RECOMMENDATION: Investigate recorded first-step progression and telemetry; do not treat
the fictional treatment as a proven winner or authorize a real rollout from this exercise.

The snapshot contains twelve eligible fictional users after excluding one internal account.
The source is assumed complete strictly before 26 January 2026 UTC. An exact repeated delivery
was removed, and two later events were withheld. Under the earliest-start, strict-order,
72-hour contract, twelve users start, eight reach a later lesson and five reach an exercise.
Overall conversion is 41.67%; conditional step rates are 66.67% and 62.50%. These counts locate
recorded stopping points; they do not explain user motivation or demonstrate learning quality.

Signup-week day-one returns are 3/4,2/4 and 1/4. Day-seven retention is 2/4 and 1/4 for mature
cohorts; the latest cohort is not yet observable. Pooled mature day-seven retention is 3/8,
or 37.50%. Dividing by twelve would wrongly classify unobserved follow-up as non-return.
The groups are tiny and observational, so the declining pattern is a hypothesis for investigation.

The separate synthetic experiment has100 assigned users per arm and 20 versus 28 completions.
TheITT difference is+8 percentage points,+40% relative, with an approximate95% interval from
-3.79 to+19.79 points. It remains inconclusive. The50:50 count check passes, but cannot prove
correct logging. Error users are 5 versus 7; sparse counts require a suitable separately verified
method and a predeclared practical harm margin before claiming safety. Outcomes were constructed
for teaching, not measured from a real intervention. The approximate plan needs 447 users per arm
under its stated assumptions; this fixture has only 100.

Next, validate event delivery and meaning, gather qualitative evidence about the transition,
and prepare a reviewed fixed-horizon experiment with adequate sample and complete follow-up.
Preserve assignment-based denominators, investigate missing observations and report uncertainty.
Commands and fixtures accompany the memo. This exercise activates no tracking, payment,
registration or live experiment, and establishes no actual commercial product effect.
