Someone does forty practice tests and stays at 6.5. The problem is rarely unfamiliarity with the format. This separates the two goals learners conflate — learning the test and learning the language — and sets out the reading and writing habits that actually have research behind them, including one uncomfortable finding about preparation courses.
There is a very familiar situation. Someone has done forty practice tests. They know the question types, they know the NOT GIVEN trap, they have an outline ready for every Task 2 prompt. And their score is still 6.5, exactly as it was six months ago.
The usual response is more practice tests. But if forty were not enough, the forty-first will not be either, and there is a specific reason why.
Preparing for IELTS is really two different jobs, and learners almost always merge them.
The first is familiarisation. Knowing how many sections there are, what the questions look like, how many words to write, what the criteria are. This is real and necessary work — but it saturates quickly. After roughly five to ten practice tests you know nearly everything worth knowing about the format. From then on, more practice tests teach you almost nothing further about the test.
The second is raising your English. That moves slowly, in months rather than weeks, and it is what determines your band once familiarisation has saturated.
The plateau everyone hits is precisely the moment the first job runs out and the second has not yet caught up.
And here the research is uncomfortable. Anthony Green (2007) compared learners on dedicated IELTS preparation courses with learners on general academic English courses. On gains in writing, the test-preparation group did not come out ahead. Teaching to the test did not deliver the advantage everyone assumes it does.
Elder and O'Loughlin (2003) added the numbers: measuring band gain across periods of intensive English study, they found progress is real but modest — on average around half a band after several months of full-time study, with large variation between individuals.
Stated plainly: if you need a whole band, plan in months, not weeks. Anyone promising otherwise is selling something.
Look at the actual numbers. IELTS Academic Reading gives you 60 minutes, three passages, 40 questions, totalling 2,150–2,750 words. And unlike the Listening test, no extra transfer time is given — those 60 minutes include writing your answers onto the answer sheet.
Almost everyone diagnoses this as a time-management problem. Usually it is not.
Take an average test of about 2,450 words and apply the coverage threshold. Hu and Nation (2000) showed that to read comfortably without help you need to know around 98% of the running words.
At 98% coverage, you meet roughly 49 unknown words in that test. Annoying, but survivable.
At 95% — still a decent level of English — you meet roughly 122 unknown words. Every pause to infer one costs a few seconds, and they compound. You are not slow; you are paying a vocabulary tax on every line.
This is why strategy tips only help so far. You cannot skim what you cannot decode. Skimming assumes that as your eye passes over the words, their meanings fire automatically. If they do not fire, the technique collapses.
So the work has two parts, and most learners do only the second.
Build academic vocabulary. This is the long game, and it is the textbook case for spaced repetition (Cepeda and colleagues, 2006). Prioritise general academic vocabulary over rare subject-specific terms.
Build recognition speed. This is the neglected half. Grabe (2009) stresses that reading fluency in a second language has to be trained separately: knowing a word and recognising it in a quarter of a second are different achievements. You train it with timed reading and repeated reading of the same passage — not with more practice tests.
Underneath both sits extensive reading. Nakanishi's meta-analysis (2015) found small-to-moderate gains in comprehension and vocabulary. Twenty minutes a day of something you actually enjoy does what one weekly practice test cannot: it gives you volume.
The Writing test gives you 60 minutes: Task 1 of at least 150 words in about 20 minutes, Task 2 of at least 250 words in about 40. Task 2 counts twice as much as Task 1.
Scripts are assessed on four criteria, each carrying equal weight: Task Achievement/Response, Coherence and Cohesion, Lexical Resource, and Grammatical Range and Accuracy.
Sit with that for a moment, because it changes what practice should look like. Three of the four criteria are not about whether you had good ideas. They are about organisation, vocabulary, and grammatical accuracy.
Yet when learners practise writing, they overwhelmingly practise one thing: generating content and getting to the word count. Three-quarters of the marks come from skills that are rarely isolated and drilled.
This is where Ericsson's concept of deliberate practice (1993) earns its place. Repetition alone does not produce improvement; what produces improvement is work on a defined sub-skill with feedback. Writing thirty essays without feedback largely rehearses the errors you already make.
This is one of the longest arguments in writing research, and it has a usable conclusion.
Truscott (1996) argued bluntly that grammar correction in writing classes is ineffective, possibly harmful, and should be abandoned. Ferris (1999) pushed back. The debate ran for decades.
Where it settled is reasonably clear: focused feedback works considerably better than comprehensive feedback. Correcting every error in a script tends not to produce lasting improvement — the essay comes back covered in red, the learner skims it, and nothing sticks. But when feedback targets one or two specific error types, accuracy on exactly those points improves and holds. Bitchener and Knoch (2010) found effects of this kind even with advanced writers.
That converts into a very concrete habit. Do not ask anyone to correct everything. Pick two error types — articles and subject–verb agreement, say — and track only those for three weeks. When they are solid, swap in two more.
For speakers of languages without inflection, the candidate list is predictable: articles (a/the/zero), plural -s, tense and -ed, and third-person singular -s. These are all short, unstressed morphemes that get swallowed in speech — which means your ear will never catch them for you. They have to be handled consciously.
The most popular preparation strategy is also the riskiest: memorising model paragraphs and fitting them to every prompt.
It fails for two reasons, both sitting inside the marking criteria.
Task Response drops. A memorised introduction was written for a generic prompt; the real prompt is specific. The more you force the template, the further the essay drifts from the actual question. Examiners are trained to identify memorised content, and language you did not produce does not count towards your ability.
Lexical Resource drops too. Overused template phrases — the In this ever-changing world... family — are not less common lexical items. They are the most common items in existence.
There is a correct version of the idea, though, and it works: learn small, flexible chunks rather than whole paragraphs. Ways of marking contrast, of conceding a point, of introducing an example, of quantifying a claim. These attach to any content. They are tools rather than scripts.
One more cheap, evidence-backed habit: spend five minutes planning before writing Task 2. Ellis and Yuan (2004) showed that pre-task planning measurably improves fluency and complexity in second-language writing. Five minutes is not time lost; it is time bought back later.
This is where most study plans fail. Not because the content was wrong, but because they did not survive two weeks.
First, drop the 21-day figure. It is folklore. Lally and colleagues (2010) tracked real people forming real habits and found a median of 66 days, with a range from 18 to 254 depending on the behaviour and the person. If study still feels effortful in week three, you have not failed — you are exactly where the data predicts.
Second, use implementation intentions. This is the most strongly evidenced technique in the whole of habit science, and it is essentially free. Gollwitzer (1999) showed that plans of the form if situation X, then I will do Y vastly outperform general intentions. The meta-analysis by Gollwitzer and Sheeran (2006) across hundreds of studies found medium-to-large effects.
In practice: I will study harder is close to useless. After I pour my morning coffee, I will read one passage at the kitchen table works, because it has pre-decided the time, the place and the trigger.
Third, keep the context stable. Wood and Neal (2007) describe habits as context–response associations: behaviour attaches to recurring environmental cues. Studying in the same place at the same time costs far less willpower than deciding afresh each day.
And fourth, set a minimum small enough to survive a bad day. A three-hour daily plan dies in the first week. Forty-five minutes does not.
They are not useless — but the reason they help is not the reason people think.
Sitting a practice test is an act of retrieval, and Roediger and Karpicke (2006) showed that retrieval itself strengthens memory far more than rereading. So practice tests do teach — just not in proportion to the hours they consume.
What decides their value is what happens afterwards. A test that is marked and filed away is a measurement. A test that is dismantled — every unknown word into the card system, every wrong answer traced to a cause (did I not know the word? not understand the question? run out of time?) — is a study session.
This is where the habit of grinding test after test goes wrong. It feels like progress because it is measurable and genuinely tiring. But if the correction loop is skipped, you are measuring the same gap repeatedly with increasing precision. One test analysed properly is worth more than five rushed through.
Concrete, and defensible when you are busy:
Daily (20 minutes): extensive reading of something you actually enjoy, at a level you mostly understand. This is the foundation, and it is the first thing cut. Do not cut it.
Daily (10 minutes): spaced vocabulary cards. Short and non-negotiable.
Three times a week (20 minutes): timed reading for recognition speed — reread the same passage twice, faster the second time.
Twice a week (45 minutes): one full Task 2 under timed conditions, with five minutes of planning. Then request feedback only on the two error types you are tracking.
Once a week (30 minutes): rewrite that same essay after the feedback. This step is almost always skipped, and it has the highest learning rate of anything in the week.
Every two weeks: one full Reading test, strictly 60 minutes, under exam conditions. Then spend longer analysing it than you spent doing it.
The great temptation in exam preparation is to aim at the number instead of at the thing that produces the number.
But the band is a measurement of your reading and writing at one moment. You can make that measurement more accurate — by understanding the test, by not losing marks transferring answers, by not writing off-topic. That is the familiarisation work, and it is worth doing, but it has a ceiling.
The rest has no shortcut. Vocabulary grows through meeting words repeatedly over months. Accuracy improves when someone points at one specific error often enough that you start catching it yourself. Reading speed rises when word recognition becomes automatic.
None of that happens the night before the test. But all of it happens, fairly reliably, if the habit survives long enough — and the data says the number to aim at is about two months, not three weeks.
The same article, told in plain words — for younger readers, or for anyone who wants the point quickly.
There is a very familiar situation. Someone has done forty practice tests. They know the question types, they know the NOT GIVEN trap, they have an outline ready for every Task 2 prompt. And their score is still 6.5, exactly as it was six months ago.
The usual response is more practice tests. But if forty were not enough, the forty-first will not be either, and there is a specific reason why.
Preparing for IELTS is really two different jobs, and learners almost always merge them.
The first is familiarisation. Knowing how many sections there are, what the questions look like, how many words to write, what the criteria are. This is real and necessary work — but it saturates quickly. After roughly five to ten practice tests you know nearly everything worth knowing about the format. From then on, more practice tests teach you almost nothing further about the test.
The second is raising your English. That moves slowly, in months rather than weeks, and it is what determines your band once familiarisation has saturated.
The plateau everyone hits is precisely the moment the first job runs out and the second has not yet caught up.
And here the research is uncomfortable. Anthony Green (2007) compared learners on dedicated IELTS preparation courses with learners on general academic English courses. On gains in writing, the test-preparation group did not come out ahead. Teaching to the test did not deliver the advantage everyone assumes it does.
Elder and O'Loughlin (2003) added the numbers: measuring band gain across periods of intensive English study, they found progress is real but modest — on average around half a band after several months of full-time study, with large variation between individuals.
Stated plainly: if you need a whole band, plan in months, not weeks. Anyone promising otherwise is selling something.
Look at the actual numbers. Academic Reading gives you 60 minutes, three passages, 40 questions, totalling 2,150–2,750 words. And unlike Listening, no extra transfer time is given — those 60 minutes include writing your answers onto the answer sheet.
Almost everyone diagnoses this as a time-management problem. Usually it is not.
Take an average test of about 2,450 words and apply the coverage threshold. Hu and Nation (2000) showed that to read comfortably without help you need to know around 98% of the running words.
At 98% coverage, you meet roughly 49 unknown words in that test. Annoying, but survivable.
At 95% — still a decent level of English — you meet roughly 122 unknown words. Every pause to infer one costs a few seconds, and they compound. You are not slow; you are paying a vocabulary tax on every line.
This is why strategy tips only help so far. You cannot skim what you cannot decode. Skimming assumes that as your eye passes over the words, their meanings fire automatically. If they do not fire, the technique collapses.
So the work has two parts, and most learners do only the second.
Build academic vocabulary. This is the long game, and it is the textbook case for spaced repetition (Cepeda and colleagues, 2006). Prioritise general academic vocabulary over rare subject-specific terms.
Build recognition speed. This is the neglected half. Grabe (2009) stresses that reading fluency in a second language has to be trained separately: knowing a word and recognising it in a quarter of a second are different achievements. You train it with timed reading and repeated reading of the same passage — not with more practice tests.
Underneath both sits extensive reading. Nakanishi's meta-analysis (2015) found small-to-moderate gains in comprehension and vocabulary. Twenty minutes a day of something you actually enjoy does what one weekly practice test cannot: it gives you volume.
The Writing test gives you 60 minutes: Task 1 of at least 150 words in about 20 minutes, Task 2 of at least 250 words in about 40. Task 2 counts twice as much as Task 1.
Scripts are assessed on four criteria, each carrying equal weight: Task Achievement/Response, Coherence and Cohesion, Lexical Resource, and Grammatical Range and Accuracy.
Sit with that for a moment, because it changes what practice should look like. Three of the four criteria are not about whether you had good ideas. They are about organisation, vocabulary, and grammatical accuracy.
Yet when learners practise writing, they overwhelmingly practise one thing: generating content and getting to the word count. Three-quarters of the marks come from skills that are rarely isolated and drilled.
This is where Ericsson's concept of deliberate practice (1993) earns its place. Repetition alone does not produce improvement; what produces improvement is work on a defined sub-skill with feedback. Writing thirty essays without feedback largely rehearses the errors you already make.
This is one of the longest arguments in writing research, and it has a usable conclusion.
Truscott (1996) argued bluntly that grammar correction in writing classes is ineffective, possibly harmful, and should be abandoned. Ferris (1999) pushed back. The debate ran for decades.
Where it settled is reasonably clear: focused feedback works considerably better than comprehensive feedback. Correcting every error in a script tends not to produce lasting improvement — the essay comes back covered in red, the learner skims it, and nothing sticks. But when feedback targets one or two specific error types, accuracy on exactly those points improves and holds. Bitchener and Knoch (2010) found effects of this kind even with advanced writers.
That converts into a very concrete habit. Do not ask anyone to correct everything. Pick two error types — articles and subject–verb agreement, say — and track only those for three weeks. When they are solid, swap in two more.
For speakers of languages without inflection, the candidate list is predictable: articles (a/the/zero), plural -s, tense and -ed, and third-person singular -s. These are all short, unstressed morphemes that get swallowed in speech — which means your ear will never catch them for you. They have to be handled consciously.
The most popular preparation strategy is also the riskiest: memorising model paragraphs and fitting them to every prompt. It fails for two reasons, both sitting inside the marking criteria.
Task Response drops. A memorised introduction was written for a generic prompt; the real prompt is specific. The more you force the template, the further the essay drifts from the actual question. Examiners are trained to identify memorised content, and language you did not produce does not count towards your ability.
Lexical Resource drops too. Overused template phrases — the In this ever-changing world… family — are not less common lexical items. They are the most common items in existence.
There is a correct version of the idea, though, and it works: learn small, flexible chunks rather than whole paragraphs. Ways of marking contrast, of conceding a point, of introducing an example, of quantifying a claim. These attach to any content. They are tools rather than scripts.
One more cheap, evidence-backed habit: spend five minutes planning before writing Task 2. Ellis and Yuan (2004) showed that pre-task planning measurably improves fluency and complexity in second-language writing. Five minutes is not time lost; it is time bought back later.
This is where most study plans fail. Not because the content was wrong, but because they did not survive two weeks.
First, drop the 21-day figure. It is folklore. Lally and colleagues (2010) tracked real people forming real habits and found a median of 66 days, with a range from 18 to 254 depending on the behaviour and the person. If study still feels effortful in week three, you have not failed — you are exactly where the data predicts.
Second, use implementation intentions. This is the most strongly evidenced technique in the whole of habit science, and it is essentially free. Gollwitzer (1999) showed that plans of the form if situation X, then I will do Y vastly outperform general intentions. The meta-analysis by Gollwitzer and Sheeran (2006) across hundreds of studies found medium-to-large effects.
In practice: I will study harder is close to useless. After I pour my morning coffee, I will read one passage at the kitchen table works, because it has pre-decided the time, the place and the trigger.
Third, keep the context stable. Wood and Neal (2007) describe habits as context–response associations: behaviour attaches to recurring environmental cues. Studying in the same place at the same time costs far less willpower than deciding afresh each day.
And fourth, set a minimum small enough to survive a bad day. A three-hour daily plan dies in the first week. Forty-five minutes does not.
They are not useless — but the reason they help is not the reason people think.
Sitting a practice test is an act of retrieval, and Roediger and Karpicke (2006) showed that retrieval itself strengthens memory far more than rereading. So practice tests do teach — just not in proportion to the hours they consume.
What decides their value is what happens afterwards. A test that is marked and filed away is a measurement. A test that is dismantled — every unknown word into the card system, every wrong answer traced to a cause (did I not know the word? not understand the question? run out of time?) — is a study session.
This is where the habit of grinding test after test goes wrong. It feels like progress because it is measurable and genuinely tiring. But if the correction loop is skipped, you are measuring the same gap repeatedly with increasing precision. One test analysed properly is worth more than five rushed through.
Daily (20 minutes): extensive reading of something you actually enjoy, at a level you mostly understand. This is the foundation, and it is the first thing cut. Do not cut it.
Daily (10 minutes): spaced vocabulary cards. Short and non-negotiable.
Three times a week (20 minutes): timed reading for recognition speed — reread the same passage twice, faster the second time.
Twice a week (45 minutes): one full Task 2 under timed conditions, with five minutes of planning. Then request feedback only on the two error types you are tracking.
Once a week (30 minutes): rewrite that same essay after the feedback. This step is almost always skipped, and it has the highest learning rate of anything in the week.
Every two weeks: one full Reading test, strictly 60 minutes, under exam conditions. Then spend longer analysing it than you spent doing it.
The great temptation in exam preparation is to aim at the number instead of at the thing that produces the number.
But the band is a measurement of your reading and writing at one moment. You can make that measurement more accurate — by understanding the test, by not losing marks transferring answers, by not writing off-topic. That is the familiarisation work, and it is worth doing, but it has a ceiling.
The rest has no shortcut. Vocabulary grows through meeting words repeatedly over months. Accuracy improves when someone points at one specific error often enough that you start catching it yourself. Reading speed rises when word recognition becomes automatic.
None of that happens the night before the test. But all of it happens, fairly reliably, if the habit survives long enough — and the data says the number to aim at is about two months, not three weeks.
These articles summarize well-established research in learning science and linguistics. Key sources and further reading:
The recording plays exactly once. That single fact governs the whole paper, and every technique worth learning is a cons...
The three parts are not three difficulty levels of the same task. They test different things — and the answer length tha...
Task 2 counts twice as much as Task 1. So the most consequential decision you make is not a good phrase — it is stopping...