AI Health Scoring
Score account health by describing what healthy, at risk, and growth opportunity look like in your own words.
How it works
In legacy customer-success tools, health scoring is a formula. You pick metrics, weights, and thresholds, and the platform multiplies things together. The output is a number with no explanation.
Cust takes a different approach. You write what each health status means in plain English. Cust runs an agent over every customer on a schedule and important events (Signals) and uses your Health scoring instructions, plus everything it knows about the customer (CRM, email, calendar, transcripts, tickets, product usage), to decide which status the customer belongs in and why.
The output is always two things:
- A health status: Healthy, At risk, or Growth opportunity.
- A reasoning trace: the deep research condensed into 3-5 reasons that explain why the customer landed in that status on this run.
Example instructions
You can mix product signals and conversational signals in the same instructions. A few examples of the kind of rule you can write:
- "If a customer mentions a competitor name in any email or call, move them to At risk."
- "If a customer is asking about additional integrations, mark them as a Growth opportunity."
- "If the primary champion has gone silent for more than 30 days and there is no other engaged contact, move to At risk."
- "If support tickets in the last 14 days have negative sentiment, move to At risk."
The rules can reference anything Cust has access to: product-usage metrics, conversation content, sentiment, and attribute values from your CRM.
Refresh frequency
You set how often the health-scoring agent runs: weekly, every two weeks, monthly, or quarterly. You can also define events that should trigger an immediate re-evaluation (for example, an at-risk customer paying an overdue invoice). When the agent re-runs, the reasoning attached to each status is regenerated against the latest context.
