How far off are our predictions?
Loading…
Every exam-prep product claims to be “accurate.” There is only one way to check: put the predicted score next to the student’s real NAATI result, then publish the sample size and the error. That number is on this page, live — we can’t massage it, it is computed from results students report back.
Sample size
—
Real results reported so far
Mean absolute error
—
Gap between predicted and real score (out of 90)
Within ±3
—
Share of predictions within 3 marks of the real score
Within ±5
—
Share of predictions within 5 marks of the real score
How it is calculated
- Where the sample comes from: students enter their real result in “My profile” after the exam. It’s voluntary and does not affect your use of the platform — except for the $199 green-light guarantee, whose terms require you to report your result after sitting.
- How the error is computed: only records that have both a predicted and a real score go into the denominator. A student who reports a real score without ever doing a mock on the platform counts towards sample size but not towards error.
- Why mean absolute error: it is literally “how many marks off, on average,” with no dressing up. We don’t use correlation coefficients that look impressive but tell a student nothing.
- We do not filter the sample: no removing “outliers,” no cherry-picking time windows. If it was reported, it counts.
The official pass mark
You pass with 63 / 90. Each of the two dialogues is worth 45, and neither may fall below 29 — both conditions must hold. A high total with one collapsed dialogue is still a fail. (Per NAATI’s CCL page, checked 2026-08-04)
Results reported so far