Your AI, with 13 years of
churn measurement behind it
Read only · Included on every plan, including free
The Churn Buster MCP connects Claude, ChatGPT, and other AI apps to your Dunning and Cancel Flows data. It applies our measurement method so your team can check what changed before changing a campaign.
Better measurement. Better advice.
A falling number can send your AI in the wrong direction. The MCP helps it check the evidence before it recommends a change.
Without MCP
Example conversationRecovery fell from 44% to 27% in September. Should we change our retry schedule?
Yes, you're right to flag this. September is down 17 points from August. That's a sharp break from a steady summer, and it suggests your current retry schedule is no longer working.
Start with more frequent retries. Tighten the schedule to four attempts in the first week so you can recover payments sooner. There's no reason to wait another month when the decline is this clear.
Refresh your emails at the same time. A/B test a new subject line on the first email and make the payment link more prominent. These changes should help reverse the drop, and next month's rate will tell you whether they're working.
Confident answer. Wrong premise. September is still in progress. This advice treats unfinished campaigns as losses and changes two things before checking either.
With MCP
Example conversationRecovery fell from 44% to 27% in September. Should we change our retry schedule?
- Checked complete months, recovery sources, and your retry changes.
No change needed yet. Recovery is steady. September isn't finished, so its 27% isn't a final result.
| Check | Result |
|---|---|
| Complete months, Apr–Aug | 41–46% recovery |
| September campaigns still open | About 610 |
| Recoveries from retries | About 60% |
| Retry schedule changed | Aug 12 |
Give the new schedule 60+ days of complete cohorts, then compare it to the summer. Your team can review that comparison with Churn Buster.
Without MCP
Example conversationOur 50% discount saved 9 of 14 customers. Should we make it the default offer?
Yes, the discount is your clear winner. Nine saves from 14 customers gives you a 64% acceptance rate, more than double the pause offer's 28%. Your instinct to roll it out is backed by the numbers.
Make it the default for every cancel reason. Put the discount first so every customer sees your strongest offer. Keep the pause option as a fallback for anyone who declines.
Check the results next month. With the discount in front of more customers, you should see more saves. If the acceptance rate stays above pause, you'll have confirmation that the rollout worked.
Too little data. Too much discount. Fourteen presentations can't establish a winner. A blanket discount also cuts revenue from customers who might stay for less.
With MCP
Example conversationOur 50% discount saved 9 of 14 customers. Should we make it the default offer?
- Checked offer acceptance by cancel reason and overall save rate.
Keep your current offers for now. The discount needs more data before you make it the default.
| Offer | Shown | Accepted |
|---|---|---|
| 50% discount | 14 | 64% |
| Pause | 212 | 28% |
Your save rate held at 31% as cancel volume rose.
Let the discount reach roughly 30 presentations within its cancel reason, then compare. The Churn Buster team can help you set up that test.
Know what changed before you change anything
Complete cohorts only
Only finished campaigns count. An unfinished month can look worse than it is.
Read moreYour normal range first
It compares results with your best and worst complete periods before calling a change.
Read moreEvery outcome counted
Card updates, retries, cancellations, and passive churn all count. Recovery counts regardless of which tool retried the card.
Read moreContext before conclusions
A platform move or a change in who signs up can shift results. It asks about those changes before blaming a campaign.
Read moreMeasure the dunning tool you run today
Keep your current dunning tool. The free Measure plan connects to supported billing platforms and measures its results. MCP access comes with it.
Set up in three steps
-
01
Turn it on
The account owner enables the MCP in Settings → Integrations → MCP.
Read the setup guide → -
02
Add the server URL
Paste
https://mcp.churnbuster.io/mcpinto your AI app. -
03
Sign in and ask
Approve read only access, then ask your first question.
Questions about the MCP
Which AI apps does it work with?
Claude, ChatGPT, Claude Code, Codex, and other apps that support remote MCP servers. See supported plans and steps in the setup guide.
What does it cost?
It's included with every Churn Buster account at no extra cost, including the free Measure plan. See pricing.
Does it work if another tool runs our dunning?
Yes. The free Measure plan measures your current dunning tool on supported billing platforms. The MCP reports on those results. Cancel flow answers need Churn Buster Cancel Flows. Book a call to talk through your setup.
Can it change anything in our account?
No. It's read only. It can't edit campaigns, settings, or customers, and the account owner can turn it off for everyone at any time.
What data can it see?
Aggregate rates, counts, and trends for accounts that have the MCP turned on. It can't see individual customers or campaigns, email or SMS content, decline codes, or test mode data. We keep a log of MCP requests for about 30 days.
Can it tell us what to change?
It shows what happened and whether a change is outside your normal range. For help deciding what to test next, our team can review your results with you.
Get the measurement right, then optimize
Start with the free Measure plan and ask about your results.