What Your Class's Common Mistake Patterns Are Actually Telling You

You mark twenty-four essays over a weekend, and on eight of them you write some version of the same comment: "watch your tenses," "check subject-verb agreement," "remember the article here." Each comment feels like a one-off note to one student. It isn't. It's the same correction, written eight separate times, on eight separate papers, without ever being counted. That's the gap between individual feedback and class-level pattern-spotting — you can catch the same mistake all semester without ever seeing it as one thing.

Why individual feedback misses class-wide patterns

Feedback is written submission by submission, which is exactly why it's bad at surfacing patterns across submissions. When you're marking essay four, you're not thinking about what you wrote on essay one — you're reading essay four on its own terms. The correction you make is genuinely useful to that one student, but it disappears the moment you move to the next paper. Multiply that by a class of twenty-plus, a semester of assignments, and it's easy to spend hours re-explaining the same grammar point one red pen mark at a time, never quite noticing that it's the point half the class needs a five-minute lesson on, not twenty individual footnotes.

The fix isn't writing better individual feedback — it's already good. The fix is a second, separate view of the same corrections: not "what does this student need to fix," but "what keeps showing up across everyone's work."

What a common-mistakes dashboard actually shows

In Writing, no kidding, every reviewed submission carries AI-assisted corrections sorted into four categories: grammar, vocabulary, structure, and register — the same actionable analytics the AI checks when it first analyses a submission. Those four categories are also what the common-mistakes view aggregates once you've reviewed enough work for patterns to appear. Instead of showing four generic labels, each category lists the specific, recurring items inside it — ranked by how often they show up — so "grammar" isn't just a count, it's a frequency-ranked list of the actual grammar issues repeating across submissions: the kind of thing that in practice tends to cluster around tense consistency, subject-verb agreement, article use, preposition choice, and word order, because those are the corrections that recur most in ESL writing at almost every level.

Two views matter here, and they answer different questions. The per-assignment view (a "Common Errors" page attached to each assignment) shows what's recurring across every reviewed submission for that specific task — useful right after a batch of grading, when you want to know what to address before the next class. The class-level view, on your analytics dashboard, rolls patterns up across everything you've reviewed for a class over time, with a class picker so you can look at one group or all of them at once. Both pull from the same underlying data: only submissions marked as reviewed feed the analysis, so the dashboard is a genuine record of corrections you've actually made, not a guess or a live AI opinion running in the background. If you haven't reviewed enough work yet, it says so plainly rather than showing you something misleading.

That last detail matters more than it sounds. It means the dashboard fills in exactly as fast as your own grading does — it's a mirror of your marking, aggregated, not a separate assessment layer making its own judgments about your students.

From a frequency list to a plan

A ranked list of recurring mistakes doesn't teach anyone anything by itself — what it does is tell you where a five-minute warm-up would help more than the fifteenth individual correction on the same point. This is manual teacher judgment, and it stays that way: the dashboard surfaces the pattern, you decide what to do with it. In practice, that's usually a short list of moves:

• Pick the top one or two items, not all of them. A frequency list can have a dozen entries; trying to reteach all of them in one lesson dilutes the point. The top-ranked item is, by definition, the one affecting the most students.
• Turn the pattern into a warm-up, not a lecture. If article omission is topping the list, five sentences on the board with gaps to fill, done as a class, does more than a slide explaining the rule.
• Check it against the CEFR level you're grading to. A pattern around article use reads differently for an A2 class (still expected, still developing) than a B2 class (worth flagging directly). CEFR writing rubrics are useful here for calibrating whether a recurring mistake is normal at this stage or actually holding students back from the next one.
• Revisit it after the next assignment. Since the dashboard is cumulative, you can come back after the following piece of work and see whether the pattern actually moved, instead of assuming a five-minute warm-up fixed it.

A worked example

Say you've just finished reviewing a class set of twenty-two "describe a memorable trip" narratives. You open the assignment's Common Errors page and the grammar list is topped by a cluster of past-tense-related corrections, well ahead of anything else on the list — more than half the class made some version of the same slip, usually switching from past simple into present tense mid-paragraph once the narrative gets more descriptive. Vocabulary and structure show a few scattered items, nothing repeating enough to stand out. Register barely registers at all, since it's a first-person narrative and formality wasn't really in play.

That's a five-minute answer to "what does this class need next": a short board exercise on maintaining past tense through a descriptive passage, not a return to teaching past simple from the start, and not twenty-two individual comments repeating the same note. The pattern told you exactly where the next ten minutes of class time is worth spending — something no single paper, read in isolation, could have told you on its own.

The bottom line

Any one piece of feedback tells you what one student needs. A common-mistakes view, built from corrections you've already made across everyone's reviewed work, tells you what the class needs — and it only works because it's counting real, reviewed corrections, not running a separate guess. The judgment about what to do with that pattern is still yours; the dashboard just makes sure you're not the last to notice something that's already shown up eight times.

---

See your class's mistake patterns. Take a Tour — no credit card required.