Checkout Recovery

Reach for this journey when a meaningful share of started checkouts is abandoned without any payment error, and the funnel report shows the loss sitting between cart and order. Match checkout start, completed purchase and genuine payment failure events to the same user, cart and order, store push, email, WhatsApp and SMS consent separately, and define the high-value order threshold before anything goes live. If users leave at one specific step, for example when the shipping cost appears or a payment method is missing, repair that step first; a reminder only calls them back to the same obstacle.

Incremental checkout completion rate vs control

The journey's contribution is the extra share of unfinished orders that the reminded group completes compared with the control group; an opened checkout link is not a completion on its own.

How it is counted: Among all checkout attempts assigned at entry to the journey group or the control group, the share that end in a purchase within the same follow-up window from assignment; the gap between the groups is the journey's effect. Delivery or click status does not affect the denominator.

Also track

• Share handed to payment failureAttempts handed off at the first check because of a genuine payment failure, as a share of all entries; when it rises, the problem sits in the payment stack, not in the reminder, and belongs to that journey.
• Link opens without completionTrack users who open the checkout link but do not complete the order; a wide gap means they come back and get stuck in checkout again.
• Result by touch and order valueCount completions after the push, the email and the final touch separately; do not compare the direct channel on high-value orders with the final email on low-value ones, since those are different users.
• Complaints and consent withdrawalsWatch WhatsApp and SMS complaints, push opt-outs and email unsubscribes per channel; the direct channel is the most sensitive.

• Trigger and entryA checkout that starts without a completed purchase triggers the journey. At that moment, split eligible attempts at random: the journey group receives the reminders, the control group gets none for this checkout.
• Who must not enterUsers who completed the purchase and users with a failure confirmed by the payment system do not enter; failures belong to Payment Failure Recovery. A cart whose checkout has not started stays with Cart Reminder; the scope is one checkout attempt, and no second flow opens for the same attempt.
• Timing and channelsFirst check at 30 to 60 minutes: purchased, payment failed or still open. If open, a push, then 1 day later an email with items and checkout link, after 1 to 2 days WhatsApp (or SMS without WhatsApp permission) for high-value checkouts and a final email for the rest. Completion is rechecked before each touch.
• Exit and handoffOnce the purchase completes, the journey ends and queued reminders are cancelled. A genuine payment failure at the first check hands the user to Payment Failure Recovery and closes the old sequence; and when 1 day passes after the last touch without a purchase, the journey ends.
• Frequency and consentAt most three reminders per checkout, each channel used only with its own consent. The direct channel opens only for high-value checkouts: WhatsApp where permitted, otherwise SMS, which needs its own consent as well.

1. 1When the journey group completes no more checkouts than the control group, the reminders landed on users who would have returned on their own; link opens or clicks do not repair that picture. Rest the decision on the gap in completed orders between the two groups.
2. 2Your effect estimate is specific to this channel order, the 30 to 60 minute first wait and the high-value threshold you set. A later purchase by a user handed off for a payment failure is credited to Payment Failure Recovery; Cart Reminder hands over when checkout starts, so it never runs in parallel on the same items, and the sale is shown in one place only.
3. 3Test a small or unclear gap in the A/B test significance calculator with each group's checkout count and completed orders; if the p value shows no significance, do not declare the journey a success, keep the first wait fixed and repeat the measurement with more entries.
4. 4Complaints that reminders reached users who had already paid mean the matching of the purchase event to user, cart and order needs fixing before you read the result; every message sent after purchase also contaminates the measurement.