Understanding Health Scores

Overview#

Health scores give you a single number (0-100) that represents how well a customer relationship is going. Instead of guessing which accounts need attention, health scores surface at-risk customers before they churn.

Klyra's health scores combine support signals (ticket volume, sentiment, response times) with engagement data (product usage, login frequency) and AI analysis (pattern detection, trend prediction) into an actionable score.

How scores are calculated#

Each customer account gets a health score based on weighted factors:

Support signals (40% weight)#

Engagement signals (30% weight)#

Relationship signals (30% weight)#

AI adjustment#

After calculating the formula-based score, Klyra's AI reviews the account's recent activity and applies an adjustment of up to +/- 10 points. The AI explains its reasoning — for example:

"Score adjusted -5: Customer opened 3 urgent tickets this week about the same billing issue. Pattern suggests systemic frustration, not isolated incidents."

Score ranges#

Setting up health scores#

Step 1: Enable health scoring#

1. Go to Settings > Customer Intelligence > Health Scores
2. Toggle health scoring on
3. Klyra begins calculating scores for all accounts with enough data

Step 2: Configure weights (optional)#

The default weights work well for most teams, but you can customize them:

1. Go to Settings > Customer Intelligence > Health Score Config
2. Adjust the weight percentages for support, engagement, and relationship signals
3. Changes take effect on the next score calculation cycle (every 6 hours)

Step 3: Set up alerts#

Get notified when accounts need attention:

1. Go to Settings > Notifications
2. Enable health score alerts for:
   • Score drops below a threshold (e.g., 60)
   • Score changes by more than 15 points in a week
   • Account moves from "Healthy" to "At Risk" status

Using health scores day-to-day#

Dashboard overview#

The Dashboard shows a health overview card with:

• Distribution of accounts by health status
• Accounts that changed status recently
• Trending scores (improving vs. declining)

Account detail page#

Click on any customer in Customers to see:

• Current health score with trend chart
• Score breakdown by factor
• AI reasoning for the latest calculation
• Health score history over time
• Recommended actions

Anomaly detection#

Klyra automatically detects anomalies — unusual changes in account behavior:

• Sudden spike in ticket volume
• Sharp sentiment drop
• Unusual activity patterns (e.g., a power user goes silent)

Anomalies appear in your Dashboard > Anomaly Alerts card and trigger notifications if configured.

Churn prediction#

Building on health scores, Klyra's churn prediction model analyzes:

• Health score trajectory (direction matters more than absolute value)
• Historical patterns from other accounts that churned
• Contract timeline and renewal history
• Engagement velocity (rate of change in activity)

The result is a churn risk percentage and an actionable save plan — specific steps your team can take to retain the customer.

Best practices#

1. Don't just look at the number — read the AI reasoning and factor breakdown to understand why the score is what it is
2. Set up weekly health reviews — use the Daily Digest to review accounts that need attention
3. Act on "At Risk" early — accounts at 40-59 are easier to save than those below 40
4. Track improvement — after taking action, monitor if the health score responds. If not, the intervention didn't address the root cause
5. Use it in handoffs — when reassigning an account, the health score gives the new owner instant context

Next steps#

• Configure the AI agent — AI-resolved tickets contribute positively to health scores
• Set up Slack — Slack activity feeds into engagement signals
• Getting started guide — return to the overview if you need to set up other features