ONISEP · Avenir(s) national programme
Upper-secondary module — building a path, not just browsing
Rethinking the guidance journey for upper-secondary students, and giving form teachers a tracking view that finally replaces the spreadsheet they kept on the side.
01
Context & problem
The upper-secondary module is the part of Avenir(s) aimed at students aged 15 to 18. This is the moment guidance stops being exploration and becomes a decision: tracks, specialisms, applications, deadlines. It is also the moment the gap widens between students who are supported and those who are not.
The existing module did what it was asked to do — deliver information — but not what students needed: build, return, adjust.
The problem, stated plainly
Students were browsing. They were not building. And the adults responsible for supporting them had no way to see who was falling behind, other than noting it themselves in a spreadsheet.
Two journeys to reconcile
On the student side: a long guidance assessment, abandoned midway, with no motivating output. On the teacher side: no overview of the class, so students were spotted late — often after the end-of-term review, by which point it is too late to act.
02
Research & insights
Research focused on the gap between what the product intended and how it was actually used, in two complementary strands.
Qualitative — understanding drop-off
Semi-structured interviews with students across all three upper-secondary years, form teachers and school psychologists, in three regional education authorities. A five-part guide: school history, how they picture guidance, use of existing tools, turning points, expectations. Verbatims coded through affinity mapping.
Six contextual-inquiry observations, in classrooms during dedicated guidance hours, revealed something interviews did not capture: students work in very short, frequently interrupted sessions, in a noisy environment, on hardware they did not choose.
“We get an hour. By the time everyone has logged in, twenty minutes are left.”
Quantitative — sizing the problem
The survey, stratified by authority and academic track (1,200 responses, ±2.8% margin of error at 95% confidence), confirmed that the observed frictions were not isolated cases.
Drop-off before the third screen of the guidance assessment
Sessions per student per term — occasional use, not sustained tracking
Of form teachers kept a parallel spreadsheet
Three structuring insights
- Effort has to be paid back immediately. A student will answer questions if they get something usable within the minute.
- Sessions are short and fragmented. Any journey that assumes twenty uninterrupted minutes is a journey that will be abandoned.
- Teacher tracking is a product feature, not a report. As long as it does not exist in the tool, it exists in a spreadsheet.
03
Exploration & solution
Break the assessment into self-contained sequences
The monolithic assessment was re-cut into short, independent sequences, each producing immediately legible output. A student can stop at the end of any sequence without losing the benefit of what they have done — which makes the journey compatible with the twenty real minutes observed in class.
Make resumption obvious
A persistent progress state, visible on arrival, shows what is done, what remains and where to pick up. Students no longer have to mentally reconstruct the previous session.
Design the class view
The teacher dashboard was designed from the spreadsheet form teachers already kept: which columns, which sort order, which alert signal. Rather than inventing a view, we formalised the one they had built themselves, then added the single element they could not produce by hand — each student's real progress through the assessment.
No student score, no ranking. The dashboard signals a level of progress and a need for support, never a performance. A tracking interface in a school context carries a responsibility: it must not turn a support tool into a judgement tool.
Three test waves
Each iteration was tested with all three audiences, measuring SUS and completion at every wave. Adjustments covered sequence length, the wording of outputs, and the density of the class view.
04
UI & design system
The module is built entirely on the DSFR and on the product patterns defined at Avenir(s) platform level — the same interface vocabulary, the same components, the same accessibility grammar.
Contributions to the system
- Sequence card: the base unit of the student journey — state, estimated duration, associated output, resumption.
- Progress bar component: a persistent element carrying continuity between sessions.
- Class view: a dense table designed for fast reading — strict typographic hierarchy, attention signals never carried by colour alone, keyboard-accessible sorting and filtering.
Accessibility constraints held in design
- No information carried by colour alone: every attention signal is doubled with a label and a shape.
- Properly structured data tables (associated headers, declared scope) so they stay readable to screen readers.
- The full journey completable by keyboard, with tab order specified in the mockup.
- Contrast verified at source across all states, including disabled and focus.
The module's screens are among the seven audited screens meeting 100% of applicable RGAA criteria.
05
Impact & learnings
Completion of the student guidance assessment (+41 pts)
Time needed by teachers to spot struggling students
SUS score across the Avenir(s) scope, over 3 iterations
What I took away
Context of use beats the ideal journey. No mockup survives a classroom with twenty minutes left after login. The six contextual observations shaped the design more than the fifteen interviews did.
A parallel spreadsheet is a free specification. It tells you exactly which columns matter, in what order, and which signal the user looks for first.
Designing for a captive audience raises the bar. A student cannot choose another tool. That imposes an extra duty: never make the interface deliver a judgement the user did not ask for.
What I would do differently. I would involve school psychologists from the first test wave rather than the second: their need for depth surfaced edge cases on the individual view later than it should have.