Skip to main content
Industries · Consumer Subscription

Autonomous AI discovery built for consumer subscription apps.

Continuous AI discovery across your onboarding, paywall, subscription, and engagement data. Every finding ships with the underlying queries, the cohort definition, and the cost of the run. The cohort-level questions that never make it past the weekly growth deck, answered before your next subscription review.

Built for Heads of Growth, Heads of Retention, VPs of Monetization, and Heads of Analytics at fitness, learning, productivity, dating, and wellness subscription apps.

CRITICALOnboarding cliff
Onboarding step 4 (set goal + pick reminder) completion fell 28% after the March UX update.
~118,000 affected signups. Users who skip step 4…
From a real DecisionBox run
Why DecisionBox

The cohort-level questions that never make it past the weekly growth deck, running overnight against your data.

Your team owns the paywall, the onboarding, the pricing, the lifecycle. The agent runs unattended against your acquisition, onboarding, billing, and engagement data — and ships findings into a queue your team triages. Every claim carries the underlying queries and the data-validated number, not the subscription-platform-reported or SDK-reported one.

See how the agent reads habit-formation behavior
The cost of late signals

The subscriber count looks fine. The cohort underneath is where the growth actually moved.

Three benchmarks every consumer subscription leader recognizes. Each points to the same gap: your data already holds the signals — habit-formation drop-off, channel-level LTV, paywall conversion drift — but they only surface when somebody asks the right question. Most weeks, nobody does.

D30 app retention
5%
Median Day-30 retention across consumer subscription apps. Habit-formation drop-off in week one drives most of the gap — and your data already holds the signals.
Adjust, Mobile App Trends Report
App-store fee
15-30%
Of subscription revenue paid to Apple and Google. The dashboard reports gross subscription; the underlying data reports net contribution after fees and cancel-friction differentials.
Apple App Store / Google Play developer terms
Free-trial conversion
~25%
Typical free-trial to paid conversion rate for subscription apps. Source-level conversion varies by 30+ points; the blended number conceals which acquisition channels convert and which buy installs that never pay.
RevenueCat, State of Subscription Apps
From real runs

Three findings the agent surfaced, paired with the recommendation it shipped.

Three shapes the team sees every quarter: a single-system onboarding-completion cliff the aggregate concealed, a cross-system habit-loop vs. price-elasticity pattern the dashboard couldn't see, and an app-store-vs-web LTV mirage the gross-revenue dashboard averaged into the wrong winner. Same auditable trail every time — severity, the underlying queries, the cohort definition, the cost of the run.

CRITICALOnboarding cliff
Onboarding step 4 (set goal + pick reminder) completion fell from 71% to 51% after the March UX update.
period: 6-week rolling, post-March 14 build · step…
Cross-checked against onboarding.step_events…
single-system onboarding cliff
P1Recommendation
Roll back UX + A/B retest at 10%
Roll back the March 14 UX update for non-iOS-promo cohorts;
Expected impact
target: restore step-4 completion + D30 retention…
Target: Roll back UX + A/B…2 concrete actions
CRITICALHabit-loop signal
The January price increase showed flat aggregate churn.
price increase: $9.99 → $12.99 (Jan 8) · aggregate…
Cross-checked across sessions.weekly_cadence…
cross-system pattern
P1Recommendation
Habit-stage price gating + comms timing
Hold price for users still in habit formation (under 3 sessions/wk);
Expected impact
target: stop training future change against in-formation…
Target: Habit-stage price…3 concrete actions
HIGHChannel LTV mirage
App-store subscriptions convert at 12% higher D1.
D1 conversion: app-store 24%, web 21% · 12-mo LTV…
Cross-checked across subscriptions.by_channel…
gross hid the net LTV
P1Recommendation
Reweight paywall + audit cancel friction
Reweight the paywall flow toward web checkout where eligible;
Expected impact
target: shift the conversion mix toward net-LTV-positive…
Target: Reweight paywall +…3 concrete actions

Point the agent at your subscriber data for a week. Triage what it finds Monday morning.

Use cases

Five surfaces the team already owns, one continuous discovery.

Same engine, same data, five categories of work your team is already doing — or wants to do, if the queue ever clears. Pick the surface that matters this quarter.

The cohort that signed up and never built the habit.

Onboarding step-level drop-off, habit-formation milestones (first session, third session, first-week cadence), and the patterns that separate users who build the habit from users who stall. The agent reads onboarding events, session cadence, and downstream subscription state together — and surfaces the cohorts whose habit-formation shape changed before the weekly growth deck shows it. The deep dive lives in how the agent decides what to investigate.

Sample analysis areas
  • Onboarding-step drop-off and its downstream retention impact
  • Habit-formation milestones (first session → third session → week-one cadence)
  • Build-version regression catches per release
  • Cohort-level signup → activation → first-week-habit funnel
  • Push notification and reminder cadence impact on session frequency
Why DecisionBox

Built for subscription apps who want a cohort analyst running overnight, not three dashboards to reconcile.

Three things every finding from DecisionBox carries. None of them changes how your team works in your subscription platform, your product analytics SDK, your paywall tooling, or whatever data tools and BI you already standardized on.

Sees what nobody had time to query for

The agent picks which cohorts, acquisition sources, onboarding steps, and channel segments to investigate from your acquisition, onboarding, billing, and engagement data. Findings ship into a queue your team triages in the morning. The cohort-level question stops sitting in the data waiting for the next QBR.

Every number you'd defend at the subscription review

The underlying queries the agent ran, the cohort definition, the data window, the cost of the run, the validation checks. Your team re-runs anything, extends anything, overrides anything. No black boxes, no billing-system-reported or SDK-reported numbers the data can't validate, no "trust the AI."

Reads all your data, not just RevenueCat and the analytics SDK

Subscription state plus onboarding, paywall, billing, sessions, and engagement — read together. The patterns RevenueCat, Adapty, Qonversion, or Glassfy can't see (because they only see subscription state) and the patterns Amplitude or Mixpanel can't see (because they only see events) surface when your data is read end-to-end. Snowflake, BigQuery, Databricks, Redshift, PostgreSQL. Open source, AGPL v3. The agent meets you on the stack you already run.

The next subscription review, with the cohort shifts already mapped.