Student voice that survives the term.

The problems education teams actually bring

Everyone is surveyed at once, and nobody finishes

Module evaluation lands in the same fortnight as the student experience survey, the parent questionnaire and the staff wellbeing check. The response you get is the response someone could complete in the four minutes between lessons, on a phone, on the school wifi.

Conditional logic is the first lever: rules match any or all of a set of conditions, so a parent with one child in Year 7 never sees the sixth-form block, and a student who did not take the optional module is not asked to rate it. Save and continue is the second: a resume link keyed on a 24-character token, which is what makes a long course-review questionnaire survivable across two sittings. Partial answers are stored in a table with no row-level-security policies, reachable only through SECURITY DEFINER functions.

The third lever is layout. Of the three available — classic (one scrolling page), conversational (one question at a time, one-tap questions auto-advancing) and paged (a section per screen) — conversational is the one that reads best on a phone handed out in a tutor group, and paged is the one that suits a long evaluation broken into modules.

The questions are mostly grids, and grids export badly

Course evaluation is a matrix by nature: eight statements, one agreement scale. A Matrix question puts the rows on one shared scale, and the CSV export produces one column per row rather than a single concatenated cell that somebody has to unpick in a spreadsheet before analysis can begin.

Alongside it, the palette covers 16 question types: linear scales with end labels for confidence questions, a rating question for a quick star response, a Recommend score from 0 to 10 for the question everyone calls NPS, ranking for prioritising options, and date, address and number types for the administrative parts of an application form.

Some forms need a document attached

A bursary application needs evidence. A student submission needs a PDF. A trip consent form needs a photograph of something. File uploads go to a private bucket, not a public URL, and are read only by members of the organisation through signed URLs valid for 300 seconds. If you clear test responses before going live, the uploads and partial responses go with them, so a pilot run does not leave orphaned files behind.

Two honest limits belong here. Signatures are simple electronic signatures, not qualified ones, so a form suits a consent declaration rather than anything demanding a qualified signature. And there is no payment collection, so a trip form cannot take the money — it can only collect the information. A parent questionnaire for a bilingual community is handled by multi-language surveys, which keep both versions in one survey and one set of results.

Accessibility is a statutory duty, not a preference

Colleges, universities and maintained schools are public sector bodies under the accessibility regulations, and a survey embedded in an institutional site inherits that duty. WCAG 2.2 AA is the standard the respondent experience is designed against; it is not an audited conformance claim, and no third party has assessed the product. The accessibility page sets out both halves of that, including which interactions have a pointer-free route. The accessibility statement is the version to attach to a procurement return.

Trusts and federations are several organisations wearing one coat

A multi-academy trust usually wants central oversight and school-level separation at the same time. Today that is done with organisations: multiple organisations are supported, roles are owner, admin and member, the last owner is protected from removal or demotion at database level, and organisation isolation is pinned server-side rather than trusting a value sent by the browser. There is no finer-grained permission model inside an organisation, and there is no SSO or SAML, so accounts are created and managed individually.

Where a trust runs a survey jointly with a local authority, a diocese or a university partner, collaborator logos are set per survey rather than per account, so one workspace can badge a different partner on each survey without disturbing the trust's own brand.

What an education survey typically looks like

A college running an end-of-module evaluation would usually build something close to this. Every element exists in the product today.

1. Cohort routing. A hidden field carrying the module code, declared per survey and read from the query string, so the link in each module's virtual learning environment tags its own responses without asking the student to pick from a list of 200.
2. A screening choice. A Multiple choice question on attendance mode, using bento option tiles — the default for choice questions — which drives the branch that follows.
3. The core grid. A Matrix question rating eight teaching statements on one agreement scale.
4. Confidence. A Linear scale with labelled ends, asked before and after the module content questions, with piping so the second question repeats what they said the first time.
5. Priorities. A Ranking question over the things they would change, with answer order randomised — seeded per page load, so going back does not reshuffle — and a "nothing to change" option marked fixed so it stays put.
6. Open comment. A Paragraph question with a maximum length and a live character counter.
7. Anonymity by omission. No name and no email question, which is the only reliable way to make a survey anonymous: a survey collects exactly what you ask for and nothing else.

Turning it into something a board will read

Results show live charts per question with response counts and completion. From there: CSV export for the analyst, with a separate column for "Other (please specify)" free text; individual responses browsable one at a time when a comment needs following up; and a print-to-PDF of any single response for a case file. There is no AI summary of free-text answers — that is planned rather than built, and it is not in the product today.

Scoring and bands are worth knowing about even outside assessment. Give choices a score and questions a weight, and a self-review — digital confidence, revision readiness, safeguarding knowledge — shows the respondent a percentage band the moment they finish. Only questions the respondent actually saw are scored, so branching never costs anyone marks.

Related

HR and people surveys covers the staff side of the same institution, local government shares the accessibility and data-residency ground, and market research goes deeper on randomisation and matrix design. Back to all solutions, or see the product overview.