Case study — Working prototype
Measuring transformation, not activity
Fuller Equip exists to form people, not just to sell courses. Every dashboard the platform came with measures the second thing. This is a working prototype of what it looks like to measure the first — built on real production data from one learner’s nineteen months on the platform.
- Mission-driven metrics
- Measurement design
- Learning science
- Data storytelling
Every metric we had was a business metric
Enrollments, completion rates, monthly actives, revenue. These metrics are necessary — a platform that can’t sustain itself forms no one. But they answer one question only: is the business working? They are silent on the question the institution actually exists to answer: is anyone being formed?
The gap is easy to ignore because commercial metrics look like mission metrics from a distance. A completion feels like formation. It isn’t — it’s an event. Formation is what happens across many events over time, and whether it holds afterward. A learning business can hit every commercial target and still fail at the thing it was built to do — and its dashboards would never notice.
What the standard dashboard sees
Enrollments
44
Completions
16
Lifetime revenue
$235
Last active
Today
Verdict: healthy customer. That is the entire depth of the insight — and not one of these numbers can say whether nineteen months on the platform changed how this person practices, thinks, or leads.
Formation has physics
If transformation is the goal, the first product task is to define what it would look like in data. Not perfectly — measurably. The working definition: sustained, reinforced engagement with a competency over time. Not a completion. A practice.
That definition has consequences a funnel can’t express, because formation behaves less like a pipeline and more like orbital mechanics. Evidence gives a competency mass. Time without reinforcement pulls it away. Learning science has known this for a century as the forgetting curve — the model simply takes it seriously: every competency decays on a 180-day half-life unless something resets the clock.
So instead of a funnel, each learner gets a sky: competencies as orbiting bodies, held close by evidence, drifting when neglected. The metaphor isn’t decoration — it’s the model.
Evidence
Completion ratio + tracked time + breadth across a competency. Not one finished course — a body of work.
Recency decay
A 180-day half-life. Formation that isn't reinforced drifts outward, no matter how deep the original evidence.
Orbit distance
Evidence × decay. The closer a competency orbits, the more alive it is in this person's practice right now.
Goldilocks zone
Where formation holds: sustained, completed, still warm. The product's job is to keep competencies here — not to maximize clicks.
One learner, nineteen months, mapped
This is the prototype, running on real data: SCORM Cloud player activity joined with Strapi accounts, enrollments, and orders. Thirty-four registrations, sixteen completions, 214 tracked hours — scored into eight competencies. Select a planet to read its orbit; select the learner for the joined profile. The name is changed; every number is real.
What the orbit sees that the dashboard can't
The renewal moment no dashboard shows
135 hours of arts formation, a complete certificate with a final project — and seven months unreinforced, it's drifting toward the outer edge. A commercial dashboard files this under "churn risk." The orbit says one well-timed course resets the decay clock on a year of formation. That's a retention play and a mission play in the same move.
Not all engagement is formation
One browsing day in March 2025 produced five enrollments. The dashboard counts five. The orbit calls it what it is — discovery behavior — and lets it drift. Refusing to count sampling as formation is the difference between a metric you can trust and a metric you can inflate.
A stall is a content signal, not a learner failure
The trauma course's February campaign cohort mostly never launched. This learner invested five real hours in three days, then stopped cold at minute 296. When your most committed learners stall at the same point, the course is the variable. The orbit turns a churn statistic into a redesign brief.
Some orbits run on the liturgical calendar
A Holy Week pairing — one short course finished in a sitting, its companion left mid-flight the same day. Seasonal, devotional engagement deserves seasonal recommendations, not a generic 14-day re-engagement drip. Knowing which calendar a learner lives on is product insight no funnel metric carries.
The healthiest signal requires nothing
The hottest competency in this sky — five completions and 51 hours in three weeks — needs no intervention at all. The discipline is recognizing it: the product's only job there is to have the next certificate ready before they ask. Most engagement tooling would spam this person anyway.
The principle travels
Nothing here is specific to theological education. Every mission-driven product has a transformation it exists to cause — health outcomes, financial resilience, craft mastery — and a set of commercial metrics that stand in for it because they’re easier to count. The proxies are necessary. The failure mode is forgetting they’re proxies.
The product leader’s job in that gap is specific: operationalize the mission into something measurable, however imperfect, and make the commercial metrics serve it as constraints rather than replace it as the goal. An imperfect measure of the right thing beats a precise measure of the wrong thing — and it changes what the team builds next. Here it reshapes the recommendation engine, the re-engagement calendar, and the content roadmap in ways a completion rate never could.
You become what you repeatedly attend to. So does a product. Point the instruments at transformation, and the roadmap starts aiming there too.
Where this stands
A working prototype against live production data — not yet shipped to stakeholders. The next proofs: cohort-level skies so program admins can see formation across a community, hooks from orbital state into the recommendation engine, and stall-point analysis feeding content redesign.