Health checks

Regular health checks help you catch issues before they impact customers. This page provides checklists and routines for maintaining agent health.

​

Quick reference

​

Daily health check

Perform this check every morning to catch overnight issues.

​

Checklist

• Review Analytics dashboard
  • Check call volume vs. baseline
  • Verify handoff rate is normal
  • Look for latency spikes
  • Identify any error patterns
• Check recent errors
  • Go to Conversation Review → Filter by errors
  • Review last 24 hours
  • Identify any new error patterns
  • Verify no critical function failures
• Monitor handoffs
  • Review handoff reasons
  • Check for new patterns
  • Verify handoff rate is acceptable
  • Identify knowledge gaps
• Verify integrations
  • Check that external APIs are responding
  • Review function logs for API errors
  • Verify no authentication failures
• Test critical paths
  • Make a test call in Agent Chat
  • Verify key functions work
  • Check voice quality
  • Confirm knowledge is current

​

Red flags

Stop and investigate if you see:

• Handoff rate > 50% (or 20% above baseline)
• Average latency > 3 seconds
• Error rate > 5%
• Call volume drop > 30%
• Multiple function failures
• ASR accuracy issues

​

Time required

10-15 minutes

​

Tools needed

• Analytics dashboard
• Conversation Review
• Agent Chat

​

Weekly performance review

Perform this review every Monday to track trends and plan improvements.

​

Checklist

• Analyze performance trends
  • Compare this week to last week
  • Review key metrics:
    • Call volume
    • Containment rate
    • Average call duration
    • Response latency
    • Task completion rate
  • Identify improving or declining metrics
• Review top unhandled queries
  • Go to Analytics → Unhandled queries
  • Identify common questions without answers
  • Prioritize knowledge gaps to fill
  • Plan content updates
• Listen to sample calls
  • Select 5-10 calls from the week
  • Mix of successful and unsuccessful calls
  • Listen for quality issues
  • Note areas for improvement
• Analyze handoff patterns
  • Review handoff reasons
  • Identify trends
  • Determine if containment is improving
  • Plan interventions
• Check function performance
  • Review function execution times
  • Identify slow functions
  • Check error rates
  • Plan optimizations
• Review Test Set results
  • Check automated test outcomes
  • Identify any regressions
  • Update tests as needed
• Plan improvements
  • Create tasks for next week
  • Prioritize by impact
  • Assign owners
  • Set deadlines

​

Deliverables

• Weekly report - Summary of key metrics and trends
• Action items - List of improvements to make
• Knowledge updates - New topics to add

​

Time required

30-60 minutes

​

Tools needed

• Analytics dashboard
• Conversation Review
• Test Sets
• Spreadsheet or doc for tracking

​

Monthly deep dive

Perform this comprehensive review at the end of each month.

​

Checklist

• Month-over-month comparison
  • Compare all key metrics to previous month
  • Calculate percentage changes
  • Identify significant trends
  • Document insights
• Containment rate analysis
  • Review containment trends
  • Break down by handoff reason
  • Identify opportunities to improve
  • Set containment goals for next month
• Knowledge coverage audit
  • Review all Managed Topics
  • Identify outdated content
  • Check for gaps
  • Plan content updates
• Function optimization
  • Review all functions
  • Identify slow or error-prone functions
  • Optimize or refactor as needed
  • Update documentation
• Voice and audio quality
  • Listen to sample calls
  • Check for pronunciation issues
  • Review voice settings
  • Test audio quality
• Test Set maintenance
  • Review all Test Cases
  • Add new scenarios
  • Remove obsolete tests
  • Update expected outcomes
• Integration health
  • Review all API integrations
  • Check for errors or slowdowns
  • Verify credentials are current
  • Test endpoints
• Version history review
  • Review versions published this month
  • Document major changes
  • Identify successful and unsuccessful updates
  • Plan future releases
• Stakeholder report
  • Create monthly summary
  • Highlight wins and challenges
  • Share with team and stakeholders
  • Gather feedback

​

Deliverables

• Monthly report - Comprehensive performance summary
• Improvement roadmap - Prioritized list of enhancements
• Content updates - Knowledge base improvements
• Stakeholder presentation - Executive summary

​

Time required

2-4 hours

​

Tools needed

• Analytics dashboard
• Conversation Review
• Test Sets
• Spreadsheet or BI tool
• Presentation software

​

Quarterly audit

Perform this strategic review every three months.

​

Checklist

• Strategic performance review
  • Review quarterly goals
  • Measure progress against KPIs
  • Identify strategic opportunities
  • Plan next quarter’s objectives
• Comprehensive knowledge audit
  • Review all knowledge sources
  • Remove outdated content
  • Consolidate duplicates
  • Reorganize for clarity
• Function inventory
  • List all functions
  • Identify unused functions
  • Remove or archive obsolete code
  • Document remaining functions
• Voice and brand alignment
  • Review agent personality
  • Ensure brand consistency
  • Test with focus groups
  • Refine as needed
• Security and compliance
  • Rotate API credentials
  • Review access controls
  • Audit PII handling
  • Ensure compliance with regulations
• Infrastructure review
  • Check for performance bottlenecks
  • Review scaling needs
  • Plan capacity upgrades
  • Optimize costs
• Team training
  • Identify skill gaps
  • Plan training sessions
  • Update documentation
  • Share best practices
• Competitive analysis
  • Review competitor agents
  • Identify industry trends
  • Benchmark performance
  • Plan differentiators

​

Deliverables

• Quarterly business review - Strategic summary for leadership
• Roadmap update - Next quarter’s priorities
• Training plan - Team development initiatives
• Budget proposal - Resource needs

​

Time required

4-8 hours

​

Tools needed

• Analytics dashboard
• Business intelligence tools
• Competitive research
• Strategic planning docs

​

Pre-deployment check

Perform this check before promoting any version to Pre-release or Live.

​

Checklist

• Run all Test Sets
  • Execute automated tests
  • Verify all tests pass
  • Investigate any failures
  • Fix issues before promoting
• Manual testing
  • Test critical user journeys
  • Try edge cases
  • Verify new features work
  • Check for regressions
• Performance validation
  • Check response latency
  • Verify function performance
  • Test under load if possible
  • Ensure no degradation
• Knowledge verification
  • Sync all Connected Knowledge sources
  • Verify Managed Topics are current
  • Test knowledge retrieval
  • Check for gaps
• Integration testing
  • Test all external API calls
  • Verify credentials are current
  • Check for rate limiting
  • Ensure error handling works
• Voice and audio check
  • Test voice quality
  • Verify pronunciations
  • Check audio caching
  • Test barge-in if enabled
• Version comparison
  • Compare to current Live version
  • Review all changes
  • Identify potential risks
  • Document breaking changes
• Rollback plan
  • Identify last known good version
  • Document rollback procedure
  • Ensure team is ready
  • Set monitoring alerts

​

Go/No-go criteria

Promote if:

• All Test Sets pass
• No critical bugs found
• Performance is acceptable
• Stakeholders approve

Do not promote if:

• Test failures exist
• Critical bugs discovered
• Performance degraded
• Integrations failing

​

Time required

20-30 minutes

​

Tools needed

• Test Sets
• Agent Chat
• Version comparison tool
• Checklist

​

Post-deployment monitor

Perform this monitoring immediately after promoting to Pre-release or Live.

​

Checklist

First 30 minutes:

• Monitor live calls
  • Watch Conversation Review in real-time
  • Listen to first few calls
  • Check for immediate issues
  • Be ready to rollback
• Check key metrics
  • Response latency
  • Error rate
  • Handoff rate
  • Call volume
• Verify functions
  • Check function logs
  • Ensure no errors
  • Verify API calls succeed
  • Monitor performance
• Test critical paths
  • Make test calls
  • Verify key features work
  • Check voice quality
  • Confirm knowledge is accessible

First 24 hours:

• Track trends
  • Monitor metrics hourly
  • Compare to baseline
  • Identify any anomalies
  • Document observations
• Review feedback
  • Check for customer complaints
  • Review handoff reasons
  • Analyze call quality
  • Gather team feedback
• Performance analysis
  • Compare to previous version
  • Measure improvement or regression
  • Identify unexpected issues
  • Plan follow-up actions

First week:

• Comprehensive review
  • Analyze full week of data
  • Compare to pre-deployment baseline
  • Measure success against goals
  • Document lessons learned
• Optimization
  • Identify quick wins
  • Fix minor issues
  • Optimize performance
  • Plan next iteration

​

Rollback triggers

Rollback immediately if:

• Error rate > 10%
• Handoff rate doubles
• Critical function failures
• Customer complaints spike
• Performance degrades significantly

​

Time required

• First 30 minutes: Active monitoring
• First 24 hours: Periodic checks (every 2-4 hours)
• First week: Daily reviews (15 min/day)

​

Tools needed

• Conversation Review (real-time)
• Analytics dashboard
• Alert system
• Communication channel (Slack, email)

​

Maintenance calendar

​

Sample monthly schedule

Week 1:

• Monday: Weekly performance review
• Daily: Daily health checks
• Friday: Pre-deployment check (if releasing)

Week 2:

• Monday: Weekly performance review
• Daily: Daily health checks
• Wednesday: Mid-month check-in

Week 3:

• Monday: Weekly performance review
• Daily: Daily health checks
• Friday: Pre-deployment check (if releasing)

Week 4:

• Monday: Weekly performance review + Monthly deep dive
• Daily: Daily health checks
• Friday: Month-end reporting

Quarterly:

• Last week of quarter: Quarterly audit

​

Customizing your schedule

Adjust based on:

• Call volume - High-volume agents need more frequent checks
• Complexity - Complex agents require deeper reviews
• Team size - Larger teams can distribute responsibilities
• Business criticality - Mission-critical agents need tighter monitoring

​

Best practices

​

Proactive vs. reactive

• Proactive - Regular health checks catch issues early
• Reactive - Responding to alerts and complaints
• Balance - Do both, but prioritize proactive

​

Documentation

• Keep logs - Document all health checks
• Track trends - Use spreadsheets or dashboards
• Share insights - Keep team informed
• Learn from issues - Document root causes and fixes

​

Automation

• Use Test Sets - Automate regression testing
• Set up alerts - Get notified of anomalies
• Create dashboards - Visualize key metrics
• Script repetitive tasks - Save time on routine checks

​

Team collaboration

• Assign owners - Rotate health check responsibilities
• Share findings - Use Slack or email for updates
• Escalate issues - Don’t let problems linger
• Celebrate wins - Recognize improvements

​

Troubleshooting

​

”I don’t have time for all these checks”

Solutions:

• Start with daily and pre-deployment checks (highest ROI)
• Automate what you can with Test Sets
• Delegate to team members
• Reduce frequency for stable agents

​

”Metrics look fine but customers are complaining”

Solutions:

• Listen to actual calls, not just metrics
• Check for qualitative issues (tone, clarity)
• Review specific customer complaints
• Test edge cases that metrics miss

​

”Too many false alarms”

Solutions:

• Adjust alert thresholds
• Add context to metrics (time of day, day of week)
• Focus on trends, not single data points
• Refine what you monitor

​

”Health checks aren’t catching issues”

Solutions:

• Add more comprehensive Test Cases
• Monitor additional metrics
• Increase check frequency
• Review what you’re measuring

​

Related pages

• Performance monitoring - Detailed performance analysis
• QA and analytics - Using analytics effectively
• Version management - Safe deployment practices
• Common issues - Troubleshooting guide