Everything candidates and instructors ask us about how BAND9AI grades against the official public band descriptors, recreates the pressure of the real exam, and pinpoints the one penalty keeping you stuck at Band 6.5. No jargon, no vague promises, just how the diagnostic engine actually works.
How a band comes out of the engine, and why it lines up with what a human examiner would give.
Does your answer fully address the prompt, take a clear position, and extend every idea?
Logical paragraphing, controlled linking, and a thread the reader never loses.
Range, precision, and natural collocation instead of memorised phrase lists.
Sentence variety and error density judged the way an examiner counts them.
Every response is assessed criterion by criterion, exactly the way a trained examiner works. For Writing that means four separate judgements (Task Response, Coherence & Cohesion, Lexical Resource, Grammatical Range & Accuracy) rather than one blended impression. The engine is calibrated against the publicly available IELTS band descriptors, so each score comes with the descriptor language that justifies it.
You do not just see a number. You see which band your writing sits in for each criterion, the specific wording in the descriptor that you have met, and the wording you have not met yet. That is the difference between "this feels like a 6.5" and "your Task Response is capped at 6 because your third body paragraph asserts a claim without developing it."
The four criterion scores are averaged and rounded to the nearest half band, following the standard IELTS Writing method. This is why a single weak criterion drags your whole score down: three sixes and one seven still average to a 6.5, no matter how strong the rest of the essay reads.
Understanding this maths is the point. It tells you where to spend your effort. Lifting your weakest criterion by one band usually moves your overall score more than polishing a criterion you are already strong in.
Listening and Reading are marked objectively against the answer key and converted to a band using the standard raw-score bands, so those results are precise. Writing and Speaking are the productive skills where the descriptor-based judgement matters most, and that is where the criterion-level feedback goes deepest.
Across all four skills the goal is the same: not just a band, but a clear reason for the band and the next concrete step to raise it.
No. General chat tools are built to be encouraging, so they tend to praise effort and suggest small tweaks. That feels good and changes nothing about your band. BAND9AI is built to be an examiner, so it is deliberately unsentimental: it tells you which criterion is being penalised and why, in descriptor terms.
The other difference is calibration. A generic model has no fixed reference point for what a Band 7 actually looks like on a given task. BAND9AI is anchored to the public band descriptors, which keeps its scoring consistent from one essay to the next. Compare the two in BAND9AI vs ChatGPT vs human tutors.
Why practice that feels comfortable rarely predicts what happens on test day.
The simulation uses the real format and the real clock. Writing runs to the standard timing, Reading holds you to the same window candidates get on the day, and Listening plays once, at exam pace, without a rewind. That single detail matters more than most people expect: many candidates score well at home only because they quietly pause, replay, or run over time.
Practising under authentic constraints surfaces the mistakes that only appear under pressure, such as rushing a conclusion, mismanaging Reading time, or freezing on a Speaking prompt. Those are the errors that decide your real band.
Because untimed, self-marked practice removes exactly the conditions that cost you marks. When you give yourself extra minutes, look up a word, or grade your own essay generously, you are rehearsing a test that does not exist.
A realistic simulation closes that gap. It shows your band under exam conditions, which is usually lower than your relaxed practice score and far closer to what you will actually get. Knowing the real number early is what lets you fix it before you pay for another sitting.
The quick Reality Check returns your estimated real band, the single penalty pattern capping your score, and the exact fix in roughly two minutes. A full section simulation takes as long as the real exam section, because the timing is part of the diagnosis.
Start short to see where you stand, then run the full simulation when you want the complete, exam-accurate picture.
Yes. Repeated timed attempts are how you turn a one-off diagnosis into a reliable band. Each retake shows whether the fix you applied actually moved the criterion it was meant to move, so you are measuring progress rather than hoping for it.
The number of retakes available depends on the plan you choose. Check the current plan details at checkout, since access levels are set there rather than on this page.
Pricing
Reality Check $15 one-time. Skill Fix $29/mo. Complete System $49/mo. Hit your target band — or money back.
$49 /mo
Save $67 vs 4 Skill Fixes separately
Hit your target band — or get your money back. That’s the moat.
$29/mo per skill
$15
Listening. You hear it once. Miss it → lose marks instantly.
Reading. Most fail because they run out of time.
Writing. One weak area caps your entire score.
Speaking. You freeze or give short answers under pressure.
Why more practice essays stop working, and what actually moves the needle.
Because volume alone does not fix a specific weakness. If you write ten more essays with the same undeveloped arguments, you get ten essays with the same Band 6 Task Response. The plateau is not a lack of effort, it is repeating an error you have not been shown clearly enough to correct.
Breaking it starts with diagnosis: find the one criterion that is holding your average below 7, understand precisely why it sits where it does, and target that. Fixing the weakest link almost always beats improving the parts you are already good at. See best AI IELTS tool for Band 6.5 students.
In Writing Task 2 it is under-developed Task Response. Candidates hit the word count but never fully extend an idea: they state a claim and move on instead of following it with an explanation, an example, and a consequence. The descriptor for Band 7 asks for ideas that are extended and supported, so a stack of thin points caps the score.
Other frequent culprits are memorised linking phrases that hurt Coherence & Cohesion, and repetitive sentence structures that keep Grammatical Range from reaching 7. The diagnostic tells you which of these is yours rather than leaving you to guess.
Stop writing more essays for a moment and diagnose which of the four criteria is being penalised. Then apply a targeted fix to that one criterion, write again under timed conditions, and check whether the score for that criterion actually rose.
Repeat the loop until the weak criterion reaches your target band. That is a deliberately narrow, measurable path, which is why it works where general practice stalls.
No honest tool can guarantee a band on your behalf, because your result depends on how you apply the feedback. What the engine guarantees is clarity: an accurate read of your current level, the exact criterion holding you back, and a concrete fix mapped to the descriptors.
Some plans include an outcome-linked refund policy. The specific terms are set at checkout, so review them there rather than relying on a summary here.
How teachers use the grading tools, and the everyday details before you start.
Yes. Teachers use the criterion-level scoring to grade faster and more consistently, and to give students descriptor-based feedback they can act on rather than a single number. Because the engine anchors to the public band descriptors, it helps standardise marking across a class or a whole cohort.
It works best as a second opinion that speeds up your workflow: the machine handles the repetitive first pass on every criterion, and you add the human judgement and coaching on top.
No. BAND9AI is an independent preparation and diagnostic tool. It is not the official IELTS test and is not affiliated with the organisations that run IELTS. It uses the publicly published band descriptors as its reference to give you exam-style feedback before you sit the real thing.
The diagnostic focuses on the four skills, Listening, Reading, Writing and Speaking, which are the areas where descriptor-based feedback makes the biggest difference. If you are preparing for a specific module, confirm the exact task types included on the current plan pages at checkout so you know what your practice will cover.
Very little. You can start by pasting a few sentences of your writing or describing the score problem you keep running into. The quick Reality Check reads that and returns your estimated band, the penalty pattern capping it, and the fix.
For a full picture, run a complete timed section so the result reflects true exam conditions rather than a short sample.
The ladder
Start at 2♠ (diagnostic) · Level up J → Q → K → A (black = Listening/Reading, red = Writing/Speaking) · Unlock Joker (complete system).
Start with the Reality Check, a full four-skill diagnostic, then decide which Skill Fix you need.
$15
All four skills · entry diagnostic
Start here, full diagnostic
Payments processed by Stripe. Seller: BAND9AI HUMAN SYSTEMS INC., Toronto, Canada.
J & Q are black suits (analytical: Listening & Reading). K & A are red suits (expressive: Writing & Speaking). Each tier: simulation, correction, breakdown, retakes until you improve.
Miss it once = lose marks
$29/mo
✓ Monthly subscription · Listening
✓ Listening simulation
✓ AI correction
✓ Mistake breakdown
✓ Retake until improvement
Step 1: Fix how you hear
Payments processed by Stripe. Seller: BAND9AI HUMAN SYSTEMS INC., Toronto, Canada.
Run out of time
$29/mo
✓ Monthly subscription · Reading
✓ Reading simulation
✓ AI correction
✓ Mistake breakdown
✓ Retake until improvement
Step 2: Fix how you process
Payments processed by Stripe. Seller: BAND9AI HUMAN SYSTEMS INC., Toronto, Canada.
Stuck at 6.5
$29/mo
✓ Monthly subscription · Writing
✓ Writing simulation
✓ AI correction
✓ Mistake breakdown
✓ Retake until improvement
Step 3: Fix how you write
Payments processed by Stripe. Seller: BAND9AI HUMAN SYSTEMS INC., Toronto, Canada.
Freeze under pressure
$29/mo
✓ Monthly subscription · Speaking
✓ Speaking simulation
✓ AI correction
✓ Mistake breakdown
✓ Retake until improvement
Step 4: Perform under pressure
Payments processed by Stripe. Seller: BAND9AI HUMAN SYSTEMS INC., Toronto, Canada.
Fix everything at once
Billed monthly: tracking across all four skills, unlimited practice until your target band.
$49/mo
Save $67 vs 4 Skill Fixes separately
Listening + Reading + Writing + Speaking
✓ Monthly subscription, all four skills
✓ Full simulations and feedback for all skills
✓ Weakness tracking across skills
✓ Unlimited practice and retakes
Final form, fix everything
Payments processed by Stripe. Seller: BAND9AI HUMAN SYSTEMS INC., Toronto, Canada.
If your situation is more specific, the fastest answer is a diagnosis of your own writing. Run the free two-minute check, see your real band, and find out exactly which criterion is holding you back.
Plan details, pricing and refund terms are confirmed at checkout.
BAND9AI is an independent preparation tool. It is not the official IELTS exam and is not affiliated with the organisations that run IELTS. Band estimates are practice guidance, not official results.