Completion tells you what a learner did. The Learning Lab tells you what they know — split into seven separate skills, built from work they were already doing. For a tutor it answers the question a roster cannot: not "did they finish" but "what is actually thin".
The seven facets
"Knows 谢谢" hides seven abilities, and a learner is rarely even across them. The Lab reports each separately:
- Meaning 义 — From reveal-mode flashcards and reading practice.
- Character 字 — From choosing the character out of meaning and sound, in lesson practice.
- Pinyin 拼 — From write mode on a flashcard, and the write step in a lesson.
- Tone 声 — From tones typed alongside pinyin, and tone drills on dictionary entries.
- Listening 听 — From listening exercises and the listen step of a lesson.
- Speaking 说 — From speak mode and pronunciation practice.
- Writing 写 — From handwriting practice, drawn and graded.
Using it to plan a lesson
- Look for the shape, not the size — A small even radar is a beginner. A large lopsided one is someone with a specific hole, and the hole is the lesson.
- Compare the group — If six of eight learners are thin on Tone, that is a class you should teach rather than six conversations you should have.
- Read Calibration before Retention — A learner whose confidence exceeds their accuracy will not accept a correction until they see that panel. It is the most persuasive screen in the product.
- Use Errors for the next warm-up — Grouped mistakes, rather than a list of wrong answers. Five minutes at the start of a lesson on the group’s commonest error is the cheapest teaching you will do.
The other panels
- Retention — What is holding and what is slipping over time.
- Errors — The mistakes actually made, grouped by kind.
- Gap — The distance between what has been covered and what the level expects.
- Focus — Where attention has actually gone — often not where the learner thinks.
- Calibration — Whether confidence matches accuracy.
Two limits you will hit
Neither is a bug, and both surprise tutors on their first week.
Facet access follows the learner’s plan, not yours. A free learner exposes three facets even to a Pro Plus tutor. If your subscription decided how much of someone else’s profile you could read, we would be selling you access to their data.
Group figures need five learners. Medians and cohort comparisons are withheld below that, because an average plus one known score identifies the rest of a group of three. Per-learner rows and plain counts are not withheld.
They see the same screens
Your view of a learner is built from the learner’s own panels. There is no teacher-only report — nothing on your screen that they cannot see about themselves.
And nothing in the Lab contains what anyone typed. It reads outcomes: right, wrong, how long, how confident. See what your learners see.
Common questions
Their plan is free. Access is gated on the subject’s plan rather than the viewer’s — a Pro Plus tutor sees a free learner’s three.