App Channel Recovery

Build this one when app analytics show a growing group of accounts that are still active but have had no session in 30 days, and you can still reach part of that group by email, SMS or WhatsApp. Before launch, check that session events are tied to the same user identity, that permission and deliverability status for each channel is current, and that the message deep-links to the relevant action in the app, not the home screen. If sessions dropped off after an update or a sign-in problem, fix that first; no message brings people back to an app that does not work.

Incremental app return rate vs control

The only proof the message works is that the share reopening the app in the journey group beats the control group; deliveries or clicks on their own prove nothing.

How it is counted: Among all users assigned to both groups at entry, the share who reopen the app within the same follow-up window from assignment (the 7-day wait in the graph); the gap between the groups is the journey's effect. Which channel carried the message, and whether it was delivered or clicked, does not change the denominator.

Also track

• Meaningful action after returnOf those who reopen the app, the share completing the action the message pointed to; one open does not mean the habit is back, so track opens and actions separately.
• Share reached through a fallback channelAcross all entrants, the share reached by SMS or WhatsApp instead of email, and the share reached by nothing; it shows how incomplete your email data is.
• Return by channelRead return rates for email, SMS and WhatsApp separately, but remember the user's data decides which channel is used, so do not read the gap as channel strength.
• Unsubscribes and complaintsWatch email unsubscribes and SMS or WhatsApp complaints per channel; if even a single message to someone silent for 30 days costs permission, revisit the copy and the destination.

• Trigger and entryA user enters when no app session has been recorded for them in the past 30 days. Right away, split them at random between the journey group and a control group that gets no message from this attempt; one recovery attempt opens per user.
• Who must not enterUsers with a closed account, or any session in the last 30 days, do not enter. Someone with no permitted, deliverable channel outside the app drops to the No reachable channel exit at entry and receives nothing. While an attempt is open, no second attempt opens for the same user.
• Timing and channelsEmail deliverability is checked first; if it passes, the email goes. If not, SMS is checked, and failing that, WhatsApp permission and deliverability. One message from the single chosen channel deep-links to the relevant app action, followed by a 7-day wait and no further touch.
• Exit and handoffOnce the wait ends, an app open closes the journey through the App channel recovered exit, and no open closes it through Recovery attempt ended. A user reachable on no channel leaves through No reachable channel. No handoff follows: the journey ends in its own exits, and silence never starts another channel.
• Frequency and consentEach user gets a single message, and email, SMS or WhatsApp is used only where that channel is permitted and deliverable. Fallback channels replace one another; they never run at the same time.

1. 1If the return gap is zero or favors the control group, the message landed on users who would have come back anyway; clicks and deliveries do not change that reading. Thirty days without a session does not prove an uninstall either, so do not write off the non-returners or start adding channels.
2. 2The reading applies to this channel priority (email first, then SMS, then WhatsApp), one message and the 7-day wait. With no handoff, a return is credited only to this journey; do not fold in the outcome of whatever flow opens inside the app after the return.
3. 3A return gap resting on a handful of users is not a result yet: put each group's assigned users and app reopeners into the A/B test significance calculator, and if it does not reach significance, let more users cross the 30-day threshold and enter before you touch the channel order or the copy.
4. 4Adding a second channel for users who stayed silent, because the return rate looks low, breaks this journey's rule: a fallback channel is chosen before the message goes, only when the one before it is unreachable, and no response never opens a new one. Write the message around observed inactivity, not an assumed uninstall.