Autoship & Save | Liki24
Build one ROI-positive retention initiative for Liki24.de | By Yevhenii Holovei
Why This Initiative
Why Autoship & Save
Four structural advantages that make this the strongest retention lever
Creates a predictable restocking habit tied to a specific product -- not to a discount or campaign. Customers opt into a repeating behaviour. This is the strongest and most durable form of retention.
Works across both surfaces, maximising the eligible user pool from day one. Yearly Membership (mobile-only) was immediately deprioritised for this reason alone.
Only products with genuine repeating demand qualify -- filtered via Product repurchase % from Bestsellers data. This avoids cannibalising one-off category buyers and targets the highest-intent users.
Unlike Personalised UX, every key metric -- frequency delta, subscription churn, LTV uplift, cannibalization -- has a clean formula and maps directly to an A/B test cell.
Alternatives Considered & Rejected
Why the other three directions were deprioritised
| Initiative | Decisive Weakness | Verdict |
|---|---|---|
| Yearly Membership7% discount | app-only | Mobile-only caps TAM. Annual commitment is a high entry barrier. | Rejected |
| Personalised UXTop-category experience | Improves experience but creates no obligation. Impact is indirect and hard to isolate. | Rejected |
| Smart Re-purchase TriggersBehavioural push/email | No customer commitment. Notifications decay in effectiveness over time. | Rejected |
| Autoship & Save10% off | web + app | Cannibalization risk -- manageable via A/B test with proper control group. | Selected |
Target Segment
Segment Definition: Product Repurchase ≥ 20%
Why product-level repurchase rate is the right filter -- not category volume
The Bestsellers file contains a Product repurchase % column -- the only signal that directly measures repeat buying behaviour at the SKU level. Customers with repurchase ≥ 20% are already self-returning; Autoship formalises and accelerates that pattern.
Why NOT category rank alone: Antiparasitic is #1 by order volume (1,824 orders) but repurchase is only 1–7%. Offering Autoship there is a pure margin discount with zero behavioural additionality.
Cohort sizing -- Jan 2026: 4,247 total → ~30% eligible (~1,274) → 15% activation → ~191 autoship subscribers
| Product | Repurchase % | Autoship Fit |
|---|---|---|
| Maraton Forte | 54% | Strong |
| Power V8 | 41% | Strong |
| Minus 30 | 33% | Good |
| Magiun afrodisiac | 27% | Good |
| Pharmatex | 22% | Marginal |
| Antiparasitic category | 1–7% | Exclude |
Core Hypothesis
LTV Uplift Hypothesis & Model
If users with repurchase ≥ 20% switch to Autoship at 10% off, frequency + retention gains will outweigh the discount cost
Incremental LTV =
AOV × (1 − 0.10) × freq × 1.10 × retention_autoship
− AOV × freq × retention_baseline| Variable | Value | Basis |
|---|---|---|
AOV | $51 | Average across cohorts |
freq_baseline | 1.05 /month | Jan 2026 cohort actuals |
retention_baseline | 4 months | Historical baseline |
retention_autoship | 6 months | Target (Amazon analogue) |
Discount | 10% | Activation optimisation |
Frequency uplift | +10% | Monthly order rate target |
= (51 × 0.90 × 1.05 × 1.10 × 6) − (51 × 1.05 × 4)Success Metrics
Northstar Metric
Primary Metric
LTV uplift > Discount cost + Cannibalization. Measured via A/B test. Large cannibalization will destroy gross margin even if subscription LTV looks positive on paper.
Early Stop Condition
Stop test immediately if gross margin in the test group drops >3% vs baseline.
| Metric | Target | What it signals |
|---|---|---|
| Autoship Adoption Rate | ≥15% | Offer resonance |
| Autoship vs non-Autoship LTV | +15–25% benchmark | Incremental value |
| Active at 1 / 3 / 6 months | Monitor churn curve | Retention quality |
| Orders/user (test vs control) | +10% vs baseline | Frequency uplift |
| Cannibalization Rate | <30% | Margin protection |
| Subscription Cancellation Rate | <35–40% | Product-market fit |
UX Flow & Product Design
Purchase & Subscription Flow
End-to-end user journey across 6 core states
ROI Logic
Revenue ROI Model -- Jan 2026 Cohort
Bottom-up from cohort sizing to net incremental value across three activation scenarios
Below activation target. Initiative is still ROI-positive but signals the offer needs iteration -- likely on discount size, placement, or SKU selection.
4,247 cohort → ~1,274 eligible (30%) → 191 subscribers. Net ROI per user $103.89 → cohort total $19,826.
Strong product-market fit. Validates scaling Autoship to a broader SKU catalogue and potentially increasing the discount in A/B iteration.
All figures are Revenue LTV, not Profit LTV. Margin impact of the 10% discount and cannibalization must be modelled against actual category gross margin data before final go/no-go.
Key Risks
Risk Assessment
Five key risks ranked by combined probability × impact -- with mitigation strategies
Customers who would have purchased anyway receive a 10% discount -- pure margin destruction. Must be isolated via A/B test with a clean control group. Treat incremental orders, not total subscriber orders, as the numerator.
Even eligible products (repurchase 20–54%) still have 46–80% of customers who don't return regularly. Track cancellation rate by delivery cycle -- a spike at cycle #2 signals frequency mismatch.
10% may be insufficient if users distrust auto-billing. Mitigate with: prominent "Cancel/Skip anytime" copy, A/B test on discount size (0% vs 5% vs 10%).
Autoship creates predictable demand but requires predictable supply. Mitigate with proactive email 5 days before if SKU is at risk of OOS.
German law may restrict auto-billing for certain near-prescription categories. Mitigate with legal review before launch; start with clearly non-regulated SKUs.
Experiment Design
A/B Test Structure
Three-cell design to isolate the mechanism effect from the discount effect | Min. 8 weeks | 2 delivery cycles
No Autoship option shown. Establishes the true repeat-purchase baseline without any intervention. All uplift in test cells is measured relative to this group.
Autoship mechanic with no financial incentive. Tests whether convenience alone drives subscription adoption and what frequency uplift it produces.
Full offer: convenience + 10% discount. Tests whether the financial incentive provides incremental activation lift above convenience alone.
| Population | Repeat buyers of qualifying SKUs, ≥2 orders in last 90 days |
| Randomisation unit | User ID -- prevents cross-group contamination |
| Duration | 8 weeks minimum (covers 2 full delivery cycles) |
| Traffic split | 34% / 33% / 33% |
| Expected cohort | ~2,000–4,000 new eligible users/month |
| Significance threshold | p < 0.05 for primary metrics |