Detect Learning Gaps Without Turning Assessment Into Surveillance

Learning gap assessment does not need to mean watching every click, recording every minute, or treating students as potential misconduct risks. A better starting point is simpler: gather small, useful pieces of evidence about what learners understand, identify the next teaching move, and discard anything that does not help that decision.

That approach keeps assessment close to its educational purpose. The Education Endowment Foundation describes formative assessment as part of the foundation for effective feedback: it helps teachers establish learning intentions and assess the gaps that feedback should address. Its feedback guidance also cautions that feedback is not automatically beneficial and carries a workload cost, so the goal is not more data or more marking. It is better evidence, used well. ([educationendowmentfoundation.org.uk](https://educationendowmentfoundation.org.uk/education-evidence/guidance-reports/feedback))

Start with a narrow definition of the gap

A learning gap is not a label for a student, a prediction about their potential, or a permanent record of every error. It is the difference between the specific knowledge, skill, or strategy a learner currently demonstrates and the intended learning for the next step.

That definition matters because it changes the question from “What can we monitor?” to “What do we need to know in order to teach well?” For example, after a lesson on persuasive writing, the useful questions may be whether students can identify a claim, select relevant evidence, and explain how that evidence supports a point. Webcam attention signals, background-noise detection, typing speed, or time spent with a tab open do not reliably answer those curriculum questions.

Make every assessment signal earn its place. Before collecting it, write down:

• the learning intention being checked;
• the evidence students will produce;
• the decision that evidence will inform;
• who will see it;
• how long it needs to be retained.

If the team cannot name a teaching or support decision, the information is probably unnecessary for learning gap assessment.

Use frequent, low-stakes evidence

Low-stakes does not mean low-value. It means the activity is designed primarily to inform teaching and learning rather than to rank, punish, or create a high-consequence record. The evidence can be gathered in a classroom, tutoring session, live online lesson, or asynchronous course without making students feel continuously watched.

Useful options include:

• One-question hinge checks: Ask a carefully chosen question where each plausible answer reveals a different misconception. Use the response to decide whether to move on, reteach, or provide a contrasting example.
• Exit prompts: Invite students to complete a sentence such as “The step I can do independently is…” or “One part I would like clarified is…”.
• Worked-example annotation: Give students a completed solution, paragraph, or model response and ask them to identify the purpose of each step.
• Short retrieval checks: Use a small number of questions from recently taught material, then review patterns across the group rather than overinterpreting one response.
• Error analysis: Present an anonymous incorrect answer and ask learners to diagnose and repair it. This reveals reasoning while normalising mistakes as material for learning.
• Brief conferences: Ask a student to explain how they approached a task, what they found difficult, and what they would try next.
• Self-assessment against success criteria: Ask learners to point to evidence in their own work, rather than simply rate their confidence.

These methods create evidence that is legible to teachers and students. They also support feedback that is tied to the task, the subject matter, and learners’ self-regulation strategies—areas highlighted in the EEF’s feedback evidence summary. ([educationendowmentfoundation.org.uk](https://educationendowmentfoundation.org.uk/education-evidence/teaching-learning-toolkit/feedback))

Triangulate before deciding that a learner has a gap

A single missed question can reflect a misconception, but it can also reflect ambiguity, fatigue, language demands, a rushed response, or a technical problem. Avoid turning one data point into a conclusion. Instead, triangulate modestly.

A practical pattern is to use three lenses:

1. Performance: What did the student produce in a short task?
2. Explanation: Can the student explain their thinking, choice, or method?
3. Transfer: Can the student use the same idea in a slightly different example?

When two or three pieces of evidence point to the same issue, the next step becomes clearer. A learner who can recall a definition but cannot use it in a new context may need modelling and guided practice. A learner who can explain a process orally but makes transcription errors may need a different support than a learner who cannot yet explain the process at all.

This is where teacher authorship matters. A dashboard can organise responses and highlight patterns, but it cannot fully interpret the context of a student’s work. The educational decision should remain with the teacher, tutor, or qualified member of the learning team.

Separate progress evidence from behaviour surveillance

Digital learning systems can make it easy to collect large volumes of activity data. Ease of collection is not the same as educational value. For this article, surveillance means collecting or inferring information about students’ behaviour beyond what is proportionate for the learning purpose—especially when the information is persistent, difficult to understand, or used without a clear route to support.

Instead of asking whether a system can track an activity, ask whether that activity provides valid evidence of learning. Time online, keystrokes, eye movement, location, or device signals may describe interaction with a system, but they are not direct demonstrations of understanding. Where a task needs an academic-integrity process, keep that process distinct from routine formative assessment and make its purpose, limits, and handling clear to learners.

This distinction is also a practical privacy discipline. The U.S. Department of Education notes that online educational services raise privacy and security considerations for tools used by students and families in school activities, including tools for viewing learning progression and completing homework. ([studentprivacy.ed.gov](https://studentprivacy.ed.gov/resources/protecting-student-privacy-while-using-online-educational-services-requirements-and-best)) In the UK, the Information Commissioner’s Office reported data-protection gaps in audited edtech providers, including weak application of data-minimisation and storage-limitation principles, incomplete data-flow mapping, and insufficiently detailed contracts. ([ico.org.uk](https://ico.org.uk/media2/13yfm55z/edtech-examined-key-findings-from-our-audits.pdf))

For schools and education businesses, the operational lesson is straightforward: collect the minimum information needed for the stated learning purpose, restrict access, explain the workflow in plain language, and review whether retained data is still useful. Local legal and contractual requirements will vary, so responsible teams should obtain appropriate privacy and legal review before introducing new monitoring or analytics practices.

Design the feedback loop before the assessment

Assessment becomes burdensome when evidence enters a system but no one has time or authority to act on it. Design the response loop first. For each regular check, decide what happens when a student is secure, developing, or stuck.

Keep the cycle short: teach, elicit evidence, interpret, respond, and check again. This gives students multiple opportunities to demonstrate improvement and prevents early misunderstandings from hardening into permanent categories.

Make the process visible and humane

Students are more likely to engage honestly when they understand that quick checks are there to shape teaching, not catch them out. Explain what is being assessed, why it matters, how responses will be used, and what support may follow. Where appropriate, show learners the patterns you see at class level: “Many of us can identify the technique, but fewer of us can explain its effect, so that is what we will practise next.”

Invite students into the interpretation too. A confidence rating alone is weak evidence, but a confidence rating paired with a reason can be useful: “I chose this answer because…” or “I am unsure because I confuse these two terms.” This supports metacognition without pretending that self-report replaces demonstrated learning.

UNESCO’s 2025 work on AI and education argues for a human-rights-oriented approach to digital education and emphasises that teachers should remain the primary interface of education. ([unesdoc.unesco.org](https://unesdoc.unesco.org/ark%3A/48223/pf0000395373.locale%3Den)) In day-to-day assessment design, that means using technology to reduce repetitive administrative work and surface evidence—not to replace professional judgement or student voice.

A proportionate workflow for teachers, tutors, and education businesses

1. Choose one teachable objective. Avoid broad labels such as “weak at maths” or “behind in English.”
2. Create a two-to-five-minute check. Make it closely match the objective and include likely misconceptions.
3. Review patterns, not just scores. Group responses by error type, strategy, or missing prerequisite knowledge.
4. Assign a small next step. Reteach, practise, extend, or ask one clarifying question.
5. Record only what supports follow-up. Keep notes concise, purposeful, and access-controlled.
6. Check again after instruction. Look for new evidence rather than assuming the intervention worked.

SubSchool can support this kind of workflow by helping teaching teams organise student-progress evidence and reduce repetitive follow-up work, while teachers retain authorship and the final educational decision. Explore SubSchool’s student-progress features to consider how a lighter, more purposeful evidence cycle could fit your teaching model.

The aim is not an assessment system that knows everything about every learner. It is a teaching system that knows enough, at the right moment, to offer the next helpful step.