A customer health score in B2B SaaS is a composite metric that combines multiple signals about a customer's engagement, product usage, relationship quality, and business outcomes into a single indicator of whether the account is healthy, at risk, or in critical danger. The goal of a health score is to give customer success managers a systematic, data-driven way to identify which accounts need their attention before the signs of churn become obvious -- not when the customer sends a churn notice, but weeks or months earlier when intervention can still change the outcome.
What to include in a B2B customer health score
A customer health score is built from multiple signal categories. Each category is weighted by how predictive it is of churn or expansion in your specific product and customer base:
- Product engagement signals (typically 30-40% of score): login frequency, active users as a percentage of licensed seats, feature adoption depth, time spent in the product. High usage is the strongest predictor of retention; declining usage is the earliest warning sign.
- Outcome signals (20-30% of score): whether the customer has achieved the outcome they defined at the start of the relationship; are they reporting positive ROI; did they complete the implementation and reach their first success milestone?
- Relationship signals (20-30% of score): responsiveness to CSM outreach; executive sponsor engagement; attendance at QBRs; NPS score; open support tickets that are unresolved for more than 14 days.
- Contractual signals (10-20% of score): time to renewal; contract expansion or contraction since last renewal; payment health (late payments are a churn predictor in some segments); multi-year vs annual contract status.
- External signals (optional, 10% of score): executive changes at the customer company; company funding status; news mentions that could affect budget; headcount changes.
Designing the health score model
Start by building the model around the signals you can actually measure in your CRM and product analytics -- a health score based on signals you cannot reliably track will not be maintained. The simplest starting model: (1) Assign each signal a score from 1-10 based on its current state (high product usage = 8-10; declining usage = 4-6; no login in 30 days = 1-3); (2) Weight each signal category by its predictive importance; (3) Calculate the weighted average as the health score; (4) Define thresholds: green (health score above 70), yellow (50-70), red (below 50). Validate the model by checking historical customer data: does your current model predict which customers churned in the last 12 months? If not, adjust the weights and thresholds.
How CSMs use health scores
Customer health scores are most valuable as a prioritisation and early warning tool: CSMs with a large book of accounts (20-50+ accounts) cannot give equal attention to all accounts; the health score tells them where to focus. Weekly routine: review all yellow accounts for the underlying signal that pulled the score down and plan an outreach; review all accounts that moved from green to yellow since last week (these are accounts that were healthy but are showing signs of risk); triage all red accounts to determine whether recovery is possible or whether churn is likely. The health score should also be visible to account managers at renewal time -- a customer with a red health score 3 months before renewal is a high churn risk that needs a proactive recovery plan, not a standard renewal conversation.
Frequently asked questions
- What is a customer health score in B2B SaaS?
- A customer health score in B2B SaaS is a composite metric that combines multiple signals about a customer's engagement, product usage, relationship quality, and outcomes into a single score or status indicator (green / yellow / red) that represents whether the account is at risk of churning or likely to expand. Health scores typically combine: product usage signals (login frequency, active user percentage, feature adoption); outcome signals (have they achieved their success criteria?); relationship signals (responsiveness to CSM outreach, NPS, support ticket volume); and contractual signals (time to renewal, payment health, contract changes). The purpose of a health score is to allow CSMs and account managers to prioritise their limited time on the accounts that most need intervention, identifying at-risk accounts before they churn and healthy accounts primed for expansion.
- What are the best signals to include in a customer health score?
- The best signals to include in a customer health score are those that are measurable in your actual data and predictive of churn or expansion in your specific customer base. The strongest universal signals: (1) Product engagement: login frequency (dropping logins are the earliest churn signal for most SaaS products); active user count vs licensed seats (low activation means the product is not being used by the intended users); feature adoption depth (customers who use only one feature are more vulnerable to churn than those who use five); (2) Time to first value: customers who achieved their first success milestone quickly retain better; customers who never completed onboarding are high-risk regardless of how much time has passed; (3) NPS score: customers who give low NPS scores are significantly more likely to churn than high-NPS customers; a low NPS combined with low product engagement is the highest-risk combination; (4) Executive engagement: accounts where the executive sponsor is no longer engaged with the CSM are at higher churn risk than accounts with active executive relationships; (5) Support ticket patterns: unresolved technical issues correlate with churn in technical products; the volume and severity of open tickets is a useful health signal.
- How do you build a customer health score model?
- To build a B2B customer health score model: (1) List all the signals you can reliably measure from your product analytics, CRM, and support tools; (2) For each signal, define what good looks like (green), acceptable (yellow), and concerning (red) based on your knowledge of your customer base; (3) Assign a weight to each signal based on how predictive it is of churn -- product engagement is typically the highest weight; (4) Calculate the health score as the weighted average of all signal scores; (5) Define aggregate thresholds: green (above 70), yellow (50-70), red (below 50) -- adjust these after validation; (6) Validate the model by back-testing: check whether the model would have correctly identified the customers who churned in the last 12 months; adjust weights and thresholds until the model has good predictive accuracy; (7) Implement the model in your CS platform (Gainsight, Totango, ChurnZero, or HubSpot) so it updates automatically and surfaces to CSMs in their daily workflow. Revisit and recalibrate the model annually as your product, customer base, and churn patterns evolve.
Keep reading
- B2B churn prevention: how to reduce churn in B2B SaaS
- Net promoter score: what it is and how to use NPS in B2B
- B2B renewal process: how to manage and improve B2B contract renewals
- B2B QBR: how to run a quarterly business review that customers value
- B2B NRR: what net revenue retention is and how to improve it