Grading guide

How to grade yourself

BandTrack just stores your scores — this guide tells you what to score. For Speaking and Writing, self-grade against the official band-descriptor rubric — that's how the real test works. AI grading is offered as a fallback, but read the warning first: it's known to inflate or distort scores. Listening and Reading you grade yourself from the answer key.

Start here

Community-vetted resources

Read these before grading anything. They cover prep material, evaluation methods, and the most common candidate questions for all four sections.

Recommended

Speaking — grade yourself with the rubric

The workflow
No AI. Use the official band-descriptor rubric, simplified.
  1. Record yourself answering one question. Don't re-record — you want an honest snapshot.
  2. Play it back once straight through, then a second time pausing to mark issues against the checklist below.
  3. For each of the four criteria (FC, LR, GRA, P), pick the column in the rubric that best matches your performance.
  4. Log the four bands in Log score. BandTrack averages them into your Speaking band.
CriterionBand 5Band 6Band 7Band 8
FC
Fluency & Coherence
Frequent pauses & self-correction; only simple linking.Willing to speak at length, but with repetition and noticeable hesitation.Speaks at length with only occasional hesitation; ideas connect logically.Fluent throughout; rare self-correction; coherent and well-developed.
LR
Lexical Resource
Limited vocab; manages basic / familiar topics only.Enough vocab for most topics, with repetition and some inaccuracy.Flexible vocab; some less common and idiomatic items; good paraphrasing.Wide vocab used naturally; skilful paraphrase; only rare slips.
GRA
Grammatical Range & Accuracy
Mostly basic sentences; frequent errors that can confuse the listener.Mix of simple and complex forms; errors don't usually block meaning.Range of complex structures; majority of sentences are error-free.Wide range of structures used flexibly; mostly error-free.
P
Pronunciation
Often unclear; first language strongly affects intelligibility.Generally clear; some L1 features and mispronounced words.Mostly clear; sustained features of connected speech (linking, stress).Easy to understand throughout; flexible intonation, sounds & stress.

Scoring shortcut

  1. Pick the column that best fits each criterion.
  2. Fits perfectly → that band. Partly fits the next column up → add 0.5. Miss one descriptor in your column → drop 0.5.
  3. Average the four → your Speaking band.
Playback checklist
Concrete things to listen for. Each one nudges a band up or down.
  • Pauses longer than ~3 seconds without a purpose? FC −
  • Same 2–3 connectors on repeat (and, but, because)? FC / CC −
  • Count distinct topic-specific words. Fewer than ~8 in a one-minute answer = LR weak.
  • At least 2 complex sentences (relative clause, conditional, passive)? GRA +
  • Word endings (-ed, plural -s) audible? P +
  • Could a stranger follow the whole answer first try? Yes = P 7+.

BandTrack averages your four criteria into a Speaking band, exactly like the real test.

Recommended

Writing — grade yourself with the rubric

The workflow
Print it (or open it on a second screen) and mark it up. Best done 24 hours after writing — distance helps you spot your own errors.
  1. Underline each part of the original task prompt. Tick each one as you find it addressed in your response.
  2. Run through the four checklists below, noting issues in the margin ("repeated word", "no topic sentence", "verb tense").
  3. For each criterion, pick the band column in the rubric that best fits, then log it in Log score.

Note: the rubric below uses Task 2 wording. For Task 1, replace Task Response with Task Achievement — same idea, just check that you covered every feature of the chart / letter / diagram.

CriterionBand 5Band 6Band 7Band 8
TR
Task Response (Task 1: Task Achievement)
Addresses the task only partially; position unclear; format may be inappropriate.Addresses all parts, though some more than others; position relevant but not always clear.Addresses all parts; clear position throughout; ideas extended and supported.Fully addresses every part; well-developed position; relevant, extended, supported ideas.
CC
Coherence & Cohesion
Some organisation but lacks overall progression; repetitive or faulty linking.Information arranged coherently; clear overall progression; some linking errors.Logically organised; clear progression; range of cohesive devices, sometimes over/under-used.Sequenced logically; skilful paragraphing; cohesion managed so well it's barely noticeable.
LR
Lexical Resource
Limited range; noticeable errors in spelling and word choice that may distract.Adequate range for the task; some less common vocab; occasional errors.Sufficient range for flexibility; some less common items; occasional collocation errors.Wide range used naturally; rare errors only as slips; skilful use of less common items.
GRA
Grammatical Range & Accuracy
Limited range; frequent errors; punctuation often faulty.Mix of simple and complex forms; errors occur but rarely reduce communication.Variety of complex structures; majority of sentences error-free; few errors.Wide range of structures; mostly error-free; only occasional non-systematic errors.
Markup checklist
Run through these in order. Each section maps to one criterion.

TR — Task Response

  • Did I answer every part of the question?
  • Word count ≥ 250 (Task 2) / ≥ 150 (Task 1)? Below = capped at Band 5.
  • Clear position in both intro and conclusion?

CC — Coherence & Cohesion

  • 4–5 paragraphs, each with one central idea?
  • Topic sentence at the start of each body paragraph?
  • Linking words used and varied (not just firstly / secondly / finally)?

LR — Lexical Resource

  • Highlight the 10 strongest words/phrases. Topic-specific or generic?
  • Any key word repeated 4+ times? Swap one for a synonym.
  • Spelling and word-form errors? Each one drags the band.

GRA — Grammatical Range & Accuracy

  • Count complex sentences (subordinators, relative clauses, conditionals). Aim for ≥ 5.
  • Count error-free sentences. Band 7 ≈ "majority error-free"; Band 8 ≈ "wide range, mostly error-free".
  • Punctuation consistent (commas, full stops, capitals)?

Self-graded from the key

Listening — count your correct answers

The workflow
  1. Take a full 4-part Listening test (Cambridge IELTS books 15–19 are the gold standard).
  2. Mark it against the official answer key. Count correct answers per part out of 10.
  3. Log each part separately in Log score (e.g. Part 3 → 7/10). BandTrack converts your raw total to an indicative band using the official IDP / British Council conversion table.
Optional: have AI explain why you missed a question
Listening review prompt
I'm reviewing IELTS Listening Part [N]. Here is the relevant transcript
section, the question, the correct answer, and my answer:

Transcript: [...]
Question:   [...]
Correct:    [...]
Mine:       [...]

In 2–3 sentences, explain why the right answer is right and which
listening skill I missed (paraphrase, distractor, number/spelling,
signposting, etc.). Then give one drill I can do this week.

Self-graded from the key

Reading — same idea, three passages

The workflow
  1. Time yourself: 60 minutes for all three Academic passages, no extra time for transferring answers.
  2. Score each passage out of ~13 (Passage 1 & 2) or ~14 (Passage 3) and log each one separately.
  3. BandTrack converts your raw /40 to a band using the same official Academic Reading table the test uses.
Optional: AI post-mortem on missed questions
Reading review prompt
I'm reviewing IELTS Academic Reading Passage [N]. For each question I got
wrong, here is the question, the correct answer, my answer, and the
sentence in the passage that contains the answer:

[paste each one here]

For every question, tell me in 2 sentences which skill I missed
(paraphrase spotting, scanning, T/F/NG logic, matching headings, etc.)
and what I should look for next time.

Now what

How to use BandTrack effectively

  1. 01

    Log every attempt — even the bad ones.

    Trends matter more than any single score. Skipping a rough day skews your average upward and hides real weaknesses.

  2. 02

    Stay consistent in your source.

    Cambridge official tests are calibrated. Random YouTube mocks aren't. Mixing them makes your bands incomparable across weeks.

  3. 03

    Paste AI feedback into the Note field.

    For Speaking and Writing, the band number alone is useless three weeks later. The reasoning and the rewritten paragraph are what you'll re-read.

  4. 04

    Check the Dashboard weekly, not after every attempt.

    Look at the radar chart and the trend lines, not the last data point. One bad attempt is noise; three in a row is signal.

  5. 05

    Focus on the weakest area card.

    Your overall band is dragged down by your lowest section. Writing especially — Task 2 counts double, so a weak Task 2 hurts twice as much.

  6. 06

    Back up at least once a week.

    Everything lives in this browser. One cache clear and it's gone. Open Settings → Export and save the file somewhere safe.

  7. 07

    Don't chase 9s on the sliders.

    Be honest with the AI's grading — overscoring yourself feels good for a day and hides the work that would actually move your band.