Retention overview
Everything below is read from tools you already run. Nothing here is stored anywhere new — Berta connects to your existing stack and layers a risk model and decision logic on top.
Revenue at risk
$48,600
via Shopify + Stripe, scored by Berta
Recovered, 30 days
$16,240
via Stripe order confirmations
Triggers, 7 days
86
via Berta trigger log
Win-back rate Conversion = first Shopify order within 14 days of trigger, confirmed via Stripe. Excludes customers in marketing suppression. Industry baseline is 5–8%.
34%
via Klaviyo opens + Stripe orders
This is pattern detection across your whole customer base — no single customer here looks unusual on their own. It's only visible in aggregate.
Customers by risk segment
This is Berta's model prediction, not a fixed rule like "no order in 30 days." Each customer's cadence and signals are weighed individually.
Marketing suppression status
At-risk customers
Click any customer to see exactly which signals are driving their score, and where each one came from.
| Customer | LTV | Risk | Signal |
|---|
Recent triggers
Berta makes the decision. Delivery always happens through your existing tools — Klaviyo, your helpdesk, or a human on your team.
All customers
Jordan Mensah
Consumables · LTV $860
Risk score, last 8 weeks
Recalculated daily from every source below, not a one-time label.
Signals driving score
Click a signal to see the real event behind it, not just a weight.
Weights are model-learned from your category's purchase history — not manually configured.
Order history · via Shopify
| Date | Amount | Channel |
|---|
Cadence model
Engagement · via Klaviyo
Support · via Gorgias
Decision and delivery
The left box is Berta's model. The right box is whichever tool actually did the sending — that split is deliberate, so nothing about who sent what is ambiguous.
Marketing suppression
Risk spike · July 12 shipping fee change
Customers affected
312
Revenue at risk
$48,600
Normal daily rate
5 / day
Shared signal across the cohort
No corresponding rise in support tickets, which points to a pricing reaction rather than a service failure.
Recommended action
Automated win-back for this cohort
Paused in Klaviyo so a blanket discount doesn't mask a pricing problem
Escalate to pricing review
Sends the spike timeline, affected customer list, and cart abandonment data to your team via Slack #retention-alerts
Sample of affected customers
| Customer | LTV | Risk | Days since fee change |
|---|---|---|---|
| Alina Costa | $1,340 | 67 | 4 |
| Marcus Webb | $610 | 59 | 3 |
| Devon Price | $920 | 64 | 4 |
Reporting
Cohort retention curves
% of customers still active at 30 / 60 / 90 / 120 days after first order
Revenue recovered per month
Via Stripe order confirmations, within 14 days of trigger
Conversion = first Shopify order within 14 days of trigger, confirmed via Stripe. 34% of triggered saves converted this month.
Win-back rate by segment
% of triggered customers who placed an order within 14 days
High-ticket is human-only outreach — smaller volume, longer decision cycle. Consumables benefit most from cadence-aware timing.
Average LTV by acquisition channel
Retained customers · via Shopify UTM + Stripe
| Channel | Avg LTV | Retention at 90d |
|---|---|---|
| Referral | $2,100 | 67% |
| Organic search | $1,580 | 61% |
| Paid social | $1,240 | 54% |
| Email campaign | $980 | 49% |
Referral customers retain at 2.2× the rate of email acquisition. Win-back budget goes furthest protecting referral-sourced cohorts.
Retention policy
Risk thresholds
At-risk starts at
Customers above this score move from healthy to at-risk
At 50, this marks 2,214 customers as at-risk.
High-risk starts at
Crossing this score is what fires an automated trigger
At 75, this marks 1,207 customers as high-risk and fires automated triggers.
Automation cutoff
Route to a human above
Automated offers stop here. Priya Desai, at $4,200 LTV, is routed to an account manager instead of a discount because of this rule.
At $2,000, this routes 23 customers (LTV $2,000–$8,400) to a human instead of an automated offer.
Offer rules by segment
Cohort safety rule
Pause automation on a risk spike
This is the rule that held the July 12 shipping-fee cohort out of automated win-back
Trigger when more than
customers cross threshold within this window
Current setting would have caught the July 12 spike (312 customers / 48h). Well-calibrated for this account size.