Will Ellis: Which EdTech Platforms Can Schools Actually Trust? A Research-Based Assessment

May 4, 2026

Not all EdTech platforms carry the same risk profile. This article assesses specific platforms used in UK schools across mathematics, literacy, retrieval practice, data privacy, SEND and assistive technology, computing, and management. Each platform is evaluated against the available independent evidence. Note that this version updates the OECD PISA finding to reflect a more precise reading of the data: the negative correlation is strongest for leisure device use specifically, while moderate purposeful use shows a more mixed picture. The assessments are drawn from the EdTech Accountability Framework compiled by Will Ellis / Reclaim Childhood.

Evidence Ratings

● Strong ● Moderate ● Weak ● None
RatingWhat it means
StrongPeer-reviewed findings with sufficient scale and replication to inform policy, including null results.
ModerateIndependent but limited in scale or context.
WeakSome independent work but too small or methodologically limited.
NoneVendor data only, or no trial found.

Key Terms

TermDefinition
The Instrument TestA tool does nothing until the child acts on it (e.g. a word processor remains empty until the child types). An agent responds, adapts or produces output on the child’s behalf. The instrument test asks which is directing whom: if the child directs the tool, it may belong in the classroom; if the tool directs the child, it does not.
Architectural GroundsA platform can be restricted based on its design, regardless of trial outcomes. Where evidence also shows harm, both grounds apply.
Engagement is Not LearningClicking, responding and staying on task is behavioural engagement. Struggling with a concept, making connections and building knowledge is cognitive engagement. Platforms built to maximise the first often undermine the second. A child completing 200 Times Tables Rock Stars questions in a session may be behaviourally engaged throughout and cognitively engaged for almost none of it.

Table 2: Platform Assessment

Platform / Tool Key Finding Evidence Source
Mathematics
Dr Frost Maths No independent RCT. Teacher assigns specific tasks; auto-marking returns results. No streaks or leaderboards. Not-for-profit. NoneVendor-funded DrFrost.org
Kerboodle No independent RCT. Digital textbook and resource platform. Requires a signed DPA from Hodder Education. NoneVendor-funded Kerboodle
Khan Academy (no Khanmigo, teacher-assigned tasks) Mixed results from several RCTs. Evidence mostly from US and developing-world contexts. WeakIndependent Pane et al., RAND (2015)
Mathletics No independent RCT. Vendor data only. NoneVendor-funded Mathletics
Mathway Improves performance on specific problem types but risks answer-without-understanding. ModerateIndependent Ghanem and Faqihi (2022)
MathsWatch No independent RCT. Teacher selects videos and worked examples. No adaptive algorithm. NoneVendor-funded MathsWatch
MyMaths No independent RCT. The one vendor study looked at teacher time-saving, not pupil attainment. NoneVendor-funded MyMaths
Sparx Maths Active use for about 1 hour per week shows modest GCSE-grade gains, but no independent RCT. Access alone showed no effect. ModerateVendor-funded RAND Europe / Cambridge (2021)
Literacy and Reading
Accelerated Reader Small short-term gains in some trials, but the large EEF primary RCT found no extra progress over other reading approaches. ModerateIndependent EEF trials (2014 and 2021)
Jolly Learning / Jolly Phonics Strong independent evidence from longitudinal studies. Jolly Phonics gains maintained seven years post-intervention in the Clackmannanshire study. Directly informed the Rose Review. The grammar-focused elements of the wider suite have a weaker independent evidence base. StrongIndependent Rose Review (2006); EEF KS1 (2021)
Lexia Reading / Core5 One moderate-sized EEF RCT shows small gains overall and slightly larger gains for FSM pupils. StrongIndependent EEF RCT (2021)
Retrieval Practice and Revision
Kahoot! Good at boosting short-term recall across 93 studies, but limited evidence for long-term retention. Novelty effect documented. Whole-class, teacher-directed use only. In practice most teachers use it as a reward activity at the end of a lesson, which is probably where it does least harm. ModerateIndependent Wang and Tahir (2020)
Pear Deck No independent RCT. Teacher-led presentation tool; pupils respond to teacher-set prompts. No adaptive algorithm. AI Copilot must be disabled. NoneVendor-funded Pear Deck
Quizlet Some evidence of better vocabulary learning, but effect sizes are small to moderate and depend heavily on how it is used. Paper flashcards outperform in some studies. ModerateIndependent Ozdemir and Seckin (2024)
Seneca Learning (Amelia disabled) Vendor study claims big score gains over revision guides, but the comparison group used passive revision guides rather than teacher instruction. No independent trial. WeakVendor-funded Seneca
Data Privacy
Google Classroom Published research raises serious concerns about data flows to Google even under education accounts. Requires a signed DPA before deployment. StrongIndependent Livingstone et al. (2024)
SEND and Assistive Technology
AAC devices (non-verbal / minimally verbal) Strong evidence that AAC improves communication for non-verbal or minimally verbal pupils and may support rather than suppress speech development. Professional prescription required. StrongIndependent Iacono, Trembath and Erickson (2016)
Microsoft Immersive Reader (SEND context) RCT in 20 UK primaries found no significant reading gains. High withdrawal rate; schools struggled to implement as intended. StrongIndependent – null result RAND Europe (2020)
Speech-to-text (for dysgraphia) UK study shows better written output for pupils with writing difficulties, but the sample is small: 30 children across 3 settings. ModerateIndependent Kambouri, Simon and Brooks (2023)
Text-to-speech (general) Meta-analysis shows moderate improvement in reading comprehension for pupils with reading difficulties, but results are mixed for fluency. StrongIndependent Wood et al. (2017)
Computing and Creative
SketchBook No evidence base as a learning tool. Child directs the tool entirely. No adaptive algorithm or engagement optimisation. None (utility)Vendor-funded SketchBook
Minecraft Education Edition (teacher-structured) Review of 29 studies shows some benefit for spatial thinking and creativity, but all studies had medium or high risk of bias. Microsoft-funded. ModerateVendor-funded Slattery et al. (2025)
Purple Mash (AI features disabled) Vendor survey of 6,000 teachers and students shows positive perceptions of engagement. No independent RCT on learning outcomes. NoneVendor-funded 2Simple (2022)
Management and Administration
Apple Classroom Staff-facing management tool. No pupil-facing engagement design. None (utility)Independent Apple Education
Class Charts (staff use only) Staff-only use for attendance and behaviour recording is reasonable. Pupil-facing version introduces social comparison through points and merit badges. Owned by TES Global: schools should check what data flows exist across TES’s wider commercial operation. None (utility)Independent Class Charts
Tapestry UK servers, GDPR compliant, data owned by school. No adaptive algorithm or engagement design. None (utility)Independent Tapestry
Context and Background
OECD PISA: general EdTech finding Large-scale PISA data show a consistent negative correlation between higher leisure device use in classrooms and attainment in reading, maths and science. The negative correlation is strongest for leisure use specifically. The overall picture on learning-oriented device use is more mixed, with moderate purposeful use associated with small positive outcomes in some systems. This is a cross-country pattern, not a verdict on any specific tool. StrongIndependent OECD PISA data
US EdTech policy reversal US schools are rolling back device-heavy policies amid declining test scores. Journalism; no causal proof. Commentary Fortune (April 2026)

Source: Will Ellis / Reclaim Childhood – reclaimchildhoodmedia.substack.com

Disclaimer: We’ve created this overview to help busy parents quickly grasp the key findings. It should not be considered a substitute for reading the original study. For accuracy and complete context, please consult the source document.

You may also like

Follow Us

Youtube Instagram Facebook