 7 still permits: hesitation, as long as it is language related. Pausing to find a word is allowed at 7. What is not allowed is losing the thread. So the fix is not speaking faster. Candidates who try to fix a 6.5 by talking quickly usually make it worse, because speed increases the error rate and the loss of coherence at the same time.

Band 6 grammar "may make frequent mistakes with complex structures". Band 7 "frequently produces error-free sentences, though some grammatical mistakes persist". Same idea as in Writing, and the same fix: shorter complete sentences, held under control, rather than long sentences that collapse in the middle.

On vocabulary, band 7 asks for "some less common and idiomatic vocabulary" and openly expects "some inappropriate choices". The examiner is not waiting to punish a word you use slightly wrong. They are waiting to see whether you have any range beyond the safe set. Silence is not neutral here. It scores lower than a near miss.

#The structure that fixes most Part 1 and Part 3 answers

Answer, reason, then one concrete detail. Three sentences, no more.

Do you enjoy cooking? Not really, no. I work shifts, so by the time I get home the last thing I want is to stand up for another hour. I cook properly maybe once a week, usually on a Sunday, and the rest of the time it is whatever can be reheated.

That answer is short, it is completely unremarkable in content, and it does four things the examiner is listening for: it commits to a position, it gives a reason, it lands on a specific detail rather than a generalisation, and it never loses coherence. Compare it with the memorised alternative, which announces itself in the first two seconds and pulls Lexical Resource down rather than up, because rehearsed language does not answer the question that was asked.

#Reading and Listening: the three or four marks, named

You now know the size of the job in each: three more Academic Reading answers, four more Listening answers. The next question is where those marks are, and it is almost never a comprehension problem. It is a small set of repeating patterns.

Our listening bank carries 498 deliberately planted distractor traps across the standalone exercises, sorted into nine kinds. The trap index that drives our weakness engine labels 920 questions across 117 exercises. Four of those kinds account for most of the marks lost between 6.5 and 7:

• Self-correction. "It's forty pounds. Actually, forty-five." The last value is the answer. Candidates write the first one because they have already moved on.
• Substitution. "We were going to meet Tuesday, but we changed it to Wednesday." Same shape, applied to a day, a room, a name.
• Negation or except. "All the facilities are included except parking." One small word inverts the answer.
• Speaker disagreement, which lives in Section 3. "I think it's thirty." "No, thirty-five." The answer is whatever the pair settle on, not whatever was said loudest.

Every one of those is a listening habit rather than a language skill: do not write until the speaker has finished the sentence. That single change is worth more than any amount of extra vocabulary at this level, and you can practise it deliberately in free listening practice, where every wrong answer comes back with the reason it was wrong.

Reading is the same story with different names. Our trap-rich reading set holds 16 passages and 208 questions, and each explanation names the trap. The eight named types are Paraphrase, Quantifier Misdirection, Time or Detail Flip, MCQ Partial-Truth Distractor, Hedged versus Definitive, Distractor Named Entity, Opinion Attribution, and Negation or Contrast. Notice how many of them turn on one small word rather than on the meaning of the passage. A statement containing only, all, never or first is usually about that word. A statement that introduces a detail the passage had no reason to mention, such as someone's motive or funding, is usually a Not Given.

If you are missing exactly three marks in Reading, do not read more passages. Redo the questions you got wrong with no clock on, and see whether they come right. If they do, you have a timing problem and you should be practising the order you attempt question types in. If they stay wrong, you have a trap problem and you should be reading explanations, not passages.

#What a four week plan looks like when it follows the arithmetic

This assumes the second row of the first table: a total of 26.0, so two half bands to find, and the receiving body wants 7 overall rather than 7 in each.

1. Days 1 and 2: measure, do not study. One timed Reading paper on the correct paper type, one timed Listening test, one Task 2 under 40 minutes, one recorded Speaking Part 2 and Part 3. Score the first two on the table above. Get the other two marked against the criteria rather than guessed at.
2. Day 3: choose two targets and write them down. Pick the two cheapest half bands, not the two that feel worst. Cheapest usually means a counted module that is within four marks, plus the single weakest Writing criterion.
3. Week 1: the counted module. Short sessions, every day, with a review that names the trap on every miss. You are hunting three or four specific marks, so do not do whole tests every day. Do one section, then spend as long again on why each miss happened.
4. Weeks 2 and 3: the judged module, one criterion at a time. If it is Task Response, write introductions and one body paragraph only, ten of them, and never a full essay. If it is Grammatical Range, take essays you have already written and rebuild the broken sentences. Whole essays are a test, not practice. They are the slowest way to fix a specific criterion.
5. Week 4: put it back together. Two full timed papers. If the numbers hold, stop. If they do not, you have found out four weeks early rather than on results day.

The part people skip is the first two days. Measuring feels like a waste when you are behind. It is the only step that tells you whether you are one half band away or four, and everything after it is guesswork without it.

#Five things that do not close this gap

1. More mock tests without review. A test measures. It does not teach. If you spend less time reviewing a test than sitting it, you are collecting scores rather than improving.
2. Longer words. Band 7 Lexical Resource asks for awareness of style and collocation, not rarity. A rare word in the wrong collocation loses more than a common word in the right one.
3. Memorised openings. "In this modern era of globalisation" and its family are recognised instantly, add nothing to any criterion, and cost you the first thirty seconds of the reader's goodwill.
4. Longer essays. There is no upper word limit, but there is no reward for length either, and long essays at this level usually mean more ideas listed and fewer developed, which is a Task Response loss.
5. Working on your strongest module. It is the most enjoyable practice and, if your target is 7 in every module, it is worth exactly nothing.

#How to know whether it is working

Two of your modules give you a number, so track the raw score out of 40, not the band. The band hides progress: going from 27 to 29 correct in Academic Reading is real, measurable movement and the band does not change at all. People give up during exactly that stretch.

The two judged modules need marking against the criteria, and marking your own writing is the hardest thing in preparation, because the errors you can see are the ones you have already fixed. Paste one Task 2 answer into the free IELTS writing checker and it returns a band for all four criteria with the sentences that hold each one down, which is the per criterion picture the whole plan above depends on. If you do not yet know which module is costing you the most, start with the free 3 minute diagnostic instead: one writing or speaking answer, scored against the official descriptors, is enough to tell you where to point the next four weeks.

We publish how closely our scorer agrees with our own labelled calibration set, every week, including the weeks the number is bad. You can read the current figure and the method on our quality page, and we would rather you judged the tool on that than on a promise.

The gap is real, but it is not vague. Add your four bands. Find the total. Count the marks. Name the criterion. Then spend four weeks on the two cheapest half bands and nothing else.

Start with the free 3 minute diagnostic and find out which half band is closest.