Teachers are already generating more data about their students than they can act on. Every exit ticket, every warm-up problem, every homework set is producing signals — but the signals aren't being organized in a way that changes what happens in class on Monday morning. The problem isn't more data. It's five specific data points, available at the right moment, that are actually decision-relevant.
Why Most Tool Dashboards Don't Help on Monday Morning
The problem isn't that STEM teachers lack data. Most department leads we've spoken with describe a data environment that's actually quite noisy: benchmark scores from assessment platforms, accuracy reports from practice tools, engagement logs from LMS systems, reading-level flags from literacy screeners. The problem is that almost none of this data is organized around the question a teacher actually asks on a Sunday night before starting a new week: What do my students not yet understand, and what should I do about it tomorrow?
Benchmark scores tell you how students performed relative to a grade-level standard — useful for compliance reporting, not useful for deciding whether to spend Tuesday on re-teaching versus moving forward. Accuracy percentages tell you students are getting things wrong — but not which things, not why, and not whether the pattern is systematic or scattered. Engagement logs tell you who's logging in — not what's happening cognitively when they do.
What teachers consistently describe needing are a different category of signals: ones that are specific enough to drive a decision, timed appropriately within the instructional sequence, and organized at the class level rather than requiring individual student review to extract the pattern. Here are the five that come up most reliably.
1. Which Misconceptions Are Active Right Now
The most decision-relevant formative signal is the one that's currently absent from most tools: a labeled, class-wide view of which specific misconceptions students are carrying. Not "struggling with fractions" — which describes a symptom, not a cause — but "applying additive reasoning to fraction equivalence" or "confusing mass with weight in gravitational context problems."
The reason this matters instructionally is that different misconceptions require different re-teaching approaches. A student who applies the additive reasoning error to fractions needs an instructional sequence that directly confronts the invalid generalization — showing them why the pattern that worked for whole number addition doesn't extend to fractions. A student who has a procedural fluency gap (they know the rule but execute it unreliably) needs a different kind of practice. Treating both as "struggling with fractions" and routing both to the same remedial material is how re-teaching sessions produce no lasting change.
When this signal is available at the class level — "6 of your 24 students are applying the additive reasoning pattern, and they're not the same 6 who are lowest on benchmark score" — it changes the re-teaching decision. You're not addressing a broad performance gap; you're addressing a specific, named conceptual error with a specific instructional response.
2. Which Students Share the Same Misconception
The second signal is closely related to the first but distinct in what it enables: not just which misconceptions are present, but the cluster structure of those misconceptions across the class. Some misconceptions are broadly distributed — present in students across the ability distribution, because they arise from the curriculum sequence rather than from individual gaps. Others are concentrated in a specific sub-group, which suggests a prerequisite knowledge gap rather than a sequencing issue.
A 7th-grade science teacher working on a unit on chemical reactions once described this to us in concrete terms: she had two clusters of students making errors on conservation of mass problems. One cluster was misapplying the idea that mass is "lost" in reactions that produce gas, a classic alternative conception. The other cluster was making arithmetic errors in the balancing steps — they understood the concept but couldn't execute the procedure reliably. These two groups of students looked similar in score reports. They needed completely different instructional responses. A class-level misconception cluster view makes that distinction visible in minutes rather than requiring individual student interviews.
3. Misconception Persistence Across Multiple Problem Types
A one-time wrong answer on a specific problem might reflect a slip, a misread, or a bad day. A misconception that appears consistently across multiple problem types — varying in context, phrasing, and surface features — is structurally embedded in the student's understanding and won't resolve through additional practice alone. The persistence signal is what distinguishes a mistake from a misconception.
This signal requires a platform that tracks error patterns across items rather than reporting only item-level accuracy. Most LMS gradebooks don't do this. A student who gets three fraction problems wrong might have three different slips, or might have one stable misconception appearing across three contexts. The distinction is not visible from item-level accuracy data. It requires error pattern tracking across items within a knowledge component cluster.
From a teacher's perspective, the practical implication is this: persistent misconceptions that appear across multiple problem types are the cases where individual practice time is not well spent without direct instructional intervention first. Knowing which students have a persistent error pattern — before you assign another homework set — prevents the scenario where a student completes 20 practice problems while reinforcing the wrong conceptual model with each one.
4. Which Gaps Are Blocking the Next Unit
Not all conceptual gaps are equally urgent. Some gaps are gaps in extension content — missing nuance on a concept that isn't load-bearing for what comes next. Others are gaps in prerequisite content — missing understanding that the next unit's core concepts depend on directly. The fourth signal a teacher needs is the prerequisite blocking map: which of the current misconceptions, if unaddressed, will directly impair understanding in the upcoming unit.
This requires a content dependency model — a structured map of which concepts in the curriculum are prerequisite to which others. When a student carries a misconception that's a direct prerequisite to an upcoming unit, the urgency of addressing it before moving on is qualitatively different from addressing a gap in a non-prerequisite concept. Teachers intuitively know this; they lack the systematic data view to act on it consistently across a full class section.
A concrete example: a Grade 5 student with an unstable understanding of equivalent fractions will hit structural difficulty as soon as the class moves to adding fractions with unlike denominators. That connection is deterministic — equivalent fractions understanding is a hard prerequisite. If a teacher can see, heading into the transition between units, exactly which students carry the specific prerequisite gap, they can plan a targeted bridging activity rather than discovering the blocker mid-unit when there's less time to address it.
5. The Cross-Section Pattern
The fifth signal is one that individual classroom teachers rarely have access to, but that department leads and curriculum coordinators need: whether a misconception or gap pattern is appearing consistently across multiple sections of the same grade level and subject. This cross-section signal is what distinguishes a student problem from a curriculum problem.
If one section of 6th-grade math shows a cluster of students with the impetus-theory misconception, it might reflect the instruction in that classroom. If three sections of 6th-grade science all show the same misconception clustering at the same point in the force and motion unit, the more likely explanation is that the curriculum sequence isn't addressing the misconception before students have the opportunity to form it. The intervention required in the second case is a curriculum revision, not a re-teaching event in individual classrooms.
This is the signal that most tools provide last, if at all — it appears in year-end summative analysis, long after the current cohort has moved on. Making it available during the school year, aggregated across sections, is what enables the kind of mid-year curriculum adjustment that department heads know they should be doing but currently can't because the data isn't organized to support it.
What these five signals share is a common structure: they're specific enough to drive a decision, timed within the window when action is still possible, and organized at the level of conceptual understanding rather than performance score. Most current tools provide approximations of some of these signals, delivered too late and at the wrong level of abstraction. The gap between what's available and what's actually decision-relevant is the gap we're trying to close.