What NAATI CCL candidates miss most: data from 2,762 real practice segments
In 2,762 real practice segments marked on naaticcl.com between 15 August 2026 and 21 October 2026, candidates most often lost names of organisations and services: 65.7% of these information points were left out or changed, against 50.3% across all information points.
- Next most often lost: conditions such as if and unless (58.2%) and words that set the order of events, such as before, after and until (57.5%).
- Learners lost 60.2% of information points when interpreting English into the other language, against 40.2% the other language into English.
- Segment length matters: segments of up to 15 words lost 30.4% of information points; segments of 26 to 35 words lost 57.4%.
- Of the topics with at least 100 segments, Insurance was the hardest (53.3% of information points lost) and Employment the easiest (45.8%).
Last updated 21 October 2026 · 2,762 segments · 243 learners · How we counted
Which information gets lost
Each bar is the share of information points of that type that learners left out or changed. Overall, 50.3% of all information points were lost.
Into English or into the other language
Longer segments lose more
Hardest topics
By language
Languages with enough practice have their own page: the five types learners lose most, each with an example from our own dialogues and how to keep it.
| Language | Segments | Learners | Lost overall | Lost most often |
|---|---|---|---|---|
| Urdu | 492 | 52 | 44.4% | Organisations and services (76.0%) |
| French | 444 | 12 | 51.3% | Conditions (73.2%) |
| Bangla | 227 | 13 | 54.6% | Order of events (70.9%) |
| Spanish | 176 | 16 | 51.8% | Order of events (60.7%) |
| Punjabi | 162 | 18 | 47.9% | Conditions (58.1%) |
| Cantonese | 122 | 10 | 46.3% | Order of events (57.6%) |
| Swahili | 118 | 10 | 52.4% | Conditions (69.2%) |
1,021 more segments in 25 other languages are counted in the totals above; each had fewer than 100 segments or 10 learners, too few for its own page.
How to stop losing these points
- Organisations and services. Say the organisation's name as it is (Centrelink, Fair Work Ombudsman, myGov) instead of a vague word like the office or the government.
- Conditions. Keep the if or unless part attached to the action it controls. Do not turn a condition into a plain statement.
- Order of events. Keep before, after, until and then, and give the steps in the order the speaker gave them.
- Dates and times. Note day, date and time as one block and keep them as said: next Monday stays next Monday, in two weeks is not soon.
- Negatives. Check the polarity before you speak. One missed not or no reverses the message.
Common NAATI CCL mistakes guide · How our AI marking works · Free practice test
How we counted
- Source. Every segment a learner recorded and had marked on naaticcl.com between 15 August 2026 and 21 October 2026: 3,650 marked segments in total. Our AI marker splits each source segment into information points and judges each one as kept, changed, left out or added. These are estimates from our marker, not NAATI marks.
- What we left out (888 segments). Our own and test accounts (21), repeats of a segment the same learner had already done (124), replays, reviews and targeted drills (40), segments where marking failed or the learner answered in the wrong language (284), unreliable recordings (48), and segments we could not match to a dialogue in our current library (371). That leaves 2,762 segments from 243 learners (signed-in accounts, or anonymous devices and practice sessions).
- Types of information. Each segment in our library has a model answer with a list of the information points to keep. A fixed set of text rules sorts each point into numbers, dates and times, money, people's names, organisations, negatives, conditions and order of events; a point can belong to more than one type, and points that fit none are "other". No AI is used for this step.
- Matching. Marker judgements are matched to the model-answer points in order. When the counts differ, they are matched by position; in 1,103 of 2,747 segments the counts were the same. Counting only those segments, the types lost most were organisations and services, conditions, order of events. A second method that does not match points at all (each segment's lost share set against the mix of information in it) puts organisations and services, people's names, negatives highest. The marker's own split and the model-answer points are made separately, so treat single percentages, and especially types with few points, as approximate.
- Lost means left out or changed in meaning. Additions are not counted. A justified omission, such as a dropped filler word, is still counted, so the rates overstate real mistakes somewhat.
- Weighting. No learner counts for more than 30 segments in any language, so one very active learner cannot set a language's result. Segment and learner counts are reported as they are.
- Thresholds. A language gets its own page only with at least 100 segments from at least 10 learners; other languages are counted only in the totals. Topics, directions and segment lengths are shown only with at least 100 segments (50 for the two directions within one language). A type is ranked only with at least 100 information points on this page and 20 on a language page.
- Limits. These are learners who chose to practise on our site, on our original dialogues, not NAATI test candidates or test material. The figures describe practice, not NAATI results. No learner's words, recordings or personal details are published: only counts.
Download the counts: atlas.json (aggregate numbers only). Recalculated every Monday.
Questions
What do NAATI CCL candidates miss most often?
In 2,762 real practice segments marked on naaticcl.com between 15 August 2026 and 21 October 2026, candidates most often lost names of organisations and services: 65.7% of these information points were left out or changed, against 50.3% across all information points. Next came conditions such as if and unless (58.2%) and words that set the order of events, such as before, after and until (57.5%).
Is interpreting into English or into the other language harder?
In our practice data, learners lost 60.2% of information points when interpreting English into the other language, against 40.2% the other language into English.
Do longer segments cost more marks?
Segment length matters: segments of up to 15 words lost 30.4% of information points; segments of 26 to 35 words lost 57.4%. Longer segments carry more details to hold in memory, so note-taking matters more.
Which CCL topic is hardest?
Of the topics with at least 100 segments, Insurance was the hardest (53.3% of information points lost) and Employment the easiest (45.8%).
Is this NAATI data?
No. These figures come from practice on naaticcl.com, an independent practice site, marked by our AI marker. They are not NAATI results and not NAATI test material.
How often is this updated?
Every Monday, from all practice marked on the site since 15 August 2026. Last updated 21 October 2026.