Customer Health Scorecard Template: 7 Signals (2026)

A useful customer health score does one job: it tells your team which accounts need attention before the renewal conversation becomes a rescue mission. This customer health scorecard template gives you seven signals, a 0–100 scoring method, and clear actions for green, yellow, and red accounts.

You do not need a data warehouse or a full customer-success platform to start. If product usage, billing, account, and support data already land in your database, you can calculate the score there and refresh it automatically.

The customer health scorecard template at a glance

Score each signal from 0 to 100, multiply it by its weight, then add the weighted results. Start with the suggested weights below. Change them only after your own churn and renewal data shows that another signal predicts risk better.

The formula is: total health score = sum of each signal score multiplied by its weight. An account scoring 80, 70, 60, 90, 50, 40, and 100 across the seven signals would receive 73.5 out of 100.

1. Activation milestone completion: 20%

Activation measures whether the customer reached the first point where your product delivers real value. Do not substitute logins, page views, or setup clicks for that outcome. For an invoicing product, activation might be sending the first invoice; for a database analytics tool, it might be connecting a database and receiving a correct answer.

Score 100 when the agreed milestone is complete, 50 when setup has started but value has not been reached, and 0 when the account remains inactive after the expected onboarding window. Keep this signal prominent for new customers, then reduce its influence after activation is complete.

2. Core feature adoption: 20%

Measure use of the two or three features most closely tied to retention. A long feature checklist creates noise because not every capability matters equally. Pick the behaviours successful customers repeatedly demonstrate.

For example, score an account on the percentage of its licensed team using a core workflow, the number of recurring dashboards created, or whether an automation runs every week. Compare the account with its own plan and use case, not with a global average that mixes very different customers.

3. Usage frequency and trend: 15%

A static usage count hides the direction of travel. Store the current 30-day activity, the previous 30-day activity, and the percentage change. Falling use is often more actionable than low use because it tells you something changed.

A simple score can award 100 for stable or growing activity, 70 for a decline under 15%, 40 for a decline between 15% and 40%, and 0 for a drop above 40% or no meaningful activity. Adjust the time window to your product rhythm; a monthly finance tool should not be judged like a daily collaboration app.

4. Customer outcome attainment: 20%

Usage is not the same as value. Record the outcome the customer bought your product to achieve, its target, its current result, and the latest review date. This may be hours saved, tickets resolved, revenue recovered, reports automated, or time-to-insight reduced.

Score the signal from progress toward the agreed target. If the outcome is qualitative, use a documented customer confirmation from a review rather than allowing the account owner to guess. This signal keeps a highly active but unsuccessful account from appearing healthy.

5. Support friction: 10%

Count unresolved high-priority tickets, repeat issues, age of the oldest open ticket, and recent escalation status. Ticket volume alone is misleading: active customers may create more tickets simply because they use the product more.

Start at 100, subtract points for unresolved severity-one or severity-two issues, and apply a smaller penalty for repeated lower-priority problems. Restore the score after the problem is fixed and the customer confirms the resolution. Otherwise yesterday's incident permanently distorts today's health.

6. Stakeholder engagement and account breadth: 10%

Single-threaded accounts are fragile. Track the number of active users, whether an executive sponsor is known, the last meaningful customer-success interaction, and whether the original champion is still engaged.

A healthy score might require activity from at least two roles, a recent value conversation, and an identified decision-maker. Do not score email opens as engagement. A reply, meeting, workflow action, or product event is harder to fake and more useful.

7. Commercial and billing health: 5%

Include contract end date, renewal notice date, payment status, failed payment count, plan fit, and any unresolved pricing concern. Billing health deserves a signal, but it should not dominate the score: a paid invoice cannot compensate for a customer receiving no value.

Score 100 for current payments and no commercial blocker, 50 for a recoverable warning such as one failed payment, and 0 for delinquency, a cancellation request, or an unresolved renewal objection.

Fields to copy into your scorecard

Create one row per customer account. Include account ID, owner, segment, plan, contract value, lifecycle stage, renewal date, the seven raw signal values, seven normalized scores, total score, score change, risk reason, next action, action owner, and last-updated timestamp.

The risk reason and next action fields are essential. A score of 42 is only a warning; “core usage fell 38% after the champion left” tells the customer-success manager what to do next.

How to calculate the score in five steps

Step 1: define one account-level record

Choose a stable account ID and map users, subscriptions, tickets, and product events to it. Decide how you will treat subsidiaries, workspaces, and multiple contracts before calculating anything.

Step 2: normalize each signal

Convert every raw value into a 0–100 score. Write the rule next to the field so another person can reproduce it. Binary signals can use 0 and 100; continuous signals should use explicit bands or a capped formula.

Step 3: apply weights

Multiply each signal by its weight and sum the results. Keep the first version simple. A transparent weighted score is easier to challenge and improve than an opaque model nobody trusts.

Step 4: handle missing data honestly

Do not silently convert missing data to zero. Mark it as unknown, show a data-completeness percentage, and either reweight the available signals or withhold the total score below a minimum completeness threshold.

Step 5: set operational bands

Start with green at 75–100, yellow at 50–74, and red below 50. Add a separate rule for sharp movement, such as a 15-point drop in seven days, because a green account falling fast can require more attention than a stable yellow account.

Turn the score into action, not dashboard decoration

Give each band a playbook. Green accounts can receive expansion discovery and advocacy requests. Yellow accounts need a named risk reason, an owner, and a review date. Red accounts require immediate triage, a customer conversation, and an agreed recovery outcome.

Add event-based actions for changes that cannot wait for a weekly review. Examples include alerting the account owner when usage drops 30%, emailing finance after a second failed payment, or sending a webhook when a high-value account moves from yellow to red.

Build the scorecard from your live database without SQL

AI for Database lets you connect PostgreSQL, MySQL, Supabase, MongoDB, BigQuery, and other databases, then describe the score in plain English. You can ask it to calculate each signal by account, create a self-refreshing health dashboard, and show which input caused each score to change.

Once the result is trustworthy, create workflows for the moments that need action: send an email, post a Slack message, or call a webhook when an account crosses a threshold. This closes the gap between seeing risk and acting on it. Start at https://aifordatabase.com and use read-only database credentials with access limited to the necessary tables.

Common scorecard mistakes

Questions teams ask about customer health scorecards

What is the simplest customer health scorecard template?

Use one row per account with activation, core adoption, usage trend, outcome attainment, support friction, stakeholder engagement, and billing health. Normalize each to 0–100, apply the weights in this guide, and record the reason behind every yellow or red result.

How often should a customer health score update?

Daily is enough for most SaaS teams. Update faster only when product activity is high-frequency and immediate intervention changes the outcome. Contract, stakeholder, and outcome fields may still need a human review after customer conversations.

Should customer sentiment be part of the score?

Yes, when it comes from a recent, attributable signal such as a survey response or documented meeting. Do not let an old satisfaction score outweigh current usage decline, unresolved support issues, or a missed outcome.

Start with version one

Ship the seven-signal scorecard, review every red and yellow account, and record whether each warning was useful. After one or two renewal cycles, compare signal values for retained and churned customers. Change the weights from evidence, not internal opinion.