How we score your practice, and what a score can tell you.
This page explains our method in plain terms, including what we have and have not measured. When the method changes, the version number changes and this page says what changed.
Writing rubric w2026.1 · Listening and reading table v2026.1 · Updated 22 Oct 2026
How each skill is scored
Listening and reading have correct answers. Writing and speaking do not, so we score them against the public IELTS criteria. Only the first two are marked by fixed rules.
Answer key and a versioned conversion table
Machine-marked by rules. No AI judgement is involved.
- 1Each answer is marked against the answer key.
- 2Correct answers give a raw score out of 40.
- 3A conversion table turns the raw score into a band. The table has a version, currently v2026.1, and your report shows the version it used.
Public criteria, each with quoted evidence
Scored by an automated scorer, never by a human examiner, and only once a scorer version has passed the checks in section 04. No speaking scorer has been released yet.
- 1Writing Task 1 is scored on task achievement (TA), coherence and cohesion (CC), lexical resource (LR) and grammatical range and accuracy (GRA). Task 2 uses task response (TR) in place of TA.
- 2Speaking is assessed on fluency and coherence (FC), LR, GRA and pronunciation (P).
- 3Every writing criterion band quotes the parts of your answer that support it. We check each quote against your stored answer. No quote, no band.
Limits
- A scorer reads your text or your recording. It does not see your handwriting, your test-day conditions or an examiner’s judgement on the day.
- Bands move in half steps. A result is a range, so two attempts can differ by half a band without any change in your English.
- The listening and reading conversion table (v2026.1) and the double weight of writing Task 2 are provisional. They follow commonly published approximations and are not yet checked against an official General Training source.
- Criterion bands, such as coherence and cohesion, never map to a CLB level. Only skill bands do.
Why results are ranges
One essay or one reading section is a small sample of your English. On another day, with another topic, the same person often scores half a band higher or lower. A single number would hide that.
So we show the range your evidence supports, and a dot for the most likely band inside it. The range reflects the marks of that one attempt.
What “Preliminary” means
A result is marked Preliminary when its evidence is not yet enough to treat it as a firm level. It gives you a direction, not a level. A Preliminary range is drawn hatched.
A result is Preliminary when
- It comes from the quick diagnostic, which samples each skill once.
- It uses a conversion table that is not yet checked against an official General Training source. Today that is every listening and reading band (table v2026.1).
The label goes away when
- You complete the full assessment of that skill.
- A listening or reading band stays Preliminary until its conversion table is checked.
How we check our scoring
Before a writing or speaking scorer goes live, we test it on answers it has never seen. Marks from IELTS examiners or trained raters are the reference. A scorer is released only if all three gates hold:
- Average error
- 0.50 band or less
- The mean absolute difference between the scorer and the reference marks.
- Close agreement
- 85% or more within 0.5
- Share of answers where the scorer is within half a band of the reference.
- Bias
- 0.15 band or less, up or down
- The scorer must not systematically inflate or deflate bands.
Status. No scorer has a published passing run yet. Writing is waiting for a licensed set of examiner-marked answers, and speaking for licensed audio with examiner marks. Until a run passes, this page publishes no accuracy figure.
| Scorer | Test set | Exact match | Within 0.5 | MAE | Bias | Status |
|---|---|---|---|---|---|---|
| Writing | No examiner-marked answers yet | no figure | no figure | no figure | no figure | Not yet evaluated |
| Speaking | No licensed audio yet | no figure | no figure | no figure | no figure | Not yet evaluated |
Exact match is the share of answers with the same band as the reference. MAE is the mean absolute error in bands. Bias below zero means the scorer gives lower bands than the reference on average. Each row names its test set and the number of answers (n). A figure describes that set only and is not a promise about your result. A synthetic set, if one is ever shown, is labelled synthetic and is not evidence for release.
What we never claim
No predicted test score
Practice results show your current evidence. They do not say what you will score on test day.
No “ready to book” verdict
We show where you meet your target. The decision to book is yours.
No guaranteed band or CLB level
No method can guarantee a result on a test we do not run.
No official status
Aptaneo is not affiliated with IDP, British Council or Cambridge. Our results are not IELTS results.
No claim that a scorer is always right
A test set is finite. A good result on it describes that set, not every learner.
IELTS bands and CLB levels
For IELTS General Training, IRCC sets the band range in each skill for each CLB level. You must reach the level in every skill to claim it. We compare skill bands only.
| CLB level | Listening | Reading | Writing | Speaking |
|---|---|---|---|---|
| CLB 10 | 8.5–9.0 | 8.0–9.0 | 7.5–9.0 | 7.5–9.0 |
| CLB 9 | 8.0 | 7.0–7.5 | 7.0 | 7.0 |
| CLB 8 | 7.5 | 6.5 | 6.5 | 6.5 |
| CLB 7 | 6.0–7.0 | 6.0 | 6.0 | 6.0 |
| CLB 6 | 5.5 | 5.0–5.5 | 5.5 | 5.5 |
| CLB 5 | 5.0 | 4.0–4.5 | 5.0 | 5.0 |
| CLB 4 | 4.5 | 3.5 | 4.0–4.5 | 4.0–4.5 |
Source: IRCC language test equivalency table, checked 2026-09-29. Open the IRCC table
Convert your own bands to CLB