This is Part 3 of our DTC Growth Intelligence series. In Part 1 we covered where customers come from. Part 2 covered what happens after the first purchase. Now: predicting who will spend more — and who’s about to leave.
Acquisition tells you how many customers you bought. Activation tells you which ones came back. But neither tells you what’s about to happen. The metrics in this article are forward-looking: they detect behavior shifts 30-90 days before they show up in revenue. One set of signals predicts who will expand. The other predicts who will churn.
Most DTC brands react to churn after it happens. They run win-back campaigns on customers who already left. The brands that retain profitably are the ones reading the signals early enough to intervene.
Quick glossary: Expansion = an existing customer increasing their spend (higher AOV, new categories, subscription conversion, or increasing purchase frequency). Customer state machine = a model that assigns each customer to a lifecycle state (Active, At Risk, Lapsed, Win-Back, Churned) based on behavioral signals. Intervention window = the period between detectable behavior shift and revenue loss, when action is most cost-effective.
Metric 7: Leading Indicators of Expansion
What it measures: Behavioral signals that predict which existing customers will increase their spend within the next 60-90 days, across four expansion vectors.
The four expansion vectors:
1. AOV Expansion: The customer starts browsing higher-priced products or bundle pages. They add larger quantities. Their cart value trends upward across orders. A customer who bought a single item at $35 and is now browsing $65 bundles is telegraphing their next move.
2. Category Expansion: Cross-category browsing and purchasing. A customer who started in skincare and is now viewing supplements or haircare. First cross-category purchase typically happens within 90-180 days of activation for brands with 3+ product lines.
3. Subscription Conversion: One-time buyers clicking subscription toggles, viewing subscription landing pages, or engaging with subscribe-and-save emails. Subscription converts are worth 3-5x the LTV of one-time buyers, making this the highest-leverage expansion signal.
4. Frequency Acceleration: Purchase intervals shortening. A customer who bought every 45 days and is now buying every 30 days is expanding without changing what they buy. This signal is invisible in AOV-only reporting.
The compounding rule: When 3 or more of these signals appear within 60 days, conversion to high-value status runs 40-60%. A single signal might be noise. Multiple signals are intent.
Benchmarks:
- 20-30% of active customers trigger at least one expansion signal per quarter
- Expansion should account for 15-25% of quarterly revenue growth (not just new acquisition)
- Cross-category buyers have 2-3x higher LTV than single-category repeat buyers
- Subscription conversion from one-time: 10-20% within 6 months is strong
Where the data lives
Browse behavior sits in GA4 (or your CDP): product page views, collection views, add-to-cart events. Purchase data comes from Shopify. Subscription intent signals come from Recharge or Skio (subscription toggle clicks, save-page visits). Email engagement signals live in Klaviyo.
The individual platforms each see one dimension. GA4 knows what they browsed but not what they bought. Shopify knows what they bought but not what they almost bought. Klaviyo knows what emails they opened but not what pages they visited after clicking.
A warehouse (BigQuery + dbt) joins all three into a customer-level expansion score: browse signals + purchase trajectory + engagement behavior = predicted expansion probability. That score reverse-ETLs into Klaviyo as a custom property, triggering targeted upsell flows to the customers most likely to convert. This is Phase 3-4 work in the implementation roadmap.
Metric 8: Revenue Retention Predictors (Customer State Machine)
What it measures: A lifecycle classification that assigns every customer to one of five behavioral states, updated continuously, with transition rules that detect churn risk before it becomes churn.
The five states:
Active (Healthy): Purchased within 1.5x their expected purchase interval. Engaging with emails. Visiting the site. No intervention needed — focus on expansion.
At Risk: Purchase interval has stretched to 1.5-2.5x expected AND email engagement is declining (open rate trending down, fewer clicks). This is the intervention window. The customer hasn’t left yet, but the pattern has shifted.
Lapsed: No purchase in 2.5x+ their expected interval AND no site visits in 60+ days. They’ve stopped engaging but haven’t hit the churn threshold. Win-back is still viable but the window is closing.
Win-Back: Lapsed for 90-365 days BUT showing re-engagement signals (email open, site visit, ad click). These customers are giving you a second chance. The conversion rate on this window is 3-5x higher than cold reactivation of fully churned customers.
Churned: No purchase in 365+ days AND no engagement signals in 90+ days. The economics of reactivation are almost never positive at this stage. Stop spending money here and redirect budget to At Risk intervention.
What good looks like:
- Monthly lapse rate (Active → At Risk): Under 8-10%. Higher means your post-purchase experience has a leak.
- Monthly subscription churn: 4-6% is average (Recharge, 2024). Best-in-class subscription brands run 2-3%.
- Win-back conversion rate: 5-15% of Win-Back state customers converting back to Active within 90 days.
- At Risk save rate: 20-30% of At Risk customers should be recoverable with timely intervention.
Intervention economics:
The cost of saving an At Risk customer is a fraction of acquiring a new one. The math is simple:
- Email/SMS win-back flows (Klaviyo): Near-zero marginal cost. Automated flows triggered by state transition. Start here.
- Direct mail (PostPilot): 3-5x ROAS for lapsed customers. Physical mail cuts through digital fatigue. PostPilot integrates with Shopify and Klaviyo for automated triggers.
- Cancellation flow saves: For subscription brands, a well-designed cancellation flow (pause option, skip, downgrade, incentive) saves 15-30% of cancellation attempts. Recharge and Stay AI both support this natively.
- Targeted discount: For At Risk customers only. Not blanket discounting — that trains healthy customers to wait for deals.
The 365-day rule: Once a customer crosses 365 days without purchase or engagement, the expected win-back ROAS drops below 1x for most DTC categories. The money is better spent on At Risk and Win-Back state customers. This is counterintuitive — brands love running “we miss you” campaigns to their entire lapsed list. The state machine tells you which subset will actually respond.
Where the data lives
The customer state machine is inherently a warehouse model. It requires joining Shopify order history (purchase intervals, recency), Klaviyo engagement data (email opens, clicks, SMS responses), GA4 site visit data, and subscription platform status (active, paused, cancelled).
Each state assignment requires calculating the customer’s expected purchase interval (from their own history, not a category average), comparing current behavior against that interval, and checking engagement signals across channels. This is a dbt model that runs daily, updating each customer’s state and flagging transitions. The state then reverse-ETLs to Klaviyo as a custom property, powering automated intervention flows.
This is Phase 2-3 work. The basic version (purchase recency + email engagement) can be built in weeks. The full five-state model with cross-channel signals is Phase 3-4.
What These Two Metrics Give You
Expansion signals tell you where growth will come from inside your existing base — before you spend another dollar on acquisition. The customer state machine tells you who’s about to leave and what to do about it.
Together, they shift the retention conversation from reactive to predictive. Instead of running win-back campaigns on customers who already left, you’re intervening during the window when it’s cheapest and most effective.
Final part: Efficiency Metrics — how fast your flywheel spins and how efficiently you convert investment into growth.
