Customer Health Score Calculation
Most health scores are built from whatever data was easiest to pull — login frequency, maybe a support ticket count — and end up rewarding vanity activity while missing the accounts that are quietly disengaging. A power user who logs in daily but has stopped using the one feature they actually bought the product for looks 'green' right up until they churn. Meanwhile CSMs distrust the score because it's flagged healthy accounts that left and flagged departing accounts as fine, so they stop checking it and go back to gut feel.
STARTING PRICE
From €299
Standard tier · Multi-step workflow with AI extraction/decisioning and 2-3 integrations.
Get a quote →Saves roughly 3-5 hrs/week per CSM, plus earlier at-risk detection ahead of renewal.
How the automation works
We build a composite health score from the signals that actually correlate with renewal in your business — feature-level usage weighted by what the account's contract is actually for, support ticket sentiment and resolution time, executive sponsor engagement, and payment or contract status — rather than a single generic activity metric. Weights are calibrated against your own historical renewal and churn data so the score reflects what predicted outcomes for your customers specifically, not a template borrowed from a different product category. Score changes are explained in plain language ('usage of the core reporting feature dropped 40% over 30 days') rather than shown as a bare number, so a CSM knows what to act on the moment the score moves.
Process flow
- 01
Signals ingest on schedule trigger
Usage events, support tickets, NPS responses, contract and payment data pull in from source systems on a daily cadence.
- 02
Weight usage by contract fit ai
Feature usage is weighted against what the account actually purchased, so usage of only peripheral features scores lower even with high overall login volume.
- 03
Calibrate against historical outcomes ai
Score weights are tuned against your own past renewal and churn data, so the formula reflects what genuinely predicted outcomes for your customer base.
- 04
Generate plain-language explanation ai
Each score change is paired with the specific driver behind it, not just a delta, so the CSM knows exactly what shifted.
- 05
Segment and rank accounts output
Accounts are segmented into health bands and ranked within each CSM's book, surfacing which declining accounts need the most urgent attention.
- 06
Push score to CRM integration
The score and its explanation sync into the CRM account record, visible alongside the account without a separate tool to check.
Inputs
- Product usage event stream by feature
- Support ticket volume and resolution time
- NPS or CSAT survey history
- Contract terms and payment status
- Historical renewal/churn outcomes for calibration
Outputs
- Composite health score per account
- Plain-language score-change explanation
- Health band segmentation
- CSM-ranked at-risk account list
Works with
Prefer a fully custom build instead of an off-the-shelf integration? We scope both options during your free consultation — most jobs like this one work fine on standard connectors, but higher-volume or non-standard systems sometimes need bespoke API work, reflected in the complex tier.
Where this goes wrong if you get it wrong
- A health score built on raw login count or session length produces false negatives on low-usage-but-high-value accounts — an executive sponsor who logs in once a month to review a dashboard their team built might be your most strategically important stakeholder, and a usage-only score will flag that account as unhealthy right before renewal.
- Uncalibrated weights borrowed from a generic template will misrank your specific customer base — the signal that predicts churn in a usage-based product (declining daily actives) differs from what predicts it in a project-based service (missed milestone deadlines), and weights need to be trained on your own historical outcomes, not assumed.
- A single company-wide score hides account-level nuance in multi-stakeholder enterprise deals — one disengaged department can offset a thriving one and average out to a misleadingly stable score, so large accounts need a per-department breakdown, not just an account-level rollup.
- Scores that update in real time on noisy signals (a single bad support ticket, one skipped login) whipsaw and train CSMs to ignore the tool entirely — smoothing over a rolling window and requiring sustained signal change before flagging an account keeps the score trustworthy.
Frequently asked questions
How is this different from the health score built into our CS platform?
The built-in score is usually a starting formula; this calibrates weights against your own historical renewal and churn outcomes and adds contract-fit weighting, which most out-of-the-box scores don't do.
Will this catch accounts that look active but are actually at risk?
Yes — that's the specific gap this closes. Weighting usage by contract fit and factoring in sponsor engagement catches accounts where surface activity looks fine but the signals that matter for renewal are declining.
How often does the score recalculate?
Daily by default, with smoothing over a rolling window so a single noisy data point doesn't cause the score to swing and lose the team's trust.
Can we use this for expansion signals too, not just churn risk?
Yes — the same signals that flag risk (rising feature adoption, growing seat usage) can be inverted to surface expansion-ready accounts.