The Metrics That Matter
Tier 1: Track These Daily
First Response Time (FRT) What it measures: How long customers wait for an initial human response. Why it matters: FRT is the single most impactful metric on customer satisfaction. Customers who wait 10 minutes are dramatically more satisfied than those who wait 4 hours, even if the resolution takes the same amount of time. Target: Under 4 hours for standard support, under 1 hour for premium. How to calculate: Time from ticket creation to first non-automated human response.
Auto-Resolution Rate What it measures: Percentage of conversations resolved by AI without human intervention. Why it matters: This is the primary efficiency metric for AI-powered support. Higher auto-resolution means less load on human agents and faster answers for customers. Target: 60-75% for a well-configured AI system. How to calculate: (AI-resolved conversations / total conversations) x 100.
Ticket Volume What it measures: How many new support conversations arrive per day, week, or month. Why it matters: Volume trends drive staffing, capacity planning, and early detection of product issues. A sudden spike in volume often signals a bug, outage, or confusing product change. Target: Varies by company. Track trend, not absolute number. How to calculate: Count of new conversations in the time period.
Tier 2: Track These Weekly
CSAT Score What it measures: Customer satisfaction with their support experience. Why it matters: CSAT is the most direct measure of support quality from the customer's perspective. Target: Above 85% (on a 5-point scale where 4 and 5 count as satisfied). How to calculate: (Satisfied responses / total responses) x 100. Watch out for: Low response rates make CSAT unreliable. If only 10% of customers respond, your data may be biased toward extremes.
First Contact Resolution (FCR) What it measures: Percentage of issues resolved in a single interaction. Why it matters: FCR correlates strongly with customer satisfaction. Customers who need to contact you twice about the same issue are significantly less satisfied. Target: Above 75%. How to calculate: (Conversations resolved in one interaction / total conversations) x 100.
Average Handle Time (AHT) What it measures: Average time agents spend on each conversation. Why it matters: AHT indicates efficiency. However, optimizing for speed at the expense of quality is counterproductive. Track AHT alongside CSAT to ensure efficiency does not hurt quality. Target: Varies by complexity. Use as a trend indicator, not a performance target. How to calculate: Total agent active time / number of conversations handled.
Escalation Rate What it measures: Percentage of conversations escalated from AI to human or from agent to senior agent. Why it matters: High escalation rates indicate knowledge base gaps (AI to human) or training gaps (agent to senior). Tracking escalation reasons helps target improvements. Target: 25-40% for AI-to-human escalation. How to calculate: (Escalated conversations / total conversations) x 100.
Tier 3: Track These Monthly
Customer Effort Score (CES) What it measures: How much effort the customer had to exert to resolve their issue. Why it matters: Low-effort experiences predict loyalty. CES captures friction that CSAT might miss — a customer might be satisfied with the answer but frustrated by how hard it was to get. Target: Below 3 on a 7-point effort scale (lower is better).
Net Promoter Score (NPS) for Support What it measures: Whether customers would recommend your support experience. Why it matters: NPS captures the holistic impression of your support, not just satisfaction with a single interaction. Target: Above 30 is good, above 50 is excellent for support-specific NPS.
Contact Ratio What it measures: Number of support contacts per 100 customers per month. Why it matters: A rising contact ratio means your product is generating more support needs per customer. This often signals product usability issues or documentation gaps. Target: Varies by product complexity. Track the trend.
Support Cost Per Ticket What it measures: Total support cost divided by total tickets resolved. Why it matters: This is the efficiency metric that connects to your budget. Track it alongside quality metrics to ensure cost reduction does not sacrifice quality. How to calculate: (Total support spend including tools, salaries, and overhead) / total tickets resolved. Use the support cost calculator to model this for your team.
Building Your Analytics Dashboard
The Executive Dashboard
For leadership, build a single-page view showing:
- CSAT trend — Current score and 12-week trendline
- Auto-resolution rate — Current percentage and trend
- Ticket volume — Current week vs. previous 4-week average
- First response time — Current average and SLA compliance rate
- Support cost per ticket — Current and trend
This gives executives the five numbers they need to assess support health in 30 seconds.
The Operations Dashboard
For the support manager, build a real-time view showing:
- Queue health — Open tickets by priority, oldest unresponded ticket, agent workload distribution
- SLA compliance — Real-time compliance percentage, at-risk tickets, breached tickets
- Volume patterns — Current day volume vs. same day last week, hour-by-hour pattern
- Agent performance — Tickets per agent, FRT per agent, CSAT per agent
This dashboard drives daily operational decisions: who is overloaded, where are SLAs at risk, and is today's volume normal.
The AI Performance Dashboard
For monitoring your AI support system:
- Auto-resolution rate — Overall and by topic category
- AI CSAT — Customer satisfaction for AI-resolved conversations
- Escalation reasons — Why the AI escalated (low confidence, customer request, topic trigger)
- Knowledge base gaps — Topics where the AI says "I do not know" most frequently
- False resolutions — Conversations the AI marked as resolved that the customer reopened
Turning Data Into Decisions
Analytics without decisions is just reporting. Here are the decision frameworks for common scenarios:
When CSAT is Declining
- Segment CSAT by channel (email vs. chat), by agent, by topic category
- Identify the segment driving the decline
- If agent-specific: investigate recent conversations, provide coaching
- If topic-specific: review the knowledge base for that topic, update if needed
- If channel-specific: investigate whether response times differ by channel
When Auto-Resolution Rate is Dropping
- Pull the list of topics where AI escalated most frequently
- Check if knowledge base articles for those topics were recently changed or removed
- Check if a product change created new questions the knowledge base does not cover
- Create or update knowledge base articles for gap topics
- Monitor the rate for 1-2 weeks after changes
When Ticket Volume is Spiking
- Categorize the spike: is it across all topics or concentrated in one area?
- If concentrated: likely a bug, outage, or confusing product change. Alert engineering.
- If broad: likely seasonal, growth-related, or caused by a marketing event. Assess whether the spike is temporary.
- Short-term: reallocate agent capacity to the spike area
- Long-term: if spikes are recurring, adjust staffing or AI coverage
When Support Cost Per Ticket is Rising
- Check if ticket volume increased (more tickets = more cost with same team)
- Check if average handle time increased (more complex tickets take more agent time)
- Check if AI auto-resolution rate decreased (more tickets flowing to human agents)
- Address the root cause: improve knowledge base for handle time, fix AI configuration for resolution rate, staff up for volume
Common Analytics Mistakes
Tracking too many metrics — Fifteen metrics compete for attention and none get it. Focus on 5-7 core metrics and check secondary metrics only when investigating problems.
Optimizing one metric in isolation — Reducing average handle time without watching CSAT creates fast, low-quality support. Reducing support cost without watching auto-resolution rate just means less coverage. Always watch metrics in pairs.
Comparing across teams without context — An enterprise support team with complex tickets will have a higher AHT than a self-service team handling simple questions. Compare agents within the same context, not across different support types.
Ignoring response rate on CSAT — A CSAT of 95% means nothing if only 5% of customers responded. The most dissatisfied customers often do not bother responding to surveys. If your CSAT response rate is below 20%, invest in improving it before trusting the number.
Monthly-only reporting — Monthly reviews catch problems too late. Daily monitoring of queue health and volume, weekly reviews of CSAT and resolution rates, and monthly deep dives on trends give you the right cadence for each metric.
Customer support analytics is not about building the most sophisticated dashboard — it is about measuring the right things, reviewing them at the right cadence, and acting on what you find. Start with Tier 1 metrics, build a simple dashboard, and develop the habit of data-driven decisions. The sophistication comes naturally as your team grows.
Want to simplify your support workflow? Try Corebee free — flat-rate pricing, unlimited agents.