20-40% of SaaS churn is involuntary: failed payments that never get recovered. AI dunning recovers 30-50% of those, which often cuts total churn by 10-15% with no engineering work.
Here's an honest comparison of the leading dunning tools in 2026 (Stripe Smart Retries, Churnkey, and Butter Payments) with pricing, fit, and what to pick for your stack.
The TL;DR
- Stripe Smart Retries: Free, built into Stripe. Best for everyone on Stripe under $250K MRR. Turn it on today if you haven't.
- Churnkey: $200-1000+/mo. Best for SaaS at $250K-$5M MRR that wants the full retention stack (dunning + email + save flows).
- Butter Payments: Performance-based pricing. Best for high-volume subscription businesses that want payment recovery only, with pricing tied to results.
For most SaaS teams, the right answer is "Stripe Smart Retries first, layer in Churnkey or Butter when MRR justifies it."
How much revenue can dunning actually recover?
Before picking a tool, work out what the problem is worth. Most teams are surprised: a small percentage of MRR failing every month compounds into real money once you annualize it. Set the sliders to your numbers.
What are failed payments costing you?
Estimate the extra revenue better dunning would recover per year.
Where this number comes from. The math is deliberately simple so you can sanity-check it. Failing per month is your MRR times the share that fails (at $100K MRR and a 5% failure rate, that is $5,000 of revenue hitting a declined card every month). The extra you recover is that failed amount times the gap between your recovery rate today and your recovery rate with better dunning. Going from 20% to 55% recovery on $5,000 of monthly failures is $1,750 more recovered each month, or $21,000 a year, for a tool that costs a fraction of that. The 30-50 point recovery uplift is the range teams typically see moving from default fixed retries to network-timed AI retries, because the model retries when the card is actually likely to clear (after payday, after a bank hold lifts) instead of on a fixed calendar.
Stripe Smart Retries
What it does: Uses Stripe's network-wide transaction data to optimize the timing of failed payment retries. Instead of retrying on day 1, 3, 5, 7, it picks retry times based on what's most likely to succeed for that specific card type, bank, and failure reason.
Pricing: Free. Included with any Stripe account.
What's good:
- Free. One-day setup. Real ML model trained on billions of transactions.
- Network effect: it learns from every failed payment across the Stripe network
- No new vendor, no integration work, no contract
What's missing:
- No personalized email sequences (just retries the card)
- No payment update flow customization
- No save flow integration
- Limited reporting on recovery performance
Pick this if: You're on Stripe and not doing dunning yet. This is the no-brainer starting point. Don't pay for a more expensive tool until you've turned this on. Full head-to-head: Stripe Smart Retries vs Churnkey.
Churnkey
What it does: Full retention platform. Dunning (with smart retries) plus personalized recovery emails, cancellation save flows, and analytics. The "Bose noise-cancelling headphones" of SaaS retention: expensive but you get the whole package.
Pricing: Roughly $200-1000+/month depending on volume. Custom enterprise pricing above ~$5M MRR. See the full Churnkey pricing breakdown.
What's good:
- Personalized dunning emails (not just card retries)
- One-click payment update flows that work cross-device
- Integrated cancellation save flows (with dynamic reason-based offers)
- Strong analytics on what's actually recovering
- Built specifically for SaaS, not generic billing
What's not good:
- Flat monthly cost: if you're below ~$250K MRR, the price-to-value is shaky
- Implementation requires connecting your billing system, ESP, and product
- Some features overlap with what your ESP already does
Pick this if: You're at $250K-$5M MRR and want a single platform for dunning + email + save flows. The bundling is the value: you stop duct-taping Customer.io + Stripe + a custom save page together. If it is not the right fit, see Churnkey alternatives.
Butter Payments
What it does: Specialized payment recovery using ML to optimize retry timing. Pricing is performance-based: they take a cut of recovered revenue, so you only pay when it works.
Pricing: Performance-based (a % of recovered revenue). No flat monthly fee. Specific cut depends on volume and contract terms.
What's good:
- Performance-aligned pricing: you don't pay if it doesn't recover
- Stronger ML than Stripe Smart Retries for some failure types (especially high-volume subscriptions)
- Built for ecommerce subscriptions, not just SaaS
- No upfront commitment
What's missing:
- No save flow, no email personalization, no cancellation flow integration
- Just payment recovery: needs to be paired with other tools for full retention coverage
- Performance pricing can get expensive at scale (you're trading certainty of cost for performance alignment)
Pick this if: You're a high-volume subscription business (ecommerce, B2C subscriptions) where every percentage point of recovery matters and you want pricing tied to results. Pair with a separate tool for save flows and emails.
What about Recurly, Chargebee, and others?
Recurly and Chargebee are billing platforms with dunning features built in. They compete with Stripe directly, not with Churnkey or Butter. If you're already on one of them, their built-in dunning is roughly equivalent to Stripe Smart Retries.
If you're not on a dedicated billing platform yet, the choice between Stripe and Recurly/Chargebee comes down to feature needs beyond dunning. See our Stripe vs Recurly comparison and Stripe vs Chargebee comparison.
The decision framework
| Situation | Pick |
|---|---|
| Pre-revenue or under $50K MRR | Stripe Smart Retries. Don't pay for more yet. |
| $50K-$250K MRR, SaaS | Stripe Smart Retries + behavioral email in your ESP |
| $250K-$5M MRR, SaaS | Churnkey (you want the full save stack) |
| High-volume B2C subscriptions | Butter Payments (or Recurly's dunning) + separate save flow |
| $5M+ MRR, complex billing | Custom evaluation: Churnkey + Butter combined often wins |
What none of these tools do
None of these tools fix the root causes of involuntary churn (cards expiring, customers' bank limits). For that, layer in:
- Proactive card expiration outreach. Email customers 30 days before card expiry with a one-click update link. See the card expiration experiment.
- Account updater services. Visa and Mastercard offer card updater services that automatically update expired cards if the customer got a new one. Most billing platforms support this; few teams actually turn it on.
For the broader retention picture
Dunning is the highest-ROI churn fix because it requires no product changes and pays for itself in weeks. But it's just one piece. See the AI dunning guide for the full implementation playbook, what is involuntary churn for the concept, and where to start fixing churn for the broader sequence.
Score your retention stack
Take the 60-second Churn Health Check. It scores your involuntary churn setup (including whether you're using AI dunning) and tells you what to fix next.