For the people who need to approve it

Start with the sources your CS lead can grant alone

Point Keply at a read-only connection or one file with a status column, and it scores the book from that in the first session. What you get back is not another dashboard. Keply reads the sentiment and the silence in every thread, weighs both against your churn history, and every account it raises comes with the move already made: the evidence behind the call, and the outreach drafted for a person to approve. The more it sees the sharper the read, so tickets, usage and email tighten the picture as they arrive. Nothing waits on another team’s backlog.

You could build the dashboard. That’s the easy half

Teams do build the part that sees, and it works. What costs is the half after: the nuances you find in month three, and something that still runs when the person who built it gets busy. That second half is what you are buying, and it is four specific things.

• It acts, it doesn't only see. Every account Keply raises arrives with the internal moves that come first and the outreach already drafted, for a person to approve. A dashboard hands you a list and the work starts there.
• It reads your own losses. Import the accounts you lost and Keply derives the silence threshold that means trouble at your company rather than a default, names the pattern in those losses, and prints a confidence tier beside it. Under 10 usable churn events it claims no calibration at all and says so.
• Nobody administers it. Your success manager runs it in their own words. No CS-ops hire, no configuration project, no weights workshop.
• It stays working. Connectors, models and scoring are maintained without a standing claim on your roadmap. An internal build competes for that room every quarter.

What you’ll want to know before you approve it

Answered here, in public. No NDA, no form, no call to book before you can read them.

• Yes, Keply reads the text of the messages you connect. Sentiment and risk cannot be read from a subject line. That text goes to Anthropic for inference under commercial terms that prohibit training on your data, written into the Data Processing Addendum you sign.
• Attachments are never opened. Never downloaded, never parsed. Keply sees the file name and flags it.
• Nothing reaches a customer without approval. Keply can send from your mailbox once you allow it, and can write four fields back to HubSpot if you switch that on, off by default. The autonomy dial starts at draft-for-approval and every action is logged.
• Keply does not hold a SOC 2 report. Controls are self-assessed against the AICPA Trust Services Criteria, in a control matrix and policy set we share on request, without an NDA.
• If something goes wrong we run a written incident response plan including the GDPR 72-hour breach clock, and a watchdog over our own background jobs.
• Your export is self-serve any time, no notice period. Leaving is a cancellation, not a migration.

The same job, at a tenth of the bill

$299 a month, flat

Mid-market teams solve this with a platform. ChurnZero does not publish pricing, and third-party trackers report a median around $36,000 a year. Keply does the same job for $299 a month, and the bill stays flat as the team and the book grow. One workspace, unlimited seats and accounts. One month free, no card. 30-day money-back guarantee, cancel anytime.

No admin. Your success manager runs it: no CS-ops hire, no configuration project.