Channel choice is where recovery programs stop being generic and start being tuned. Email is universal; SMS is expensive but immediate; push is free but only reaches your active users. The best recovery cadences use all three deliberately, not interchangeably.
# The channel comparison
| Metric | SMS | Push | |
|---|---|---|---|
| Cost per send | ~$0.001 | $0.01–$0.10 | $0 |
| Open / delivery rate | 40–60% open (post-Apple MPP inflation) | 95%+ delivery, 90%+ read within 15 min | 20–35% seen |
| CTR on primary CTA | 3–8% | 15–35% | 8–20% |
| Compliance burden | DMARC, unsubscribe | TCPA, opt-in proof, quiet hours | App permission |
| Reach | Anyone with an email | Only opted-in phone numbers | Only active app installs |
| Best for | Bulk of the cadence | Time-sensitive, high-value | Users mid-session |
The costs are order-of-magnitude, not competitive. SMS is 10–100x the cost of email per send. Push is free but has no reach outside your app installs. Any comparison that ignores cost or reach is oversimplified.
# Email: the workhorse
Email is 60–80% of most recovery cadences by volume, and it should be. Reasons:
- Universal reach — every subscriber has an email.
- Cheap enough to send freely.
- Rich formatting for context (invoice details, product summary, one-click CTA).
- Regulated but well-understood (GDPR, CAN-SPAM, CASL).
Weakness: inbox placement is not guaranteed. Even a perfectly authenticated domain lands in Promotions or Spam some fraction of the time. Which is why email cannot be the only channel.
# SMS: the accelerator
SMS wins when time matters. Failure day 4, email #2 not opened, WhatsApp not available → SMS is the right escalation.
Rules for SMS in recovery:
- Opt-in with a paper trail. TCPA in the US requires express written consent. Store timestamp, IP, and the exact opt-in copy shown.
- Quiet hours enforced in the customer's timezone. No sends between 9pm and 8am, regardless of your ops schedule.
- A single template per failure cycle. Never chain two SMS reminders — it looks desperate and drives complaints.
- Include the brand and a short URL. "NETFLIX: Your card failed. Update: nflx.co/x — Reply STOP to opt out."
- Track STOP replies as suppression events — cross-channel suppression, not just SMS. A customer who STOPs on SMS should not receive email marketing either.
SMS conversion on payment recovery is 2–4x email in isolation, but the addressable audience is smaller. Overall program impact is usually 5–15% of recovered revenue.
# Push: the mid-session nudge
Push notifications reach the user only if they have your app installed and permissions granted. That is a smaller pool, but it is a pool with high engagement — these users are your best customers.
Push is optimal for:
- In-session recovery — "Your subscription is at risk. Update card [now]." shown when the user opens the app.
- Deep links to the update flow — no browser hop, no re-login.
- Quiet nudges — a persistent banner, not a modal, on the failure state.
Push is not a broadcast channel for recovery. Sending push at 3am wakes the user's phone and burns permission for future notifications. Send push only in the customer's local waking hours, and only if you have a reason to believe they're active.
# In-app: technically not a channel, always the highest-converting
The single best-converting "channel" is not one of the three above — it's an in-app modal on next login. Rules:
- Fire on the first authenticated session after the failure.
- Non-blocking (dismissible) after showing once.
- Deep link to the card update flow.
- Do not repeat more than once per 24 hours.
In-app conversion on payment recovery is 30–60% — higher than every other channel by a wide margin. Every recovery cadence should include an in-app step; the question is where in the sequence.
# Orchestration: the RRLabs default cadence
For a B2C consumer subscription in a card-first market:
| Hour | Channel | Purpose |
|---|---|---|
| 0 | (retry) | Silent smart retry |
| 2 | Email #1 | First notification |
| 24 | In-app | If user logs in |
| 48 | Email #2 | Different subject line |
| 72 | Push | If app installed |
| 96 | SMS | If opted in and high-value |
| 120 | (retry) | Second smart retry |
| 168 | Email #3 + In-app | Final notification |
| 168 | Grace ends | Reactivation flow |
Every step is conditional. Users who paid at hour 3 do not receive email #2. Users without SMS opt-in skip the SMS step. Cadence is not a script; it is a state machine.
# Compliance quick reference
- Email: DMARC required by 2024+ for bulk senders. One-click unsubscribe required.
- SMS US: TCPA — express written consent, quiet hours, opt-out via STOP.
- SMS EU: Consent, quiet hours, opt-out via STOP. GDPR applies.
- SMS Brazil/India: register the sender ID with regulators before sending.
- Push: user permission required per platform (iOS, Android). Do not spam or the OS revokes permission.
- WhatsApp: opt-in required, utility templates only for business-initiated (see the WhatsApp article).
Every channel has a compliance floor. Under-doing it triggers regulatory action; over-doing it triggers customer complaints. The sweet spot is boring and documented.
# The RRLabs default
The Revenue Recovery Labs engine orchestrates all four channels with per-customer state — opt-in status per channel, app-install status, quiet hours in local timezone, cross-channel suppression on STOP or unsubscribe. Cadence is data, not code. Every send is logged with channel, template, and outcome. That's what turns "we send emails and sometimes texts" into a program you can improve every quarter.