ISMGenius research, October 2026
The Draft Report 2026
What 970 Queensland assessment drafts tell us about where marks are won, lost and won back.
The short version
- +29points
Drafts that end on a clear answer to their own question score 29 points higher than drafts that don't. No other habit is worth as much.
See all eight habits - 69%
Evaluation criteria average 69%, the lowest of any skill. It is also among the issues students fix least often on a redraft.
Where marks go missing - 94%
of the fixes in our feedback take an hour or two at most. Most drafts are closer to done than they feel.
How big the fixes are - +9points
Redrafted assignments rose from 77% to 86% on average. Two in three were redrafted within a day.
What a redraft changes - 21%
of drafts arrive after 9 pm. Friday is the busiest day, Saturday the quietest.
When the work happens
Judging beats explaining
We asked one question of every draft we read in full: what is the deepest kind of thinking it keeps up? Nothing else we measured tracks the result as closely.
Drafts that only describe or explain averaged 32%. Drafts that analyse reached 55%. Drafts that evaluate, weighing their evidence and arriving at a judgement, averaged 85%.
The step from analysing to evaluating is worth 29 points on its own. That is what verbs like evaluate, justify and critique ask for in a task sheet: a decision, backed by reasons, not a tidier summary.
Show the numbers
| Deepest thinking sustained | Drafts | Average predicted result |
|---|---|---|
| describes or explains | 13 | 32.2% |
| analyses | 28 | 55.2% |
| evaluates | 207 | 84.6% |
Eight habits, ranked by what they're worth
Each row compares drafts that have a habit with drafts that don't. The number on the right is the difference in average predicted result.
Show the numbers
| Habit | With (drafts) | With (avg) | Without (drafts) | Without (avg) | Gap (points) |
|---|---|---|---|---|---|
| Ends with a conclusion that answers the question | 206 | 83.5% | 42 | 54.4% | +29.1 |
| Explains every piece of evidence it uses | 150 | 88.9% | 98 | 62.7% | +26.3 |
| Discusses its own limitations | 217 | 81.5% | 31 | 58.1% | +23.4 |
| Includes a reference list | 184 | 84.5% | 64 | 61.6% | +22.9 |
| Writes in a consistently academic voice | 99 | 88.2% | 149 | 72.1% | +16.1 |
| Cites sources in the text | 178 | 83.0% | 70 | 67.2% | +15.8 |
| Uses subject terms precisely | 75 | 87.8% | 173 | 74.5% | +13.3 |
| States a clear question or thesis up front | 178 | 79.9% | 70 | 75.0% | +4.9 |
The surprise is at the bottom. Stating a clear question up front is worth just 5 points. Ending on an answer to it is worth 29. Finishing moves beat opening moves: a conclusion, a limitations paragraph (+23) and evidence that is explained rather than dropped in (+26).
These are associations, not causes. A draft that concludes well tends to be a more finished draft overall, and finished drafts score higher.
Evaluation is where marks go missing
We sorted 1,412 marking criteria by the skill they assess, then averaged how much of each criterion students reached.
Evaluating and concluding sits at 69%, almost 19 points below knowing and explaining (88%). Students know their content. The marks slip away when the task asks them to judge it.
Communication criteria lose marks more often than any other: 79% of them dropped at least one. When they do, it is usually small: 1.6 marks on average, and 95% of those fixes take an hour or two. That makes communication the cheapest place to win a mark back.
Show the numbers
| Skill assessed | Criteria | Average reached | Lost at least one mark |
|---|---|---|---|
| Evaluating and concluding | 347 | 69.4% | 66.3% |
| Communicating | 213 | 71.6% | 78.9% |
| Researching and planning | 197 | 75.9% | 55.3% |
| Data and mathematics | 218 | 77.5% | 67.0% |
| Analysing | 249 | 84.1% | 39.8% |
| Knowing and explaining | 40 | 88.1% | 45.0% |
The one thing holding each draft back
For every report we read, we asked: what is the single biggest thing between this draft and a higher result? Each square is one draft.
Show the numbers
| Biggest blocker | Drafts | Share |
|---|---|---|
| Data, calculations and graphs | 77 | 27.4% |
| Missing or thin evaluation | 41 | 14.6% |
| Unfinished sections | 36 | 12.8% |
| Method and design choices | 25 | 8.9% |
| Spelling, grammar and expression | 22 | 7.8% |
| Not answering the task | 16 | 5.7% |
| Referencing | 14 | 5.0% |
| Inaccurate knowledge or terms | 13 | 4.6% |
67 drafts (19%) had nothing material holding them back. The feedback on those was polish, not repair. Knowing you are already there is worth something too.
Data, calculations and graphs tops the list because so much of this year's work came from science and maths investigations, where one wrong unit or a missing uncertainty runs through every result after it. Outside the sciences, missing evaluation and unfinished sections do the most damage.
Most fixes are smaller than they feel
Show the numbers
| Effort | Actions | Share |
|---|---|---|
| Quick edit | 433 | 48.8% |
| An hour or two | 402 | 45.3% |
| Substantial rework | 10 | 1.1% |
| New research or data | 43 | 4.8% |
A draft can feel a long way from done. The work it needs is usually short: 49% of actions are quick edits, and only 6% need new research, new data or a rebuilt section.
What the priority actions ask students to fix, as a share of all 888 actions.
Redraft once, and do it the same day
65 assignments came back for a second look or more. Here is where they started and where they finished.
Show the numbers
| First draft started at | Assignments | First draft | Last draft |
|---|---|---|---|
| 50-69% | 14 | 57.9% | 72.9% |
| 70-84% | 12 | 76.1% | 89.1% |
| 85-99% | 34 | 92.6% | 93.7% |
| All redrafts | 65 | 77.0% | 86.0% |
What a redraft fixes, and what it doesn't
Hard-to-follow structure gets resolved most often on the next draft (60%). A vague question or thesis (36%) and missing evaluation (40%) get fixed least. Those need new thinking, not editing.
Rewriting has a cost. Of the 27 new problems that appeared in redrafts, 16 were spelling, grammar and expression. New sentences bring new slips, so proofread the paragraphs you changed.
Show the numbers
| Issue | Issues tracked | Resolved next draft | New regressions |
|---|---|---|---|
| Hard-to-follow structure | 10 | 60.0% | 2 |
| Data, calculations and graphs | 50 | 54.0% | 2 |
| Unfinished sections | 14 | 50.0% | 0 |
| Not answering the task | 26 | 46.2% | 0 |
| Spelling, grammar and expression | 82 | 45.1% | 16 |
| Inaccurate knowledge or terms | 18 | 44.4% | 2 |
| Referencing | 40 | 42.5% | 2 |
| Missing or thin evaluation | 20 | 40.0% | 0 |
| A vague question or thesis | 11 | 36.4% | 2 |
Every subject loses marks somewhere different
Where criteria fell short, by subject. Each cell is that gap's share of the subject's total, so read across a row.
Show the numbers
| Subject | Gaps | Missing or thin evaluation | Data, calculations and graphs | Method and design choices | Spelling, grammar and expression | Inaccurate knowledge or terms | Claims without evidence | Evidence left unexplained | Referencing | Hard-to-follow structure | Style and genre conventions |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Physics | 98 | 22% | 35% | 16% | 8% | 0% | 0% | 5% | 0% | 0% | 0% |
| Chemistry | 48 | 17% | 17% | 15% | 6% | 17% | 4% | 0% | 13% | 0% | 0% |
| Biology | 25 | 24% | 24% | 8% | 8% | 20% | 0% | 0% | 12% | 0% | 0% |
| Psychology | 143 | 22% | 4% | 0% | 11% | 0% | 20% | 15% | 8% | 0% | 0% |
| Mathematics | 239 | 26% | 25% | 26% | 5% | 10% | 0% | 4% | 0% | 0% | 0% |
| English & Literature | 22 | 0% | 0% | 0% | 32% | 0% | 0% | 9% | 0% | 14% | 32% |
| Humanities | 64 | 36% | 3% | 0% | 16% | 13% | 14% | 0% | 5% | 0% | 0% |
Physics drafts lose most to data and calculations. Psychology drafts lose to evaluation and to claims made without evidence. Mathematics splits three ways between method, evaluation and calculation. More than a third of humanities gaps are about evaluation. English and literature are the outlier: their gaps are about style and expression, not reasoning.
When the work happens
Every report, by the hour it arrived (Brisbane time) and the day of the week.
Drafting is a school-day habit: 86% of reports arrive Monday to Friday, with Friday the busiest. Activity dips around dinner, climbs again from 7 pm, and 21% of reports arrive after 9 pm.
The assessment calendar, week by week
Show the numbers
| Week starting | Reports |
|---|---|
| 9 Mar | 12 |
| 16 Mar | 30 |
| 23 Mar | 13 |
| 30 Mar | 1 |
| 6 Apr | 0 |
| 13 Apr | 0 |
| 20 Apr | 1 |
| 27 Apr | 2 |
| 4 May | 0 |
| 11 May | 3 |
| 18 May | 14 |
| 25 May | 9 |
| 1 Jun | 28 |
| 8 Jun | 74 |
| 15 Jun | 69 |
| 22 Jun | 37 |
| 29 Jun | 3 |
| 6 Jul | 5 |
| 13 Jul | 41 |
| 20 Jul | 75 |
| 27 Jul | 42 |
| 3 Aug | 58 |
| 10 Aug | 79 |
| 17 Aug | 43 |
| 24 Aug | 45 |
| 31 Aug | 104 |
| 7 Sep | 77 |
| 14 Sep | 15 |
| 21 Sep | 12 |
| 28 Sep | 29 |
| 5 Oct | 15 |
| 12 Oct | 34 |
ISMGenius grew through the year, so later weeks carry more reports than earlier ones. Within each term, demand builds toward the Term 2 and Term 3 due dates and falls away in every school holiday.
What students ask Google
1,519 different searches showed an ISMGenius page between March and October 2026. We sorted each one by what the person wanted, weighted by how often it appeared.
Students search for the rules of the game far more than for how to play it. ATAR calculators, the meaning of a cognitive verb and the exam timetable make up 77% of these searches. Searches for how to write a task, or for examples, are about 2%.
Not shown: searches for ISMGenius by name (7%) and unclear searches (5%).
Show the numbers
| What the searcher wanted | Different searches | Appearances | Share |
|---|---|---|---|
| ATAR and score calculators | 339 | 7,553 | 31.2% |
| What a word or acronym means | 176 | 6,186 | 25.6% |
| External exam timetable | 133 | 4,999 | 20.7% |
| ISMGenius by name | 18 | 1,808 | 7.5% |
| Other or unclear | 422 | 1,271 | 5.3% |
| Rules, limits and integrity | 81 | 1,055 | 4.4% |
| Marking guides and criteria | 108 | 430 | 1.8% |
| A tool to check their work | 130 | 418 | 1.7% |
| How to write a task | 54 | 291 | 1.2% |
| Examples and exemplars | 22 | 99 | 0.4% |
| Subject content | 36 | 71 | 0.3% |
These are searches where ISMGenius appeared, so they reflect what the site already ranks for as well as what students want. Example searches are shown only where at least 40 appearances make them common phrasing, not one person's words.
Length tracks completion, then flattens
Under 1,000 words, drafts averaged 52%. Around 2,500 to 3,000 words they reached 94%. Past that, extra words stopped helping.
Short drafts are mostly unfinished drafts. Length is a sign that the sections exist, not a way to earn marks, and every task has a limit. How QCAA word limits work
Show the numbers
| Words | Reports | Average predicted result |
|---|---|---|
| <1k | 33 | 52.4% |
| 1-1.5k | 52 | 69.4% |
| 1.5-2k | 158 | 81.5% |
| 2-2.5k | 272 | 89.5% |
| 2.5-3k | 113 | 93.5% |
| 3-4k | 34 | 93.1% |
| 4k+ | 11 | 89.8% |
What to do with this
Five things worth doing on your next draft, in the order the data says they matter.
End on an answer
Write your conclusion as a direct answer to your question or thesis, then check every section leads to it. It is the most valuable habit in this report.
Explain every piece of evidence
After each quote, figure or source, add the sentence that says what it shows and why it matters for your point.
Write the evaluation early
Limitations, reliability and implications are where marks go missing, and the hardest thing to fix in a rush. Draft that paragraph before the night before.
Redraft the same day
Most fixes take an hour or two. Act on the first priority while the feedback and the draft are both fresh.
Proofread what you changed
Rewritten paragraphs bring new slips. Read them aloud before you submit.
Teachers: evaluation is the skill to model out loud. Show a class the difference between explaining a result and judging it, using one of their own drafts.
Get feedback on a draftHow we made this report
Every number here comes from draft reports generated on ISMGenius between 11 March and 16 October 2026.
- 970
reports collected
Every draft report in the window, across 872 assignments. Each adds its marks, subject, length and timing.
- 348
read in full
Read criterion by criterion: what each draft does well, where it lost marks, and what its feedback asked the student to do next.
- 12,359
categorisations
1,412 marking criteria and 888 priority actions, each sorted into a fixed set of categories: the skill it assesses, what is missing, how big the fix is.
- 99%
matched the marks
The categorising never saw the marks, so the marks make a fair check. 99% of criteria that earned full marks were found to be missing nothing, and 98% of criteria that lost marks were matched to a specific gap.
Reading the numbers
Predicted, not official
Every result here is ISMGenius's predicted result against the task's own marking guide, not a QCAA result.
Associations, not causes
Differences between groups show what tends to go together. They don't prove that a habit causes a higher mark.
Who is in the data
Students who chose to use ISMGenius, weighted toward the sciences and toward Year 12 internal assessments. It is not a random sample of Queensland students, and a few students who redraft often appear more than once.
Weighted to recent months
ISMGenius grew through the year, so the data leans recent: 69% of reports arrived after Term 3 began in July. Earlier months are thinner because fewer students used the site then, not because less work was due.
Searches
Search data comes from Google Search Console for ismgenius.com, 24 March 2026 to 14 October 2026. Google names the search behind about a quarter of all appearances.
Next edition
New reports are categorised the same way as they arrive, so the 2027 edition will cover a full school year and can show what changed.
Cite this report
ISMGenius. (2026). The Draft Report 2026: Where Queensland assessment drafts win and lose marks. https://www.ismgenius.com/reports/draft-report-2026