Everything looks fine—until you look closer
The overall success rate has not changed. Look at the groups, then set a warning that catches the hidden problem.
Ninja · Year 10Baselines, group results and alerts
Compare with a normal week
Fictional results. Success means a completed task in this demo. Each group has 100 attempts per week.
Who is getting different results?
| Group | Normal | Selected week | Change |
|---|
A change from 80% to 70% is a fall of 10 percentage points. Both groups count equally in the overall rate.
Choose what the helper watches
What happens on replay?
The overall rate is unchanged. Show the groups, then replay with the warning on.
Helper ready
- Waiting for a replay.
Illustrative sample results, not measurements of real people or AI. The helper and its pause-for-review behaviour are simulated. A curriculum practice example, not student work.
What the child
is learning.
Explore the learning journeyYear 10 · Baselines, group results and alerts
A baseline gives a comparison point. Breaking results down by group can reveal a problem that an overall average hides. A warning needs a tested response, such as pausing for review.
Now try this…
Replay the changed week with the warning off, then on. Next replay the normal week with the warning on. Does it pause only when the rule is met?