What 617 IELTS Writing Practice Submissions Show — and What They Do Not

Key Findings at a Glance

This dated snapshot covers 617 scored practice submissions from 69 accounts between 22 August 2025 and 10 July 2026. It contains aggregate product activity and unofficial model outputs from a self-selected sample with repeat attempts. The values were not validated against examiner-awarded scores and do not show how IELTS candidates generally perform.

• 617 scored submissions: 302 Task 1 and 315 Task 2 responses.
• 528 rows contained all four comparable stored criteria.
• 6.48 was the internal four-criterion mean across those 528 complete rows.
• 237.49 words was the average across the mixed Task 1 and Task 2 sample.
• 0 rows contained a recorded time-taken value, so this snapshot reports no timing result.

How the Internal Averages Were Calculated

The comparison uses only the 528 rows that stored numeric values for Task Achievement or Task Response, Coherence and Cohesion, Lexical Resource, and Grammatical Range and Accuracy. Each row was reduced to a four-criterion arithmetic mean for this internal audit.

• Coherence and Cohesion: 6.66
• Task Achievement or Task Response: 6.54
• Grammatical Range and Accuracy: 6.43
• Lexical Resource: 6.29

These labels organise English AIdol model outputs around the published IELTS criteria. They are not examiner-awarded component scores. The 0.37 difference between the highest and lowest internal averages is a product-audit signal, not evidence of a general learner weakness.

Distribution of Internal Four-Criterion Means

The 528 complete rows were grouped to half steps for auditing. The largest groups were mean 7.5 (95 rows), mean 6.0 (84 rows), mean 6.5 (68 rows), and mean 8.0 (63 rows). This distribution combines outputs from models used at different times and includes repeat attempts. It is not a distribution of official IELTS bands.

Writing Patterns the Snapshot Can and Cannot Show

The overall average of 237.49 words combines 302 Task 1 submissions and 315 Task 2 submissions. It should not be compared with the minimum for either task as though every row had the same prompt type. Because no selected row stored a time-taken value, reporting an average completion time would be unsupported.

The 617 submissions came from 69 distinct accounts, so a submission count is not a learner count. The sample is self-selected and includes repeat attempts. It cannot establish common IELTS mistakes, examiner agreement, scoring accuracy, improvement, causation, or performance among IELTS candidates generally.

Using AI and Human Writing Feedback

AI can support rapid iteration by returning repeatable, criteria-based suggestions. It may still miss context, reasoning, or language nuance, and any band shown is an unofficial practice estimate. A qualified teacher can provide contextual coaching, while only IELTS can award an official result.

Use automated feedback to identify one revision target and compare similar attempts. Use current official preparation materials or qualified human feedback for high-stakes judgments. Neither this internal dataset nor a single AI estimate predicts an official result.

Method and Privacy Notes

• Only aggregate counts and averages are shown; no essay text, feedback text, account identifier, or individual result is published.
• The inclusion rule selected scored IELTS Writing rows with stored band-score data at extraction time.
• The self-selected sample includes repeat attempts and model outputs produced at different times.
• No examiner comparison was available, so the snapshot cannot support accuracy, agreement, causation, or score-improvement claims.
• A new dated snapshot is required for future updates; this historical extract must not be silently overwritten.

Review the dated machine-readable aggregate snapshot used for these figures.