You can spend three hours revising and walk away confident in exactly the wrong topics. The problem isn’t effort—it’s that the internal signals students use to judge their own readiness are systematically unreliable. Robert Bjork identified this pattern as “illusions of competence”: states in which training performance and subjective confidence diverge sharply from durable learning. When retrieval feels fluent, confidence rises; when study and test conditions differ from revision conditions, that confidence can be seriously misplaced in ways the learner cannot detect from the inside. Koriat, Ma’ayan, Sheffer, and Bjork (2006) traced the specific failure: judgments of learning are systematically biased by cues that differ between how material was studied and how it will be tested. A student can finish a revision session convinced of mastery in the very areas most likely to fail under exam conditions.
The bottleneck in exam preparation is rarely a shortage of content. It is the learner’s systematic inability to locate their own competency gaps accurately. A meta-analysis by Sitzmann and colleagues (2010) found that self-assessed knowledge correlates more strongly with affective variables—satisfaction and motivation—than with actual cognitive outcomes, making it a poor guide to what still needs work. Performance dashboards offer a corrective by turning accumulated practice behavior into an external competency map. Whether they deliver on that promise depends on a harder question than it appears: is the dashboard designed to help the learner act, or merely to show them what the platform knows?
Measuring the Wrong Thing, Confidently
Analytics platforms produce two structurally different kinds of dashboards. Those that surface engagement behavior—logins, time on platform, lessons completed—are administratively convenient but limited predictors of achievement. A 2022 systematic review and meta-analysis by Wang and Mousavi found that common engagement-log variables show small-to-modest integrated effects: login frequency (β ≈ 0.04) and login time (β ≈ 0.17) are both moderated by context, meaning the same metric can signal different things across settings. A 2021 study in Computers & Education cautioned that behaviors like online activity and logging in invite misleading interpretations without additional framing. Corrin and colleagues (2016) found that students themselves recognized the limitation—dashboard-visible login time, they noted, is not necessarily an accurate proxy for actual engagement or learning. Dashboards that foreground competency-relevant signals and explain what those signals mean occupy fundamentally different informational territory.
Evidence that metric choice changes outcomes came from a 2026 peer-reviewed study published in Interactive Learning Environments. The study’s dashboard intervention measured genuine student interaction with course content rather than passive time on platform, and the results were instructive: the intervention significantly increased engagement with course elements and reduced the academic achievement gap between at-risk and non-at-risk students in online learning environments. The dashboard did not merely provide convenience to students already on track; it produced a measurable equity effect.
Metric quality alone doesn’t settle the design question. A study published in Education and Information Technologies in July 2026 tested whether presenting the same performance data differently changes its cognitive effect. In a five-week university programming course, 30 students were assigned to one of three conditions: no agent, a “telling” agent that delivered performance data directly, and an “eliciting” agent that required students to self-assess before seeing system metrics. Students in the eliciting condition showed more reflection and more accurate self-judgments of mastery. The data were identical across conditions; the mechanism of presentation determined the metacognitive outcome. Those same questions don’t stop mattering when the user is a teacher rather than a student—but unlike a student reading their own data, a teacher’s interpretation of the dashboard shapes what an entire group of students experiences next, turning a personal calibration problem into a structural question about how instruction gets distributed.
What Teachers See—and What the Dashboard Doesn’t Show Them
The difference between a useful teacher dashboard and an elaborate attendance record is whether visibility translates into teachers actually teaching differently. A quasi-experimental study by Karademir and colleagues (2024), published in the Journal of Learning Analytics, examined a teacher dashboard with an integrated feedback tool—the “LA Cockpit”—across 16 teachers, 22 classes, and 403 secondary students. Teachers used the dashboard to identify students facing learning difficulties and sent targeted feedback through the system; the dashboard condition produced significantly higher pre/post knowledge gains than controls. The knowledge gains resulted from teachers acting on what the dashboard surfaced—specifically, sending targeted feedback to students they identified as struggling—rather than from having greater visibility alone.
Zearn Math, a K–8 digital mathematics platform used in U.S. schools, provides teacher-facing Progress Reports that show each student’s current position in the digital learning sequence and are designed to support grouping for small-group instruction. In a year-one evaluation conducted by Johns Hopkins University (2019), teachers reported using these reports at meaningful rates—74.0% used the Pace Report at least frequently, and 52.2% used the Progress Report—and described incorporating them into instructional planning and differentiation routines, characterizing the real-time data as information their prior curricula had not provided.
Independent research identifies what these dashboards still miss. A study of the MATHia intelligent tutoring system, led by Qiao Jin—assistant professor of computer science at North Carolina State University—and presented at the 2026 Learning Analytics & Knowledge Conference, analyzed more than 1.4 million student–system interactions across 14 middle- and high-school math classes. Teacher help was “sticky”: even after accounting for each student’s current engagement state, teachers repeatedly assisted the same subset of students, while those who remained persistently idle received less sustained attention. The help that did occur coincided with within-session improvements but did not predict skill acquisition in later sessions. Dashboards that make student placement and progress visible can help teachers base grouping and planning decisions on data they would not otherwise see—though current tools generally don’t yet track or equalize how teacher attention is distributed across a class.
Diagnosis Without a Remedy Is Just Anxiety
The risk of dashboards inducing anxiety rather than action is real, but it’s design-sensitive rather than inevitable. A 2021 study of dashboard design in an online statistics course found that norm-referenced visuals function as a “double-edged sword”: below-average signals triggered feelings of embarrassment and failure even when no significant differences in final exam or quiz scores appeared between conditions. Corrin and colleagues (2016) documented the anticipatory version of the same pattern—students expected that poor signals would make them feel “dejected” and ready to give up, with unmanageable remediation options amplifying that response rather than relieving it. The design variable is not whether gaps are surfaced but whether the platform immediately routes the learner somewhere actionable.
A case study at The Open University, presented at the 2025 Learning Analytics & Knowledge Conference, deployed a student-facing dashboard combining descriptive, predictive, and prescriptive analytics with 30 distance-learning undergraduates over four to fifteen weeks. Students described the prescriptive elements—recommended actions, suggested learning materials, and prompts to contact support—as especially valuable for motivating study planning and remedial action. Outcomes were largely self-reported, supporting mechanism plausibility rather than effect magnitude. Ioana Jivet, a postdoctoral researcher at the Leiden-Delft-Erasmus Center for Education and Learning and TU Delft and a PhD researcher at the Open University of the Netherlands, has argued that most dashboards stop too soon—making performance data visible but failing to add the scaffolds that help students decide what to do with what they’ve just seen. She states the principle directly: “Awareness is not enough. For learning dashboards to have an impact, interpreting the displayed information needs to be followed by decision-making and action.”
Revision Village, a preparation platform for International Baccalaureate (IB) Diploma and IGCSE (International General Certificate of Secondary Education) students used across more than 135 countries, builds this logic into its analytics architecture. Performance dashboards track topic-level progress, flag areas needing further work, and route students to specific practice questions with written mark schemes and step-by-step video solutions—all within the same environment. That continuity matters more than it might seem: a gap signal with no remedy attached tends to produce exactly the kind of unresolved awareness that tips into anxiety rather than action. Teachers using the School Partnership Program access class-level views of the same data for monitoring and planning. Diagnostic output and remediation content share a single platform, so a student who identifies a weak topic does not have to leave the environment that surfaced the gap in order to address it.
The Same Infrastructure, Turned Against the Learner
The same analytics that track whether a student has mastered integration by parts can also log their IP address, disability status, school-lunch eligibility, and the time spent on each question—and that data doesn’t always stay within the learning context it was collected for.
Three 2026 developments establish what those failures look like in practice. California parents filed a lawsuit against Curriculum Associates, maker of the widely used i-Ready platform, alleging the company collected student data—including race, gender, disability status, school-lunch eligibility, assessment responses, time per question, and IP address—and used it to refine commercial products without adequate parental consent; Curriculum Associates denied all allegations. The UK Information Commissioner’s Office, publishing its “Edtech examined” report after auditing 28 edtech providers operating in UK schools, found sector-wide failures in data minimization, storage limitation, and privacy documentation when children’s data was used for analytics, issuing 596 recommendations, of which providers accepted and implemented 98%. A separate analysis in Frontiers in Education examined Moodle, Canvas, and Blackboard and concluded that engagement tracking and learning analytics structurally enable surveillance, aggregation, secondary use, and distortion through ordinary platform use—harms that basic consent and regulatory compliance do not resolve because students have limited visibility into and control over downstream data uses. The most serious privacy risks here aren’t violations of policy—they’re features of how the systems are designed to operate every day. The common thread is that platforms are frequently engineered around institutional oversight and commercial data interests, with learner transparency and control treated as secondary.
Who Dashboards Are Really Built to Serve
As performance analytics become a standard expectation in digital learning environments, metric choice, presentation, remediation, and governance don’t function as isolated design decisions—they all express the same underlying intention: who the platform is actually built to serve. When that answer is the learner, the diagnostic and the remedy arrive together. When it isn’t, the same infrastructure that could close competency gaps instead accumulates data for purposes the learner cannot see.
A platform may know more about a learner’s competency gaps than the learner does. The question is what it does with that knowledge: whether it closes the gap—as Zearn Math’s teacher-facing reports demonstrate when they drive grouping and differentiated instruction, or as Revision Village’s integrated analytics and remediation architecture demonstrates for self-directed exam preparation—or routes it toward institutional and commercial interests the learner had no say in shaping. The student just lives with it.
