Lotus Partners

Industry

Developing Practice Tier 3 Recurring Compounder

Educational Technology

Consumer edtech churns because it cannot prove anything happened. Engagement is rented month to month, and no library size substitutes for evidence of mastery.

The fix is assessment science as the product — skill models, spaced repetition, mastery gates — with content volume demoted to what those systems schedule. A subscription business whose renewal case is measured progress. Engagement mechanics borrowed from games hold attention for a while; only evidence of mastery holds a customer — and evidence requires instrumentation most of the sector never builds.

Industry research

The economics

Weeks the motivation half-life Learning subscriptions are bought on a motivation spike and abandoned within weeks unless the product schedules practice for the learner instead of waiting on willpower.
Rented the engagement subscription A subscription renewed on streaks is rented month to month; one renewed on measured progress is owned. The sector’s churn problem is the distance between the two.
COPPA-shaped products, by design constraint Children’s data law — COPPA, FERPA, the state ed-privacy statutes — is product architecture, not legal cleanup: consent flows, family accounts, and data minimization cost little designed in and a fortune retrofitted.

FIGURES FROM THE FIRM’S RESEARCH · SOURCES ON REQUEST

Growth opportunities

Where the 2× lives

  1. The assessment spine

    Skill modeled before content is written: rated dimensions, mastery gates, and adaptive selection that always serves the measured weakness. The spine is built first, and the lessons hang on it.

  2. Reinforcement as scheduling

    Spaced repetition treated as the core loop, not a feature: recall scheduled by the forgetting curve — FSRS-class scheduling, implemented and tested in the firm’s build — so what was learned on Tuesday survives the month.

  3. Content that proves itself

    Wherever a domain has checkable ground truth, the content pipeline verifies its own correctness in CI: every position, problem, and solution machine-checked before a learner sees it — in the firm’s own build, 3,733 curriculum entries are verified by engine. Content scales by pipeline, not headcount.

  4. A learning product, not a casino

    The deliberate refusal of engagement mechanics that optimize time-in-app over mastery. It reads as a values choice; it is also a durability choice — products paid on attention while bought for outcomes eventually meet their own gap.

Our perspective

What most firms get wrong

Answering churn with volume

The sector answers churn with volume — more lessons, more video, more gamification varnish. But a learner who cannot see measured progress has no reason to renew, and no library size fixes that. Retention is an assessment problem; the sector keeps answering it with content.

Where we’d start

Cohort decay read against the skill model: decay curves mapped to measured-progress events, the gap between engagement metrics and mastery evidence quantified, and the assessment spine — dimensions, gates, recall schedule — specified before another lesson is commissioned.

Put it to us

How we engage here