For tutors
师
Evidence you can teach from. Not who logged in most, but which learners are thin where, why a mistake keeps recurring, and what to assign on Tuesday because of it.
The Lab reads your own account, so opening it asks you to sign in first.
Why it exists
You could already see who finished what. That has never told you who understood it, and it has never told you what to do next — which is the only question a lesson plan actually asks.
Completion moves the same way for the student who followed every step and the one who clicked through it on the bus. Planning from it means planning from attendance.
The circle view counts retrieval separately from presence, so “done” and “knows it” are never the same column.
One number over twelve learners tells you nothing about the three who are quietly stuck, and it hides the far more useful pattern: several people failing in the same specific way.
Errors are classified by cause, so a learner who keeps missing one specific tone pair reads as exactly that, with the count behind it — not as a slightly lower average.
The student who answers fastest and never asks is often the one furthest from mastery, and is the easiest to miss in a room.
The Lab compares stated confidence against actual accuracy, and names the mismatch rather than the ranking.
What actually happens
Four steps, in order. Nothing about a learner is inferred from other learners, and the last step is allowed to refuse.
Ordinary practice — a review, a lesson exercise, a spoken phrase. Nothing is added to their workload, and there is no test for you to set.
What was asked, what they said, whether it was right, how long it took, and which of the seven facets it exercised. Appended, never edited.
A wrong answer is classified — a look-alike character, a homophone, a specific tone pair, a sandhi rule — so a recurring mistake reads as a cause rather than a tally.
The circle view orders learners by the reason they need attention, with the evidence that triggered it attached. Below the evidence floor it says so, instead of putting a red dot next to a name.
What gets measured
This is why a student can read fluently and freeze when asked to speak. Each facet grows at its own rate, and teaching to the strongest one is how a class looks fine until the exam.
Seven ways of knowing a word — meaning, character, pinyin, tone, listening, speaking and writing — counted apart, because in practice they come apart.
A free account opens the first three — Meaning, Character and Pinyin — with real evidence behind them. The other four are measured on Pro, on each learner’s own plan rather than on their teacher’s or their parent’s.
Create a free account to see what it measuresThe boundary
Your learners read the same boundaries, in the same words, on their own ethics page. Nothing about the teacher view is hidden from the people it describes.
How to trust it
Every figure carries its evidence count. A cohort below the minimum shows the shortfall instead of an average — a class of four does not get a chart that implies it means something.
There is no leaderboard and no class ranking. A learner’s accuracy is compared with their own earlier window, so a drop reads as a change in them rather than as a position in the room.
No webcam, no eye-tracking, no emotion or attention inference — not now and not later. What is recorded is what was answered and how long it took.
Questions
For learners in a circle you teach: mastery per facet, retention, which errors recur, and where they are thin. You do not see their cognitive-task results, and you do not see free text they wrote. You see nothing at all about learners outside your circles.
Yes — it is stated plainly on their own ethics page, in the same words as everything else. Nothing about the teacher view is hidden from the people it describes.
Aggregates are withheld below a minimum cohort size, because a “class average” over three people identifies them individually. You still see each learner you teach; it is the group comparison that waits.
Neither. There is no leaderboard, no class ranking and no teacher comparison anywhere in the Lab. Where movement is reported it is a learner against their own earlier window, which is the only comparison that survives a mixed-ability room.
Ordering learners by the reason they need attention rather than by score, and attaching the evidence behind it — “72% of 25 classified errors are one tone pair” — so the row tells you what to assign and why. It is one row per learner: it names the cause, it does not yet merge learners who share a cause into a single finding.
No. It reads the practice your learners already do. There is no separate assessment to administer, nothing to schedule, and nothing extra for them to complete.
You can, but read the evidence counts first. A facet below its floor is explicitly not a measurement, and the Lab says so rather than handing you a number that looks gradeable and is not.
A managed minor never appears in any cross-circle or public comparison, cannot consent to research on their own behalf, and their guardian sees a plain-language view rather than raw responses. Those rules hold regardless of what a teacher would prefer.
The cohort view fills from the practice your learners already do — there is no assessment to administer and nothing extra for them to complete. Until there is enough behind a figure, it says so.
What we record, and what we never do