SMS Dunning: Baremetrics' 2026 Data Says the Best Message in Your Sequence Isn't an Email
Baremetrics analyzed 123,810 dunning emails and found SMS open rates beat email 90%+ to 20-30% by day 7. Here's how to add it without breaking TCPA.
Most dunning advice stops at email: three or four messages, an increasingly urgent subject line, a link to update the card. That's the whole playbook in most Stripe dunning guides, ours included. What it usually skips is what happens to open rates after the first email — and a 2026 benchmark report from Baremetrics, built from actual recovery data rather than survey responses, puts a number on exactly how much that costs.
That gap doesn't show up on day one. It shows up exactly where a typical dunning sequence is weakest: the later retries, after the customer has already ignored one or two payment-failed emails and your open rate has quietly collapsed.
What Baremetrics actually measured
The report tracked 119 US B2B SaaS companies running Baremetrics Recover through May 2026 — 123,810 dunning emails sent, 10,999 charges recovered, $1,236,764 in revenue pulled back from failed payments in that single month. The median company saw a 12.7% attempted-recovery rate and an 808% median per-customer ROI, with 95% of customers hitting payback on their dunning spend within the first month. This isn't a survey of what SaaS founders believe works. It's what actually cleared, measured against real invoices.
The detail that matters most for sequencing: the report identifies the day-13 message as typically the single most valuable message in a subscription dunning sequence — not the first email, which most guides treat as the important one, and not a later message either. Day 13 sits deep enough into the retry window that easy recoveries (a temporary insufficient-funds bounce, a one-off card glitch) have already resolved themselves, but early enough that the customer hasn't yet mentally written off the subscription.
Why email alone degrades by day 7
The mechanism isn't mysterious once you look at it. A first payment-failed email lands with genuine surprise value — the customer didn't expect it, so they open it. By the third or fourth email on the same topic, you're competing with an inbox that's started auto-filing the thread, a spam folder that's grown suspicious of repeated sends from the same subject line pattern, and simple fatigue. We've written separately about the DNS-level reasons some dunning emails never arrive at all — SPF, DKIM, DMARC enforcement — but even a perfectly authenticated domain can't fix a customer who's stopped opening a message they've already seen twice.
Source: Baremetrics, Subscription Payment Recovery Benchmarks (2026). Email figures approximate the reported day-0 (41.29%) and day-30 (26.83%) open rates; SMS reflects the reported 90%+ figure for later cadence steps.
Notice what this chart argues against: sending SMS from day zero. A text on the first failure treats a routine, often self-correcting glitch — a card that expired mid-month, a bank hold that clears in a day — with the urgency of a final notice. Baremetrics found the accounts getting the best returns used SMS specifically for the day-7-and-beyond steps, while treating everything before that as email-only. That's a deliberate sequencing choice, not a blanket "text everyone every time" policy.
The TCPA line you can't cross
Before wiring up SMS, it's worth being precise about what kind of message you're sending, because US telecom law treats two categories differently. A payment-failure notice — what happened, the amount, a link to fix it — is a transactional message, and under the TCPA that only requires prior express consent, which can be given orally or in writing. A message that includes a discount, a retention offer, or anything promotional gets reclassified as marketing, which needs the stricter prior express written consent. Standard TCPA violations run $500 per message; willful ones run $1,500 per message — numbers that add up fast if your dunning SMS accidentally drifts into marketing copy.
| Channel | Typical open rate | Consent needed | Cost per send |
|---|---|---|---|
| Email (day 0-4) | 40%+ | Existing customer relationship | Near zero |
| Email (day 7+) | 20-30% | Existing customer relationship | Near zero |
| SMS (transactional only) | 90%+ | Prior express consent (oral or written) | ~$0.012-0.013/msg |
| SMS (with any offer/discount) | 90%+ | Prior express written consent | ~$0.012-0.013/msg |
| In-app banner | Depends on login frequency | None — already logged in | Near zero |
Sources: Baremetrics (2026), Twilio US SMS pricing, FCC TCPA guidance on transactional vs. marketing messages.
The practical fix is to capture SMS consent once, cleanly, at a point unrelated to any specific billing event — during signup, or in an account settings page with its own explicit "text me about billing and account issues" checkbox. Don't repurpose a phone number collected for two-factor authentication; that consent was given for a different purpose and a plaintiff's attorney will make exactly that argument if it ever comes to it. And keep the message itself boring on purpose: amount, reason if you know it, one link. The moment you add "here's 20% off to stay" to a payment-recovery text, you've turned a cheap, low-consent-bar message into one that needs a standard you probably haven't cleared.
Building the sequence
Layering SMS onto an existing dunning schedule doesn't mean rebuilding it. It means adding two touchpoints to whatever retry schedule you're already running:
- Day 0-1: Email only. Soft tone, assumes it's an accident, direct link to the Stripe customer portal.
- Day 4-7: Email, plus your first SMS if the customer has opted in. This is where email's open rate has already started sliding and SMS starts pulling its weight.
- Day 13: Email and SMS together. Per Baremetrics, this is the single highest-value message in the whole sequence — worth writing deliberately rather than letting it be an auto-generated copy of the day-1 email with a different date.
- Day 21-28: Final notice on both channels, with a firm date for what happens to the account — this is also the point to consider pausing rather than cancelling outright, since a paused subscription is cheaper to win back than one that's fully lapsed.
None of this replaces the underlying retry logic — Smart Retries picking sensible payment attempt times still matters independently of when you notify the customer. What it adds is a second channel for the exact window where email's usefulness is already declining, instead of running four increasingly ignored emails and calling the sequence complete.
Is it worth building?
For most SaaS products already running any dunning sequence, yes, and the math isn't close. At roughly a cent per message, even a modest lift in day-13 recovery covers the SMS cost many times over — which is consistent with the 808% median ROI Baremetrics reported, and with 95% of the companies in its sample recovering their dunning spend within a single month. The exceptions are products with genuinely low failed-payment volume, where the fixed cost of building consent capture and a second send pipeline isn't worth it yet, or products where customers reasonably don't expect to be texted by a B2B tool at all — some enterprise buyers will read an SMS as an odd channel for a vendor relationship, so read your own customer base before assuming this generalizes.
Involuntary churn from failed payments still accounts for a large share of total cancellations at most SaaS businesses, and almost none of it reflects a customer who wanted to leave — recovering it is close to free money compared to the cost of replacing that revenue with a new sale. Run your current recovery rate through our churn calculator to see what a few extra points of dunning recovery are actually worth before you decide whether SMS earns its place in the sequence. And for the payments that don't recover — the customers who genuinely do want to leave, not the ones who just missed an email — that's exactly the moment CancelFlow is built to catch, turning a silent cancellation into a recorded reason and, often, a saved subscriber.
Frequently asked questions
Do I need separate consent to text a customer about a failed payment?+
Yes, but the bar is lower than for marketing texts. Under the TCPA, transactional and informational messages — which a payment-failure notice qualifies as, provided it stays free of promotional content — only need prior express consent, which can be given orally or in writing. Marketing SMS needs prior express written consent, a stricter standard. The safest way to clear even the lower bar is to capture phone-number consent explicitly at signup or in account settings, worded for billing and account notifications specifically, rather than reusing a number a customer gave you for an unrelated purpose like two-factor authentication.
What's the best day to send an SMS in a dunning sequence?+
Not day 0. Baremetrics' 2026 recovery data found the highest-performing accounts reserved SMS for day 7 and later in the sequence, while treating the first failure as email-only. Email still recovers well early — open rates run 40%+ on the initial notice — and texting a customer the moment a card fails reads as aggressive for what's often a one-off glitch. SMS earns its cost once email engagement has already dropped, which is where its 90%+ open rate does the most work recovering payments email alone would lose.
Is SMS dunning worth it for a small SaaS product?+
Almost always, on unit economics alone. Twilio's published US rate is $0.0083 per SMS segment, with carrier pass-through fees typically bringing the effective cost to around $0.012–0.013. Even at the high end, recovering a $30/month subscription with one extra text message is a trivial cost against the revenue saved — which is consistent with the 808% median per-customer ROI Baremetrics reported across the 119 companies in its 2026 sample. The cost only becomes worth worrying about at very high failure volumes, and even then it's dwarfed by the MRR a single recovered subscriber represents.
Can I include a discount or retention offer in a dunning SMS?+
Not without changing its legal classification. The FCC and TCPA guidance treat a message as promotional the moment it includes marketing content — a discount code is the textbook example — which moves it out of the lower-consent transactional bucket and into the stricter prior-express-written-consent requirement for marketing texts. Keep dunning SMS to the failure itself: what happened, the amount, and a direct link to update payment. If you want to offer a pause or discount to reduce voluntary churn, that belongs in your cancellation flow once the account is already at risk of leaving, not folded into a payment-recovery text.
Stop losing subscribers today
One script tag. One function call. A live cancellation flow in under 10 minutes.
Start free trial →