`ripllo.abandoned_cart.recovered.v1`

Fires when a previously-abandoned cart is recovered — the buyer came back and completed a checkout that originated from a reminded cart. The primary metric for "did the recovery email pay off".

Reserved — not currently emitted in v1. markRecovered stamps recoveredAt on the reminder row but writes no outbox row, so nothing is dispatched to your endpoints. Subscribe defensively; for now, poll GET /abandoned-cart/stats or GET /abandoned-cart/reminders (rows with a non-null recoveredAt), or react to your own /recover call's {recovered: true} response.

When it will fire

When POST /api/v1/abandoned-cart/recover sets recoveredAt on a reminder row.

/recover is not idempotent on (accountId, checkoutSessionId). It matches the most recent unrecovered reminder for (accountId, customerId) whose sentAt is inside a hard-coded 72-hour window, and only stores checkoutSessionId on the row afterwards. A retried partner-platform webhook carrying the same checkoutSessionId therefore marks the next unrecovered reminder for that customer recovered as well, inflating the recovery count — and, once the event ships, would emit twice. Dedupe on the partner side: call /recover once per completed checkout.

If the partner platform calls /recover for a checkout that has no matching reminder (the cart was never abandoned, never reminded, or the reminder is older than 72 hours), nothing is stamped — the recovery endpoint responds with recovered: false and Ripllo is silent.

Payload

The intended shape, for when this ships (ids are opaque cuids — AbandonedCartReminder has no status column, and no delivery state is tracked):

{
  "id": "clx2k9f0a0002v8pq5w1kghi",
  "type": "ripllo.abandoned_cart.recovered.v1",
  "createdAt": "2026-05-13T10:43:22.187Z",
  "accountId": "acc_01HX...",
  "data": {
    "reminder": {
      "id": "clx2k9f0a0000v8pq7h3m1abc",
      "accountId": "acc_01HX...",
      "customerId": "cus_storlaunch_4471",
      "cartId": "cart_storlaunch_19823",
      "email": "alice@example.com",
      "valueAtSend": 250000,
      "currencyAtSend": "IDR",
      "discountCodeId": "clx2k9f0a0003v8pqz0p4jkl",
      "sentAt": "2026-05-13T06:42:00.000Z",
      "recoveredAt": "2026-05-13T10:43:22.140Z",
      "recoveredBySessionId": "cs_storlaunch_01HX...",
      "externalSource": "storlaunch",
      "externalRef": "cart_storlaunch_19823"
    },
    "checkoutSessionId": "cs_storlaunch_01HX...",
    "timeToRecoveryMs": 14482140
  }
}

The timeToRecoveryMs is a convenience: the milliseconds between sentAt and recoveredAt. Useful for histograms without doing date math in your handler.

Handler examples

// Node
if (event.type === 'ripllo.abandoned_cart.recovered.v1') {
  const { reminder, checkoutSessionId, timeToRecoveryMs } = event.data;
  analytics.track('cart_recovered', {
    customerId: reminder.customerId,
    valueIdr: reminder.valueAtSend,
    discountUsed: reminder.discountCodeId !== null,
    hoursToRecovery: Math.round(timeToRecoveryMs / 3_600_000),
    checkoutSessionId,
  });
}

# Python
if event['type'] == 'ripllo.abandoned_cart.recovered.v1':
    reminder = event['data']['reminder']
    analytics.track('cart_recovered',
        customer_id=reminder['customerId'],
        value_idr=reminder['valueAtSend'],
        discount_used=reminder['discountCodeId'] is not None,
        hours_to_recovery=round(event['data']['timeToRecoveryMs'] / 3_600_000),
    )

// Go
if event.Type == "ripllo.abandoned_cart.recovered.v1" {
    var data struct {
        Reminder            ripllo.AbandonedCartReminder `json:"reminder"`
        CheckoutSessionID   string                       `json:"checkoutSessionId"`
        TimeToRecoveryMs    int64                        `json:"timeToRecoveryMs"`
    }
    _ = json.Unmarshal(event.Data, &data)
    analytics.Track("cart_recovered", data.Reminder.CustomerID, data.Reminder.ValueAtSend, data.TimeToRecoveryMs)
}

What to do

Update recovery analytics. Once the event ships it's the push counterpart to GET /abandoned-cart/stats, which is where recovery rate comes from today. Aggregate by discountCodeId !== null to compare bundled-discount vs no-discount recovery.
Optionally thank the customer. Sending a "thanks for coming back!" follow-up can deepen engagement, but tread carefully — too many emails will provoke an unsubscribe.
Compute true recovery value. Compare reminder.valueAtSend (cart size at abandonment) to the actual checkout total (from the partner platform's order data) — the buyer may have added or removed items between abandonment and recovery.
Stop any "second reminder" sequences. If you've layered automations on top of Ripllo's first reminder, this is the cancel signal.

Common pitfalls

Counting valueAtSend as recovered revenue. It's the cart-snapshot value at abandonment time, not the final checkout total. The two diverge if the buyer added or removed items. For revenue reporting, use the partner platform's order-completed data and cross-reference by checkoutSessionId.
Treating recovery as instant. The buyer might come back days later. timeToRecoveryMs can easily span hours or days; bucket appropriately when reporting.
Assuming the reminder caused the recovery. Attribution is "the buyer was reminded and then completed a checkout". Some of those buyers would have come back anyway. The honest measure is incremental recovery rate vs a holdout group, which Ripllo doesn't currently A/B-test for you — you'd need to design that on the partner-platform side.
Sending the recovery event back to the partner. This event is for you (the integration owner) to track. Don't bounce it back to the partner platform — the partner already knows the checkout completed.

ripllo.abandoned_cart.reminder_sent.v1 — reserved-not-emitted; the "we sent the email" counterpart.
ripllo.abandoned_cart.suppressed.v1 — reserved-not-emitted; buyer opted out.
ripllo.discount_code.redeemed.v1 — if the recovery used the bundled discount code, this event fires too.

Abandoned cart resource — the full config + recovery API.
Webhooks reference — envelope, retries, signature verification.

ripllo.abandoned_cart.recovered.v1