Typing insight, privacy, and research
Plain-language transparency for schools and families: what we observe from routine homework, when we withhold conclusions, and how sensitive indicators stay behind student-first, school-governed controls.
What we observe (sources of signal)
Qwixl:Homework is designed around routine work, not separate lab tests. Signals can include:
Typing telemetry collected while a student composes in the assignment workspace (for example timing between keystrokes, pauses, edits, and deletions). This is gathered only within Qwixl, when the student is completing real assignments.
Written responses in submissions (spelling, grammar, phrasing, and structure) to complement movement-based signals.
Marking and feedback performance and rubric-aligned comments, if added by the teacher, can be read alongside process data.
Optional patterns from the personal tutor (topics revisited, types of help requested) where the school uses that feature, as an additional lens — not a replacement for teacher judgement.
No single stream is treated as definitive; we combine modalities conservatively and prioritise stability across multiple pieces of work.
Multimodal screening (not a label machine)
Our screening-style analytics are built to highlight learners who may benefit from further attention or structured follow-up within the school's usual pathways. They are not diagnoses and never appear to students as clinical labels or risk scores.
Analysis runs only when there is enough signal across the modalities your organisation uses. If evidence is thin, the product answers with "Not enough data for analysis" rather than speculative indicators.
Student-first privacy and school governance
We treat SEN-related insight as high-trust data. In line with our mission, visibility of sensitive indicators to staff roles follows school policy and explicit sharing rules (for example SENCO-mediated sharing where your configuration requires it). Families receive transparent first-login messaging about what the product does and how to exercise data rights.
Cookie and optional analytics choices are separated from core teaching features; see our Privacy Policy and Cookie Policy for detail.
Research-grounded methodology
We maintain an internal evidence register tying major modelling choices to peer-reviewed literature and recognised guidance (for example patterns described in typing and literacy research, SEND early-identification good practice, and safe use of AI in education). Methodology and citation snapshots are versioned so schools can see what generation of logic they are on.
For a longer read on undiagnosed need in UK schools, see Insights & Research. Pedagogical framing of how Qwixl supports early identification is on Our approach.