A platform enters a school to simplify assignments, personalise exercises or communicate with families. Within months it organises homework, marks, messages and part of students’ attention. Asking how it works can then feel late or suspicious. Yet the most serious cost is not always printed on the invoice. It emerges when nobody can explain which decision the system makes, which data it needs, which students it disadvantages or how the school can step back.
Looking under the surface does not require every parent or teacher to learn programming. It requires something more ordinary: identifying the educational problem, asking for proportionate evidence and knowing who answers when a promise fails. A tool may be useful while also imposing invisible work, gathering unnecessary information or turning a teaching difficulty into an automated score. The choice is not between embracing and rejecting technology. It is about preventing its presence from displacing professional judgement and democratic conversation.
Start with the problem, not the product#
Before a purchase, a school should be able to complete a simple sentence: we need this tool because students face this specific problem, and we will know it helps when we observe this change. If the answer is to innovate, modernise or prepare for the future, there is no diagnosis yet. Those words describe an aspiration, not an assessable need. A school might need to reduce marking time, provide accessible support or preserve continuity during absence; each aim requires different evidence.
Before a purchase, a school should be able to complete a simple sentence: we need this tool because students face this specific problem, and we will know it helps when we observe this change.
Families can ask which alternatives were compared, including reorganising existing resources without adding a platform. It is also worth asking who defined the problem: leaders, teachers, students or the vendor. When the description comes from a sales catalogue, the solution tends to look inevitable. When it comes from the classroom, important detail appears: age, subject, language barriers, disability, connectivity and teacher workload. This initial conversation can prevent years of payments for a polished answer to a question nobody had.
The real map behind a small decision#
Clicking accept may activate many organisations: the school, an education authority, the main vendor, hosting, analytics, video and support services. Each may process different information under different terms. Recognising the visible logo is therefore not enough. The school should have an understandable map showing which information enters, where it travels, how long it is retained and who can gain access.
The map must include obvious data, such as names and email addresses, as well as derived information: connection time, response speed, mistakes, navigation paths or predictions about attainment. An inference can affect a student even when it is wrong. If a platform labels a girl as inconsistent, teachers may offer simpler work and eventually confirm the label. Asking whether profiles, automated recommendations or metric-based decisions exist is not technical curiosity. It protects every learner’s ability to change, develop and surprise us.
The educational evidence that matters#
Many demonstrations display features rather than outcomes. The ability to generate exercises in seconds does not prove that an application improves understanding. To find out, the school needs a baseline, a reasonable trial period and several forms of evidence: student work, teacher observation, independence, participation and learning that transfers beyond the screen. Login counts and minutes of use describe activity, not progress.
The school should agree in advance which result would justify continuing and which signal would require a change. After one term, for example, the class might explain a concept more clearly, make fewer meaningful errors or need less support with planning. Staff should also examine who improves and who does not. An average may conceal that a system helps students who were already thriving while burdening those who share a device, use a screen reader or need more time. Examining actual groups and experiences turns innovation into a testable, revisable hypothesis.
Invisible labour and total cost#
A licence price is rarely the full cost. Schools must add setup, training, account creation, troubleshooting, adaptation of materials, family communication, replacement equipment and the time required to export information. If teachers spend ten minutes each day correcting synchronisation failures, that burden belongs in the evaluation. So do the fatigue of repeatedly learning new interfaces and the attention diverted from planning a good lesson.
During a pilot, the school can record every new task and who performs it. It should distinguish initial effort from permanent maintenance. A cheap tool may be expensive if it requires recent devices or constant assistance. A costly tool may be reasonable when it removes a demonstrated barrier without creating dependency. The calculation must include what happens after a grant ends or a price rises. If the community could not sustain the platform from its ordinary budget, that fragility exists from day one rather than appearing at renewal.
Privacy people can understand#
Privacy should not be reduced to handing families a long notice. A school must be able to explain in ordinary language which data is necessary, why it is used, what legal basis permits the processing, when it is deleted and how people can exercise their rights. Spain’s data protection authority provides dedicated material through its Education and minors area. The responsibility for choosing and configuring a suitable tool does not pass to a child simply because they click accept.
A useful test is to imagine the information outside its original setting. Would it be acceptable for a company to retain a learner’s reading difficulties, behaviour or conversations for years? Can the school disable analytics, advertising, facial recognition or model-training functions? The most protective setting should be the default, not a reward for families who understand complicated menus. Refusing an optional use should not exclude a child from an essential activity or force them to reveal their choice publicly.
Equity: identify who pays for convenience#
A convenient solution for the majority may transfer its cost to a smaller group of families. Requiring a phone for identity checks, printing at home, keeping a camera on or answering messages at night creates obligations not everyone can meet. Evaluation should test the tool on a slow connection and small screen, with assistive technology and with different levels of digital confidence. There must also be an equivalent alternative that does not punish people with extra bureaucracy because they lack the expected equipment.
Students should participate safely. A satisfaction survey is insufficient because an entertaining interface may be popular while still distracting or exposing data. Better questions include: do you understand why it recommends this task; can you correct a mistake; do you feel safe asking for help; does the score represent you; and can you work without classmates seeing private information? Their answers are not decorative anecdotes. They reveal effects that aggregated statistics miss and help the school design support before wider deployment.
Human oversight with genuine power#
Saying that a human is always involved means little if that person does not understand the system’s limits, lacks time to review it or fears departing from its recommendation. Genuine oversight requires access to relevant information, training, authority to correct and a route for recording disagreements. No automated score should drift into becoming a mark, sanction, pathway decision or diagnosis.
High-impact decisions need a firm rule: a system may provide a signal, but an accountable person checks context and evidence before acting. Students and families should know when automation contributed, how to request an explanation and where to challenge an error. The European digital competence framework for educators, DigCompEdu, shows that professional capability includes selecting, managing and evaluating digital resources. It is much more than learning which buttons to press.
Contracts, exit and institutional memory#
A school keeps its freedom to decide when it can leave. Before adoption, it must know the export format, deletion timetable, migration support and which functions will stop when the contract ends. Student work and marks must be recoverable in usable formats rather than locked in an unreadable archive. Contracts should also prevent silent, unilateral changes that add new purposes or subcontractors without sufficient warning.
An exit plan is not pessimism; it limits dependency and strengthens negotiation. The school can test it during the pilot by exporting one class, deleting test accounts and checking what remains. It should preserve a short record of the version assessed, incidents, adjustments and final decision. When leaders change or the vendor announces a new feature, the community will not have to start again. Institutional memory prevents repeated mistakes and also identifies the tools that genuinely delivered value.
A ten-question method for families#
Families can bring ten questions to a meeting: what problem does it solve; what evidence is expected; which data does it collect; which inferences does it create; who has access; what does it cost in money and time; what alternative exists; how is an error corrected; when will it be reviewed; and how can the school leave? Not every question needs an instant answer, but each needs a named person and a date. If reasonable questions produce defensiveness, the family is not the problem; governance is missing.
UNESCO’s Global Education Monitoring report on technology and education, available at Technology in education, proposes judging technology through relevance, equity, scalability and sustainability. Those principles offer a useful compass for organising the questions. The European Commission’s Digital Education Action Plan also provides a public framework for high-quality, inclusive and accessible digital education.
What a school can do tomorrow#
The first step does not require a grand audit. The school can choose one familiar platform and bring together a leader, two teachers, students, families and the data-protection lead for an hour. On one page they record purpose, data, costs, barriers, evidence, accountable person and review date. Unknown facts are marked as questions rather than filled with assumptions. The group then requests missing information from the vendor and compares it with classroom experience.
Thirty or sixty days later, the group examines three specific cases: someone the tool helped, someone for whom it made little difference and someone who encountered a barrier. It decides to retain, adjust, restrict or remove the tool and records its reasons. This small cycle teaches more than an abstract declaration about innovation. Looking at how a system works costs time and attention; failing to look may cost privacy, equality, money and trust. A questioning school does not obstruct the future. It learns to govern it with the people who must live there.
