This screen needs a bigger window
It's a wide grid with a lot of small controls — on a phone it would be fiddly enough to make a mistake on somebody's paper. Open it on a computer and it'll be here.
The rest of the app works fine on this phone: your tests, your students, who's sitting what, and what's waiting to be marked.
Tests
Nothing matches these filters.
Test
Each paper in this test has its own timer. Students sit them in any order, any day inside the window.
Combined score across every paper in this test, and where each student ranks among others who've fully finished and been marked. Preview run-throughs and in-progress attempts are never ranked. Per-question breakdowns live on each paper's own Results button, under Papers.
Each student gets their own link, bound to their name. Send it the same way you send test invites today. A link can be reused to resume, but never mints a second attempt.
Who marks the writing
Open link
Marking
Student
Your mark
Settings
Basics
How you file it. It groups the Tests list and nothing else, and students never see it. Whether this test is a week of a Horizon programme is the tick below, not this.
The shelf this test sits under on a student's tests page. Students do see this one, so put nothing here they shouldn't know before opening the paper. Name a programme or subject, never a writing genre.
This choice decides what a student gets back. A practice paper is one that is not an assessment and not a week of Horizon: it shows them their score and every question with its answer options, and they can sit it again whenever they like. A paper that counts shows them their score and nothing else, because the whole cohort sits the same paper and a student who has the questions can pass them on. There is no separate setting for it further down, so the two can never disagree.
The cohort a student is ranked against. Sets sharing this name are the weeks of one programme, in the order you arrange them in Horizon. The category above only files the paper, and renaming it moves nothing.
When students can sit it
Leave both blank to keep it open indefinitely.
Question order
Mostly for retakes. Sat a second time in the same order, a paper starts teaching “the third one is C” rather than the content, and the retake stops telling you anything. It also means one student can’t pass another the answers in order.
The other half of a retake. A student who has sat a paper twice often remembers “the answer was C” without remembering the question at all, and the question order alone does nothing about that. Multiple choice only; gap-fills and matching blocks carry their own answer structures and are left alone.
A reading text and its questions always move together. A group is never split, so a student can never get a poem question, then a cloze, then another poem question. A question with no reading text is its own part and moves on its own, which is how a real paper reads: extract and its questions, a cloze, another extract and its questions, another cloze. A group can be told it is safe to shuffle inside itself, on the group in the questions list; that is off unless you say so, because some papers build (“read the first stanza”, then “now read on”) and shuffling those makes a broken paper, not a harder one. A paper of nothing but independent questions is one run, so all of them scatter, which is right for maths.
Nothing else changes: every report, export, the printed paper, the answer key and the cohort figures stay in the order you wrote, questions and options both, so “question 12” and “option C” mean the same thing to you, to Horizon and to a comparison as they always did. Where you look at one student’s paper, it shows both. Preview as student is never shuffled, so you can still check the paper against the original.
Honest Exam Mode
Four things are recorded during a live paper:
- leaving the page — a tab or window switch
- trying to copy, including dragging text out
- trying to paste anything in
- pressing a screenshot key
All four share the one warning limit. Students are told before they start, warned each time, and you see the breakdown per question in Results. Copying and pasting are always blocked outright; screenshots can only be detected, never prevented.
Counts page-leaves, copy attempts, paste attempts and screenshot attempts together. Only enforced in the third mode, where the paper ends at the limit. In “Warn & record” nothing stops. 3 is a good default: real notifications and accidental clicks do happen, and the student gets a clearly-marked last warning before the final one.
What your student sees, in order
Everything below is their side of the test, top to bottom.
1 Their link
The first thing they open. It names the test and who it's registered to, so a forwarded link is caught before anyone sits it. Nothing here to set. It's built from the roster.
2 How the test works
3 The page they see before they start
Bring this page across from another test. It is about the test as a whole, how the session works and what to have ready, and stays separate from each paper’s own instructions page.
Fills in the time and question count for you. Replace anything in [square brackets]. The test can't be activated while they're there.
Paper name
The timer starts as soon as you choose Begin, and it can't be paused.
4 The paper itself
The questions, in the exam chrome. The only way to judge it is to sit it. This mints a preview attempt that never touches your results, and never counts against a student's link.
5 After they finish the last paper
The end of the whole test, shown once every paper in it has been handed in. Finishing a single paper with others still to come is a different screen, written on that paper’s own settings page.
Horizon Plus students see exactly the same screen as everyone else. The upgrade is not something this page controls: it is the report and the copies of each paper, sent by email after the week is confirmed. There is no Plus setting to find, and nothing here needs changing for them.
A written response has no mark until a person gives it one, so there is nothing for the setting above to reveal. Until it is marked and released the student sees “with a tutor” on that paper — never a zero, and never a total that counts it as one.
- On the day — no score for that paper, and no partial total.
- Once marked and released — the mark appears and the overall total is worked out for the first time.
- Marks are entered by hand in the marking queue. Nothing is generated and nothing is estimated.
Their link reopens an unfinished paper so they can carry on, but it never starts a second attempt. That is what makes a score comparable across the cohort, so it holds for every set that counts.
Answers save automatically the whole way through. There is no separate “save” step for the student, and no way to lose progress by closing the page. Exporting a PDF or CSV report is a tutor-only tool, from Results: students never see it and cannot generate one themselves.
What the test says once everything is in. Leave it empty and they just see “All done”.
Bring the wording across from another test. Ten weeks of Horizon say much the same thing.
All done, Emma. That's the test finished.
Paper settings
Basics
What the student sees on the list of papers, and the heading on every report.
This paper has been sent out, so its time limit and instructions are fixed. A student starting now must get the same paper as one who started earlier. The name can still be corrected. To change the rest, take it back to Setup if nobody has sat it, or duplicate it and change the copy.
Where this paper sits
The window, the link, Honest Exam Mode and whether they see a score are decided once for the whole test. They are the same for every paper in it.
Before they begin
The paper-specific bit: how many questions, how many answers each needs, whether calculators are allowed. How to flag, navigate and use the timer is explained automatically on the pages before this one, so it does not need repeating.
After they finish this paper
Shown the moment this paper is handed in, with the other papers still listed underneath it, for example "Thanks for completing the Reading paper." Leave it empty and they simply see the test marked as finished and get on with the next one.
This is not the end of the session. When the last paper goes in they get a different screen, with the test's own closing message and the way out. That one is written in the test's settings.
Copies the wording here so you can change the paper's name in it, rather than retyping the same paragraph for all forty papers.
New reading text
Questions on this text
Questions
Add ▾
More ▾
Consultations
Papers chosen for one family, handed out in one act. Open a pack to see the screen you share on the call.
Consultation
Consultation
Results
Item analysis
How the cohort did on each question, across everyone who's submitted this paper.
Question library
Live now
Sittings in progress, refreshing every 15 seconds. This shows the last time each student's work reached the server. It can't tell you whether they're at their desk. A closed laptop still counts as in progress until its timer runs out, because the clock runs from the moment they pressed Begin. Nothing here changes a sitting; to grant extra time, open that test's Results.
Free sittings
Everyone who came through an open door. They are their own group: never on a roster, never in a cohort, never in a band. This is the only screen that finds one by first name — which is all a parent has when they email.
Writing rubrics
How a written response is marked, written down once and pointed at by every paper marked that way. Selective writing is two sets judged separately, so a piece carries a band on each rather than one band overall. A practice piece that is not selective needs no rubric at all: leave it without one and it is marked out of its own marks, exactly as before.
Staff
Everyone who can sign in. There is no signup and no reset email. You create accounts here and tell each person their password; they can change it themselves afterwards. Switching someone off keeps their name on the work they did and ends any session they still have open.
Import log
Every paper imported, and everything the importer said about it at the time. Nothing here is cleared. It stays even after a question has been fixed by hand, so a format that keeps causing trouble shows up as a pattern rather than being forgotten one import at a time.
Horizon
A week is a test that says it belongs to a programme. Open one, go to Settings, and tick Part of a Horizon programme, then give the programme its name. Tests naming the same programme gather here as its weeks. A test that does not name one stays out, because a report should only ever compare a student against people who sat the same thing.
Programme
What families see about this programme
Shown on every student's own page, under their week markers. Say anything the weeks cannot — which weeks are which phase, how long a sitting runs. Leave it empty and nothing is shown.
Across the programme
Skills on this exam
Week
One row per student who sat this week. A row with no report says which of the two things is missing: marking you can clear now, or a cohort that only more sittings will fix.
Plus reports
Marks from outside this platform
Students who sat the same paper elsewhere belong in the same cohort. Without them a rank is out of seven rather than out of a real field. Paste the sheet or drop the file in; you say which column is which, so a sheet that gets rearranged next term costs a re-map rather than a bug report.
- A name is matched against the roster once. An exact match after tidying spaces and case is silent; anything close is reported and never merged on its own. A wrong merge welds two children's histories together and is close to unfindable afterwards.
- An import is a snapshot, not a link. Re-importing a corrected sheet updates the same rows rather than adding a second copy, so fixing a cell on Tuesday cannot change what Monday's report said.
- These marks are never treated as answers. They join rank, percentile, band and the distribution, and they stay out of item analysis, the marking queue and every per-question figure. Nobody outside answered a question here.
- How the outside sitting was marked is not recorded. The total it was marked out of is kept, so a 28 of 30 and a 28 of 35 never land in one distribution, but if another centre marks writing to a different rubric, nothing on this screen will tell you.
Generate questions
Build a paper of spatial or abstract reasoning questions. Every one is checked before it is written, the numbers behind each check are on screen, and the whole paper can be exported for editing by hand.
Students
Student
More ▾
What they see right now
Their portal, exactly as it renders for them.
How to reach them
Every sitting
No submitted tests yet.
Results
Every completed sitting across all tests, newest first. Click a row to open the full review.
No completed sittings yet.
Review
Question editor
More ▾
This question is written on a reading text, so it can’t be banked on its own. The extract wouldn’t travel with it. Bank the whole reading set instead, from the reading text’s own Save to bank button.
Saved to bank.
Set automatically: 1 mark per gap/statement, like the real test.
What this question tests. Groups the marks-by-skill breakdown; untagged marks appear under no skill. Never shown to students.
Use the [gap] toolbar button to insert numbered gaps into the text above, then set each gap's choices here.
One answer per line. Add extra lines as distractors. Students see them all, shuffled together.
Defaults to the passage's tab labels. One per line.
You mark each response yourself from the Results view. Nothing is auto-scored.
Students can only type. Bold and underline are the only formatting available either way, and images can never be inserted. Anything else is stripped on the server too.