Scoring from footageTue 6:15 pm · Strength & Stretch · Dana

How good was this class?

InstructorReach, 1.5 body heightsFirst visit

0.0 s

5 people

Block plan · room movement against the Strength & Stretch templatePlannedMeasured movement—

Try it · the score recalculates

Capacity

Start illustrative

Plan template

First-timer check-in, before or after illustrative

0Class scoreScoring…

Reading pose data…

Starting weights. In the pilot they're refit to your own 7-day rebook data.

Modeled12 weeks · one studio · not customer data

Zoom out: one class is a data point.

Every class gets the same score, every week. The last point on each chart is the class above, scored live; everything else is a model of what a studio would see.

This slot, week by week

Tue 6:15 pm · Strength & Stretch

Modeled + today

Instructor spread, anonymized

HQ view · distributions only, never names · last 12 weeks

Modeled

Each row is one instructor's classes: middle half and range. HQ sees how wide the spread is, not who is where. Managers see each instructor's own four-week trend; instructors get a private recap after every class.

Does the score predict who comes back?

The pilot tests it · each dot is one class · share of its members who rebooked within 7 days

Modeled

By format

Median, middle half and range · amber ring is today

Modeled

Dana's private recap · sent after class · only Dana and the manager see it

Three things for next Tuesday.

Class Quality Score

One number for how a class went against its own plan, built from six things a camera and the booking system can see. It turns "this class feels flat" into a private, specific note the instructor can use next week.

What you're seeing

A real class clip with real pose tracking. As it plays, Room scores six components and the gauge fills, one segment per component, sized by its weight. Change the capacity, the start time or the plan template and the score recalculates. Below, the same score over twelve weeks, the anonymized spread across instructors, formats, and rebooking, all modeled.

How it's scored

Fill (15): members present against capacity. 85% full earns full marks.
On-time start (10): full marks within 30 seconds, then 10 points off per 30 seconds. Illustrative here; a real install reads it from the door camera and the schedule.
Block plan (20): the room's movement against the format's planned blocks. Barre is plank → thighs → seat → core, so holds should look still; HIIT alternates work and rest; yoga builds to a peak pose, then savasana. Shape match is a correlation. This is movement, not effort: effort comes from a heart-rate band or bike power.
Sync (20): how closely the room moves together. An average of 0.75 earns full marks.
First-timer check-in (20): every first-timer got one individual check-in, before, during or after class. In class, that's the instructor within 1.5 body heights for 8 seconds straight. Leading from the front is right for most blocks, so distance walked is never scored.
Modifications & early leavers (15): who took the planned modification (here, hands on hips instead of the overhead reach), and who left before the end. Up to a quarter of the room modifying is normal; past that, the block was pitched above the room.

To be clear

The clip is ten seconds long, so it stands in for a full class: the block-plan curve here is the shape of those ten seconds, squeezed onto a 45-minute template. The instructor is found automatically (nearest the camera, centre, facing the room). The first-timer comes from the booking system; the start time, the before-or-after check-in and everything in "Zoom out" are illustrative or modeled. The weights are a starting point, to be fit to your 7-day rebook data in the pilot. The link between score and rebooking is what the pilot tests, not a finding.

Why owners care

Members cite the instructor as the top reason they return, and nobody has a fair, repeatable way to coach it. Each instructor gets a private two-minute recap after class; the manager sees four-week trends; HQ sees anonymized spread only. It's built to produce credible instructor-variance data, validated against your rebooking in the pilot.