A health score is only worth acting on if it tells the truth about an account. The usual version watches product usage and calls it health. But an account can log in every day and still be slipping away, because its champion left, its sentiment soured, or its invoice has been overdue for two months. Airhop's health read is built to catch what usage alone cannot see.
Four signals, one read
Rather than a wall of separate metrics, Airhop folds the things that actually predict retention into a single read per account:
- Usage. Product events and activation milestones from Segment, so a drop in the actions that signal value registers early.
- Billing. Failed payments, overdue invoices, and upcoming renewals from Stripe and Measure, fed straight in.
- Sentiment. The tone of recent conversations and tickets, so a frustrated thread is not buried in a queue.
- Call notes. Context from Fireflies, so what was said on a call shapes the read, not just what was clicked.
Why revenue-aware changes the answer
Making billing a first-class input is the difference that gives the page its name. A large share of churn is involuntary and recoverable: a card expires, a payment fails, an invoice lapses. A usage-only model is blind to all of it. Because Airhop wires Stripe and Measure directly into health, a failed payment or an overdue invoice drops the score on its own, with no rule for you to write, and the at-risk account surfaces while it can still be saved. This is the engine behind churn prevention.
No scoring rules to maintain
Rule-based health scoring rots. Every edge case becomes another weight to tune and another branch to remember. Airhop reasons over the live picture instead. You are not maintaining a brittle if-then tree. The agent reads the combined signal and tells your CSM not just that an account is at risk but exactly why, which is the part a raw number never gives you.
From a read to a move
A health score that just sits on a dashboard is a measurement, not a save. When health drops, Airhop surfaces the account with the reason attached and, where you have allowed it, drafts the outreach to address it, ready for your approval. The read becomes a move. That handoff from signal to drafted action, governed by the approval layer, is what makes the health read part of an agent rather than another chart in agentic AI customer success.
Common questions
What feeds the health score?
Four signals fold into one read per account: product usage and activation from Segment, billing from Stripe and Measure, conversation sentiment, and call notes from Fireflies. Billing is first-class, so a failed payment or overdue invoice moves health on its own.
Why call it revenue-aware?
Because billing signals feed the score directly. Most health scoring looks only at usage, which can hide an account that is active but sixty days overdue. Folding revenue in means the health read reflects real money, not just clicks.
Do I have to write scoring rules?
No. A failed payment or overdue invoice drops health without a rule for you to author. The agent reasons over the live picture rather than running a brittle if-then tree you have to maintain.
What happens when health drops?
The agent surfaces the slipping account with the reason attached and, where you have allowed it, drafts the outreach to address it for your approval.