Conduct a Test Sitting
Five steps, one page. Language Tests → Testing Center → pick an organization → Conduct a test.
1. Name the sitting and set the window
Give the event a name your colleagues will recognise months later — "Feb intake placement — Campus B" beats "Test 3".
| Field | What it does |
|---|---|
| Opens | Before this time, a candidate's link or code will not start the test |
| Closes | After this time, the sitting is closed |
| Leave Opens empty | Candidates can start as soon as they are invited |
| Leave Closes empty | The sitting stays open until you close it |
The window is per event, so two sittings on the same day can have different hours.
2. Choose the paper
Two options:
- One fixed test — everyone sits the same paper. Simplest, and right for most sittings.
- A pool — the system gives each candidate a paper from a set you choose. Use this when candidates sit side by side. See Test pools.
Exam conditions
Once you pick a paper, the wizard shows the exam-condition switches that paper supports. They are the same choices a student sees when starting the test themselves, decided once here for the whole sitting. Candidates cannot change them after the test starts.
| Switch | Shown for | What it does |
|---|---|---|
| Listening audio plays once | Papers with a listening section | Each part auto-starts after a 30-second preview; no pause, rewind or replay. On by default for real-exam papers, off for practice papers |
| Real exam speaking flow | Papers with a speaking section | Preparation and recording start automatically, with no review or re-record. Off by default |
| IELTS layout | Real-exam IELTS papers | Real exam layout numbers questions 1–40 exactly like IELTS; Classic uses the standard practice panel. The choice is pinned for the sitting, so the in-test layout switch is hidden |
What you will see, by exam:
| Exam | Listening audio once | Speaking flow | IELTS layout |
|---|---|---|---|
| IELTS real-exam papers | Yes | Yes | Yes |
| IELTS practice, TOEFL, PTE, OET, CELPIP | Yes | Yes | — |
| Custom tests | If the paper has a listening section | If the paper has a speaking section | — |
| GRE, GMAT, SAT, dMAT | — | — | — |
| Duolingo (DET), Adaptive Language Proficiency | — | — | — |
GRE, GMAT, SAT and dMAT have no audio. DET and ALP play listening and speaking inside their own adaptive flow, which already runs exam-style, so there is nothing to switch. A paper with no switches shows a short note saying so; for a pool each candidate gets only the conditions their own paper supports. The candidate's start page repeats the conditions before they press Start, and the event page shows them as chips.
3. Decide how results are released
| Policy | What happens when a candidate finishes |
|---|---|
| Release immediately | The candidate sees the score straight away |
| Hold for examiner review | The result waits in the review queue. An examiner checks it, can adjust the score, and releases it |
Choose review for anything that goes on the record — admissions, employer reports, certificates you sign. Choose immediate for quick internal placement.
Writing and speaking are marked by AI either way. Review is about whether a human sees it before the candidate does.
3b. Find a sitting, a day, or a candidate
The organisation page opens on Sittings: every sitting under the day its candidates sit, with booked / entered / sat / to mark / released and the average, rolled up per day. Pick a period, or a campaign. A day header opens the day view — every sitting on that day, results by level or band, a per-group and per-venue breakdown across all of them, and the merged roster. The schedule's day cells open the same view. The Candidates tab is the flat list with scheduled-day, completed-date, sitting, group and sort filters.
3c. Who is here, and sending each group its results
Attendance (sitting page → Attendance) shows expected / entered / sat / not yet / absent for the sitting, per group and per venue, with the lists of who has not entered yet and who is absent. Entered means the candidate started the test online or used their printed code — it is not proof of physical presence. Absent means their window closed without them entering: nothing is clicked, the clock is the record. The numbers refresh every 30 seconds; the register is downloadable as CSV, for the whole sitting or one group.
Send results (sitting page → By group → Send results) emails one group its released results as a CSV: name, reference, group, venue, slot, overall score, CEFR level, verification code and the public verify link — nothing else. Phone numbers, parent emails and every other roster column never travel. Results still under examiner review are held back automatically, and a group with nothing released cannot be sent. The preview shows the recipient, the exact columns and every row before you confirm; every send, successful or failed, is recorded under the group with who sent it and when.
4. Load the roster
| Method | Best for | How |
|---|---|---|
| Paste | A handful of people | One per line: Name, email, Group — or Name, reference, Group for a printed code instead of an email. The group is optional |
| CSV upload | Hundreds or thousands | Download the template, fill it in, upload it. Up to 5,000 rows and 5 MB per upload. A header row names the columns |
| Codes only | Candidates with no email | Supply a name and a reference number (roll number). The system generates a printed access code for each |
Metadata: tell us who each candidate is, once, at import
Anything you know about a candidate goes in the roster and comes back in every report, export and seat sheet. A header row decides what each column means:
| Column (aliases) | Used for |
|---|---|
name |
The candidate's name |
email |
Email invite. If the candidate already prepares with us under this email, the result lands in their own account too |
ref (roll, id, reg no) |
The group's own reference. Required for a printed code; used to match re-uploads |
group (school, college, institution, campus, cohort, batch, class) |
The group every report is segregated by. The first of these present becomes the group; any others are kept as extra columns |
venue (hall, room, centre, site, lab) |
Where the candidate sits. Filterable on its own, so a group split across two halls, or a hall shared by two groups, can each be counted |
slot (or date + time as two columns) |
That candidate's own start time, e.g. 2026-10-14 09:00. They cannot start before it, and may start up to the sitting's late-start allowance after it (60 minutes by default); after that their slot has closed even if the sitting is still open. A test already under way is never interrupted |
| anything else | Kept exactly as given (class, section, phone, centre code …) and shown on the candidate page and in every CSV |
One roster can mix both access kinds: a line with an email is invited by email, a line with only a reference gets a printed code. A candidate's identity in a sitting is their group plus reference, so two groups can both have a roll number 001. Re-uploading a roster with corrected metadata updates the metadata; it never creates duplicates, as long as the group label is the same one used the first time. A repeated column name, a value longer than 200 characters, or an unreadable slot is reported on that line and nothing is silently dropped or cut.
What the numbers mean
| Number | Definition |
|---|---|
| Booked | Candidates on the sitting, not cancelled |
| Entered | Entered the test online: started it, or used their printed code. This is not proof of physical attendance |
| Sat | Finished the test. Kept even if the underlying test session is later removed |
| To mark | Finished and waiting for examiner review |
| Released | Finished and released to the candidate |
| Average | Released scores only. Shown only when every result in the set is on one score scale; a day mixing IELTS bands and another scale shows no average |
Give several sittings the same Campaign name and they are reported together — the right shape for one drive spread over many groups, sites or days.
A large upload does not run while you wait. The event page shows a progress bar and you can leave and come back — invitations go out steadily in the background rather than all at once.
If some rows fail — a malformed email, a duplicate — the import tells you exactly which lines and why. The rest still go through.
5. Let them sit
Invited by email: the candidate gets an account and a personal link. They only ever see their own test.
Given a code: the candidate goes to /language-tests/start/, types the code, and
begins.
Either way it is one timed attempt with AI help switched off. If the timer runs out, the attempt is submitted automatically and the result is still produced — a candidate is never locked out of their own result.
After the sitting
The event page is the control room:
- Roster — who has been invited, started, finished, or is awaiting review
- Resend an invitation, or reissue a lost access code
- Cancel a candidate who is not going to sit — cancelling before they take the test means you are not charged for it
- Export CSV of the whole roster, or of whatever is filtered on screen — includes reference, group, venue, slot and every extra column you imported. The cross-organisation registry export carries the same extra columns
- By group panel: expected, arrived, not arrived, sat, to mark, released and average per group, with roster / CSV / seat-sheet links per group
- Certificates ZIP — every certificate for the event in one download