Why did two teachers give your IELTS essay different band scores?
Two teachers, one essay, two bands. Usually neither is wrong: they weighted different criteria. How to compare the feedback and find the real fault.
Three retakes, same band? The same flaw is being written into every script and nobody has pointed to it. The three that repeat, and how to break the loop.
If you get the same IELTS writing band every time you retake, your English has usually not stalled. You are writing the same flaw into each script, and nobody has ever shown you where it is — so the examiner finds it again, and the four criterion scores land where they landed last time.
That is a more specific problem than "I need to practise more", and a more fixable one. What follows is why unguided practice tends to lock a band in place, the three flaws that most often travel from one test to the next, and why a result that refuses to move is actually useful information.

Three scripts, three examiners, the same line holding the band.
Each test is marked from scratch. The examiner reading your third script has never seen your first two; they read the essay in front of them and, for each of the four criteria, pick the band whose descriptor fits it best. Task Response (Task Achievement in Task 1), Coherence and Cohesion, Lexical Resource and Grammatical Range and Accuracy each get a whole band, and those scores are averaged and weighted into the number on your report.
So when three different examiners, on three different days, reading three essays written to three different prompts, all arrive at 6.5, that is not a coincidence and it is not bad luck. It means the three scripts share the features those examiners were responding to. Different topic, different vocabulary, same underlying shape — and the shape is what gets scored.
The candidates in this position have rarely been idle. Most have written dozens of practice essays between attempts. The problem is what those essays were checked against, which in most cases was nothing, or a model answer, or their own reading of their own work. None of those can see the thing that is holding the score, because if they could, it would already have been fixed.
Writing practice does two jobs at once. It builds fluency — you get faster, you run out of ideas less often, you finish on time. And it makes whatever you do repeatedly more automatic. The first job helps your band. The second only helps if what you are repeating is right.
Take a candidate whose Task 2 essays consistently answer only one half of a two-part prompt. Every practice essay rehearses that same reading of the question. By the fiftieth essay, reading the prompt that way is no longer a decision; it is how they read prompts. The fifty-first essay, written in the exam hall under time pressure, does exactly the same thing, only faster and more confidently.
This is why the second and third attempts often feel better than the first while scoring the same. The writing genuinely is more fluent. But fluency mostly shows up as speed and comfort, which examiners do not score directly, while the flaw shows up in a criterion, which they do.
Self-checking does not break the loop, for a structural reason: you check an essay with the same understanding you used to write it. If you did not see that the prompt had two parts when you wrote the essay, you will not see it when you reread it. The same is true of a grammar pattern you believe is correct, or a paragraph plan you believe is standard. The blind spot writes the essay and then marks it.
Retake scripts that come back on the same band tend to share one of three patterns. Each of them survives a change of topic, which is why a new prompt does not shake them loose.
1. One part of the prompt goes missing, every time.
Consider this prompt: Some people think governments should spend more on public transport than on roads. Discuss both views and give your own opinion.
A typical conclusion from a candidate on a repeated 6.5 reads:
In conclusion, both public transport and roads have advantages and disadvantages, and governments should consider both carefully when making decisions.
The essay discussed both views. It never gave an opinion — and "consider both carefully" is not one. At band 6, Task Response allows that "the main parts of the prompt are addressed (though some may be more fully covered than others)"; band 7 asks for the main parts to be "appropriately addressed" with "a clear and developed position". A candidate who habitually hedges in the conclusion caps Task Response at 6 on every prompt that asks for an opinion, whatever the topic, and cannot see it, because to them a balanced ending feels like the careful, academic choice.
2. The paragraph plan never changes.
Many candidates learn one Task 2 structure — introduction, one advantage paragraph, one disadvantage paragraph, conclusion — and apply it to every prompt. It fits discussion questions. It fits "to what extent do you agree" questions badly, and problem–solution or two-question prompts worse. An opinion essay built on that frame spends a body paragraph arguing a side the writer does not hold, and the position arrives only in the conclusion.
Here is the opening of such a body paragraph, written for a prompt asking how far the candidate agrees that university education should be free:
On the one hand, there are several advantages of making university free. Firstly, more students can attend. Secondly, society benefits from educated people.
Nothing in that paragraph is wrong at sentence level. But the structure makes Task Response read as undeveloped, and the reflexive "On the one hand… Firstly… Secondly" is the pattern the band 6 Coherence and Cohesion descriptor calls cohesive devices that "may be faulty or mechanical". Because the frame is used for every prompt, the same two criteria come back at the same level on every test.
3. A grammar error that has stopped being visible.
Every writer has a few errors they no longer notice, because they have been making them long enough that they read as correct. Common ones are article use with general nouns, a missing -s in third-person verbs after a long subject, and comma splices between two full clauses:
The number of cars in cities have increased rapidly, this has caused serious pollution problems.
Two errors, both habitual. The band 7 Grammatical Range and Accuracy descriptor asks for "frequent" error-free sentences. A writer who produces one of these in every second sentence cannot reach that line, and more complex sentence structures only give the habit more places to appear. Because the errors feel correct to the person making them, rereading does not catch them. A fresh reader catches them in seconds.

A stable cause is one you can find and lift.
A result that stays at exactly the same band across attempts is frustrating, but it is more useful than a result that jumps around.
If your band moved half a band up and down between attempts, the causes could be anywhere: an unfamiliar Task 1 type one time, a harder Task 2 prompt the next, a bad day. There would be little to aim at. A band that holds steady across different prompts and different examiners tells you the opposite. The cause is stable, it is in your writing rather than in the test, and it is probably one or two criteria doing the same thing each time.
Stable causes can be found and fixed. Once the missing opinion, the one-size paragraph plan or the invisible grammar habit is identified, it is usually one specific change, and the criterion it was holding down is free to move on the next test. The difficulty was never the fix. It was that nobody pointed at the problem.
Because the reported band averages the criteria and weights Task 2 double — (Task 1 + Task 2 × 2) / 3, rounded to the nearest quarter band — one criterion stuck a band below the others is often the whole difference between two bands on your report.
Start from the scripts, not the scores. You cannot change a result you cannot trace to a criterion.
Get one essay marked criterion by criterion. An overall band says nothing about where it came from. You need four separate scores for Task 2, and ideally for Task 1 as well, with the specific sentences that set each one. If you reconstruct your exam essay while it is still fresh, mark that one; otherwise use a practice essay written under timed conditions.
Look for the criterion that is lowest in every essay. One marked essay tells you about that essay. Two or three marked essays, on different prompts, show you which criterion keeps coming back at the same level. That repeated criterion is your retake problem.
Change one thing and check it held. Fix the habit — a stated position in every opinion introduction, a paragraph plan chosen for the prompt type, one grammar pattern at a time — then write the next essay and check that the target criterion moved and the others did not drop. Changing everything at once makes it impossible to tell what worked.
Only then book the test. Another attempt without a diagnosed cause is mostly a repeat of the last one.
If you do not have a teacher, how to get IELTS writing feedback without one compares the realistic options. If you already know your four criterion scores, which criterion to fix first shows how to decide where the next month goes.
Why do I keep getting the same IELTS writing band on every retake?
Usually because the same flaw appears in every script. Each examiner marks your essay fresh against the four criteria, and if the same criterion is held down by the same habit — a missing part of the prompt, a fixed paragraph plan, a recurring grammar error — the scores land in the same place on every attempt, whatever the topic.
Does getting the same band mean I have stopped improving?
Not necessarily. Practice usually improves fluency and speed, which feel like progress but are not scored directly. The band stays put because the feature an examiner is responding to has not changed, not because nothing else has.
Will writing more practice essays break the plateau?
Only if someone checks them against the criteria. Practice without feedback repeats whatever you are already doing, including the flaw, and rereading your own essay rarely finds it because you check with the same understanding you wrote with.
How many criteria usually need to change to move up half a band?
Often one. The reported band averages the four criteria in each task and weights Task 2 double, so a single Task 2 criterion moving one band shifts the result by about a sixth of a band, which is enough when you are close to a rounding boundary.
Should I book another test straight away?
It is worth finding the criterion that is holding the band first. Have one or two essays marked criterion by criterion, fix the repeated weakness, and check that it has moved in a timed practice essay before the next attempt.
Are the band estimates on this site official?
No. They are estimates made against the public band descriptors. Only a certified examiner produces an official band.
All descriptor wording quoted here is from the public Writing band descriptors updated in May 2023, published by IELTS and the British Council.
Bandly grades your essay or letter against all four criteria and shows the gap to your target band, sentence by sentence.
Two teachers, one essay, two bands. Usually neither is wrong: they weighted different criteria. How to compare the feedback and find the real fault.
Not always your lowest one. Why the cheapest criterion to move comes first, how the averaging decides it, and the order that usually works.