The rocks are formally measured on September paid cohorts — but paid relaunched mid-July (two Meta iOS campaigns, ~$2.6k July / ~$1.3k August-so-far spend), so the July–August cohorts are an early read on where paid quality is landing before September starts.
Rock 1 · Week-1 retention — ⚠ still tracking below baseline. The paid-scaled weeks are returning in week 1 at ~24% — under the 29.8% May baseline and well under the 35.8% September target. A small uptick from last week (Jul 26 settled to 26.5%) but not near target.
| Cohort week | New users | Week-1 return | |
|---|---|---|---|
| Jul 5 | pre-scale | 1,175 | 33.2% |
| Jul 12 | paid scaling | 1,626 | 23.2% |
| Jul 19 | peak paid | 1,342 | 23.5% |
| Jul 26 | paid | 1,322 | 26.5% |
| Aug 2 | paid | 1,429 | 24.3%* |
| Paid-scaled (Jul 12–26, mature) | tracking | 4,290 | 24.3% |
*Aug 2 W1 not yet fully matured; Aug 9 pending. Read up, not down: these are blended paid+organic cohorts — organic retains higher, so pure-paid W1 is lower than the 24% shown. Paid can't be cleanly isolated post-May (no attribution split; Adjust paused). Net: after ~4 weeks of scaled paid, installs are still retaining worse than the May baseline — Rock 1 needs a real onboarding/quality lift before September, not just more spend.
Rock 2 · LTV — pending. The July cohort isn't mature yet (Ad-LTV unlocks ~Sept 7), so a comparable per-install LTV can't be measured. Mixed early signals: paid conversion improved (July ~2.2% paid-30d vs June ~1.0%, the trial-paywall lift) but retention fell (fewer active days → less sponsor value). Direction resolves at cohort maturity.
Share of each week's new installs that reopened the app the week after they joined (GA4 weekly cohorts). The two paid-surge weeks are the baseline; the organic weeks bracketing them show the house average for context.
| Cohort week | New users | Week-1 return | |
|---|---|---|---|
| May 3 | organic | 1,181 | 45.8% |
| May 10 | paid surge | 3,166 | 30.6% |
| May 17 | paid surge | 3,099 | 29.0% |
| May 24 | organic | 1,289 | 32.0% |
| May 31 | organic | 993 | 42.9% |
| Paid cohort (May 10–17) | baseline | 6,265 | 29.8% |
Paid installs returned in week 1 at 29.8% — vs a ~44% organic baseline (May 3 & May 31). Paid users come back at roughly two-thirds the organic rate; the +20% rock lifts that first-week return to 35.8%.
Same cohort-survival model as the all-traffic LTV page, run on the paid cohort's own inputs (its measured retention, its conversions). A paid install came out at $1.67 — about half the app's blended $3.28.
| Component | Paid cohort | All-traffic |
|---|---|---|
| Upgrade LTV | $1.23 | $2.43 |
| Sponsor LTV | $0.44 | $0.85 |
| Total LTV / download | $1.67 | $3.28 |
| Retention | Paid | Blended |
|---|---|---|
| D1 | 19.9% | 26.4% |
| D7 | 9.0% | 12.3% |
| D30 | 3.8% | 5.5% |
| Active days (5-yr) | 19.0 | 27.1 |
5-yr horizon, $4.50 sponsor CPC, 78% annual renewal, geometric sub-LTV. Lower LTV is driven by both worse retention (fewer active days → less sponsor value) and lower conversion to paid (upgrade LTV $1.23 vs $2.43). The +20% rock lifts a paid install to $2.00.
They're the same lever pulled twice. Week-1 retention (Rock 1) is the leading indicator; LTV (Rock 2) is the lagging dollar outcome. Retaining more paid installs past week 1 increases their active days (more sponsor-click value) and gives them more chances to upgrade — so a win on Rock 1 mechanically feeds Rock 2. Both are measured on September paid cohorts against these Q2 baselines.