Students typically have one instrument for gauging exam readiness: how confident they feel. It’s a natural proxy and an unreliable one. The gap between subjective preparedness and actual retrieval under test conditions is wide enough that closing it requires more than a confidence check. The diagnostic data that meaningfully measures readiness is granular by necessity: not just whether a student is online, but which topics trigger errors, how performance shifts as difficulty rises, and how accuracy holds across separate sessions.
Metacognition research explains why that level of detail matters at all. When students reread notes or work through familiar exercises, the cues available during study are largely absent when the exam begins; Bjork’s research on metacognition shows these cue mismatches can inflate judgments of learning and misdirect study time as a result. Repeated studying can raise confidence even when delayed retention falls behind what active testing would have produced. The implication is pointed: both the subjective feeling of readiness and the objective results of a recent practice session can be poor predictors of what a learner will retrieve when conditions shift. Robert A. Bjork, Professor of Psychology at the University of California, Los Angeles (UCLA), whose research examines how learners assess their own competence and where those assessments diverge from later retrieval, states this directly: “current performance, measured subjectively or objectively, can be a highly unreliable basis for predicting future performance.” What examinations demand is recall. What revision typically produces is recognition — a different cognitive operation wearing familiar clothes.
The Analytics Gap — Why It Persisted
Traditional preparation tools were built to deliver content, not to map where a learner’s understanding actually breaks down. Textbooks, class notes, and past examination papers expose students to the right syllabus areas and question formats, but their feedback is coarse. A marked past paper shows which questions were wrong on that sitting — not whether errors cluster around a particular concept, recur across sessions, or emerge only once questions cross a certain difficulty threshold. The signal is single-instance and retrospective, which makes it a poor instrument for steering revision in real time.
Teacher feedback is richer, but it runs into structural limits that have nothing to do with teacher skill. In most classrooms, the ratio of students to teachers and the volume of ongoing assessment make sustained individualized diagnostic feedback — per student, per topic, repeated — hard to deliver. The OECD’s TALIS 2018 survey reports that teachers average 6.5 hours per week on planning and 4.2 hours on marking and correcting work across participating systems. That is, of course, the normal condition: granular diagnostic feedback is hardest to produce precisely when the number of learners requiring it is largest, which describes every classroom. Periodic quizzes and class-level assessments can identify where a cohort is collectively struggling, but they rarely generate the continuous, per-topic performance history that would show each student where their accuracy degrades.
Private tutoring has historically filled that gap. A tutor working repeatedly with one student can notice which topics trigger errors under pressure, which mistakes are stable versus random, and how performance shifts as difficulty rises — then adjust practice accordingly. That diagnostic function depends on time and individual attention, and it has always been more accessible to students whose families can pay for it. The result is an information gap as much as a teaching one: the diagnostic precision that shapes effective preparation has been a purchased advantage, not a structural feature of most students’ experience. Analytics represent the proposed corrective — a way to distribute that diagnostic function without requiring the same price of admission.
Signal vs. Noise — Measuring What Matters
Most platforms can already log how long a student spends on a page or how many questions they attempt — engagement metrics that inherit precisely the miscalibration problem they’re meant to correct. The competency signal looks different: topic-level accuracy, stability across sessions, and performance as difficulty rises. A study in the Journal of Learning Analytics by Kovanović and colleagues makes the measurement stakes concrete: different methods of estimating time on task can materially change the observed relationship between time spent and performance outcomes. If a foundational metric shifts its story depending on how it’s calculated, it cannot reliably anchor readiness decisions. Engagement data is context; demonstrated performance is the signal.
That measurement instability has a natural downstream implication: platforms positioning themselves as readiness instruments need to be grounded in demonstrated competency, not logged activity — which is where the commercial assessment market is now placing its development effort. Instructure’s expansion of its Mastery Predictive Assessments — rolling from 13 states to 27 more beginning in the 2026–2027 school year — illustrates the direction. The product is positioned explicitly as a readiness predictor: designed to measure previously taught content mid-year and generate state-aligned signals about likely summative exam performance. Instructure points to an Every Student Succeeds Act (ESSA) Level III study in which districts pairing the tool with other Mastery assessments saw students gain up to 38 percentile points per two assessments completed — a figure drawn from the company’s own citation of the study rather than independent evaluation. The commercial positioning, regardless, speaks for itself: benchmark platforms are increasingly being sold on the premise that they can read exam readiness before the exam arrives.
In exam preparation, platforms that close the diagnostic gap embed analytics directly into the practice workflow. Students preparing for International Baccalaureate (IB) Diploma and International General Certificate of Secondary Education (IGCSE) examinations face the core problem that private tutoring historically addressed: hours of study can accumulate without any precise signal of where performance actually falls short of exam standards. Exam-focused practice environments that track how students handle questions over time — not just deliver them — are the structural answer to that gap. Revision Village operates on this basis: a bank of syllabus-aligned, exam-style questions, filterable by topic and difficulty, attempted under timed conditions, generates a stream of performance data tied to specific concepts rather than to aggregate hours logged. Performance analytics dashboards then aggregate this activity, flagging topics where accuracy data indicates gaps remain; through the School Partnership Program, teachers gain the same topic-level view at class scale. In this configuration, a student’s readiness is expressed as a pattern of topic-level results rather than a tally of hours studied.
Evidence and Access — Who Benefits?
Without access to analytics, learners typically organize revision around instinct: topics that feel familiar get less attention; those that trigger anxiety get more. Those signals track emotion and study history, not exam vulnerability. Topic-level performance data can shift that calculus — but only when it is presented in ways that are interpretable and actionable. An experimental study by Kim, Jo, and Park found that students given a learning analytics dashboard achieved higher final scores than a control group, suggesting dashboards can support better regulation of study effort. A systematic review by Matcha and co-authors found that many learning analytics dashboards are not grounded in learning theory and cannot generally be recommended as tools for supporting metacognition. The platforms most loudly promising self-regulation gains are often the ones least grounded in the theory that would make self-regulation achievable. Both findings together point to a conditional claim: dashboards can improve how students plan and monitor their study, but only when the display makes clear where performance is unstable and what specific adjustments that instability calls for.
Scaled to a full class, the same logic shifts from self-regulation to instructional decision-making. When a teacher can see that students are accurate on a topic at lower difficulty but consistently falter as questions become harder, that pattern calls for a different instructional response than a handful of random errors scattered across the cohort. Aggregated, topic-level analytics can show where reteaching or targeted practice is warranted before an examination — turning diagnostic precision into an instructional asset rather than a gap discovered only in the results afterward.
Diagnostic granularity of that kind has historically been a paid-for advantage — the private tutor’s observational edge, purchased per hour. When that same capability is embedded in widely accessible platforms, it extends that form of insight to students who otherwise would never receive it, narrowing at least one dimension of the preparation gap tied to family resources. Access and assessment literacy still matter: students need connectivity, and they need enough understanding of what the data shows to act on it. Both considerations inform policy-level thinking. The National Governors Association’s policy academy — an 18-month initiative helping participating states build longitudinal student-readiness dashboards — reflects the same institutional bet: that continuous, disaggregated readiness evidence can guide better decisions than periodic, aggregate test scores. The more precisely that data captures where a learner struggles, however, the more sharply the question of what happens to it afterward demands an answer.
Accountability and Data Sensitivity
The educational case for dashboards rests on granularity — accuracy by topic, time spent on each question, patterns across attempts — and that is precisely the kind of detail at the center of the California i-Ready lawsuit against Curriculum Associates. The complaint alleges that performance data is paired with sensitive demographic information and shared commercially without adequate consent; the company denies the claims and the case is unresolved. The complaint’s catalogue shows what diagnostic data actually consists of at the level of detail that makes it useful — and the same specificity that powers the diagnostic function is what becomes acutely sensitive when questions of sharing and purpose arise.
Regulators and investigators have documented that this risk is real rather than hypothetical. The US Federal Trade Commission found that edtech provider Edmodo had “unlawfully used children’s personal information for advertising” and stated that schools can authorize collection only “for an educational purpose,” not commercial use; a 2022 FTC policy statement separately raised concerns about “persistent identifiers used to target advertising to children.” Human Rights Watch, reviewing 163 education technology products, reported that 145–89% of those examined — “appeared to put children’s rights at risk,” frequently through behavioral advertising and third-party data flows.
These findings sharpen, rather than weaken, the case for analytics: if fine-grained performance data can close the gap between feeling prepared and being prepared, the systems that collect it must treat that information as both educationally valuable and intrinsically sensitive — with consent frameworks, transparency, and strict limits on secondary use built in by design, not retrofitted afterward. Accountability is not an addendum to the dashboard’s promise. It is the test of whether an instrument that sees learners so clearly is actually working in their interests.
Building a Meaningful Dashboard
A meaningful dashboard is harder to build than it sounds. The design test isn’t whether a platform can surface a number — nearly all of them can. The real test is whether a student who opens the dashboard can immediately translate what they see into a concrete, prioritized plan: which topics to revisit, at what difficulty, and in what order. Research suggests dashboards can support more effective self-regulation when they’re designed to do exactly that — and that many current designs don’t, offering raw traces and progress bars that visualize activity without telling learners what to do with the information. The challenge isn’t visualization complexity; it’s alignment between what the display shows and the decisions students and teachers actually need to make.
The debates now surfacing around student data collection show why that alignment can’t remain purely pedagogical. The same logs that reveal where a learner’s accuracy collapses, where performance degrades under time pressure, and which concepts produce consistent errors also constitute a detailed record of a specific mind under stress. Who holds that record, who can access it, and for what purposes are not peripheral governance questions — they are definitional. A diagnostic instrument that tells you, earlier and more precisely than any prior tool, where you are likely to struggle earns that credibility only if the data behind it was gathered with meaningful consent, protected from repurposing, and used in ways that leave the learner more capable rather than more legible to someone else.
About the Author
