1. First Response Time (FRT)
First Response Time (FRT) measures how long a customer waits before receiving a meaningful reply. Not an auto-acknowledgment — an actual response that addresses their question. For email and chat support, the benchmark for B2B SaaS is under 4 hours for email and under 2 minutes for live chat (Zendesk CX Trends Report). AI-powered support has reset expectations here: customers who interact with an AI agent expect a response in under 10 seconds. If your AI is handling first responses, your FRT should be nearly instant for the majority of conversations. Track FRT by channel and by time of day to identify coverage gaps.
2. Customer Satisfaction Score (CSAT)
Customer Satisfaction Score (CSAT) remains the most direct measure of support quality. Ask customers to rate their experience after a conversation is resolved. The industry benchmark for B2B SaaS is 85-92% (Salesforce State of Service). Anything below 80% signals a systemic problem — either response quality, resolution time, or agent knowledge. Track CSAT separately for AI-handled and human-handled conversations. A well-implemented AI system should achieve CSAT within 5 percentage points of your human agents. If the gap is larger, your AI needs better training data or tighter guardrails.
3. Resolution Time
Resolution Time measures the total elapsed time from when a customer opens a conversation to when the issue is fully resolved. This is different from first response time — it captures the entire support experience including back-and-forth, escalations, and waiting for internal investigation. The benchmark for B2B SaaS is under 8 hours for standard issues and under 24 hours for complex technical problems. Track median rather than average, because a few outlier tickets with week-long investigations will skew the average and mask real performance.
4. First Contact Resolution (FCR)
First Contact Resolution (FCR) rate tracks the percentage of conversations resolved in a single interaction without follow-up or reopening. High FCR means your team is solving problems correctly the first time. The benchmark is 70-80% for human agents and 55-65% for AI-handled conversations. Low FCR usually indicates knowledge gaps, insufficient agent training, or an AI knowledge base that provides partial answers. Improving FCR is one of the highest-leverage activities because every reopened ticket costs roughly double the resources of a single-touch resolution.
5. Auto-Resolution Rate
Auto-Resolution Rate is the percentage of conversations your AI resolves without any human intervention. This is the defining metric of 2026-era support operations. A well-configured AI with a comprehensive knowledge base should achieve 60-75% auto-resolution. Track this metric alongside CSAT for AI-handled conversations — a high auto-resolution rate is only valuable if customers are actually satisfied with the AI's answers. If your auto-resolution rate is below 50%, the problem is almost always your knowledge base, not the AI model.
6. Ticket Volume and Trends
Ticket Volume and Trends matter more than the absolute number. Track total volume, volume by category, and week-over-week trends. Rising volume in a specific category signals a product issue, a documentation gap, or a recent change that confused customers. Declining volume after a knowledge base update confirms the update is deflecting tickets effectively. Segment volume by source — widget, email, in-app — to understand where customers are reaching out and optimize each channel accordingly.
7. Escalation Rate
Escalation Rate measures how often conversations are transferred from AI to human agents or from frontline agents to specialists. A healthy escalation rate is 25-40% of total conversations. Below 20% suggests the AI or frontline agents may be attempting to resolve issues beyond their capability. Above 50% means the AI or first-tier support is not handling enough independently. Track the reasons for escalation to identify patterns — if 30% of escalations are about billing, that is a clear signal to improve your billing documentation and AI training.
8. Cost Per Resolution
Cost Per Resolution divides your total support costs by the number of resolved conversations. This is the metric that connects support quality to business economics. The benchmark for B2B SaaS is $8-15 per human-resolved conversation and under $0.50 per AI-resolved conversation (Forrester Research). Track this monthly and watch for trends. Rising cost per resolution often indicates increasing complexity (your team spends more time per ticket) or inefficiency (process problems, tool friction, or undertrained agents).
9. Net Promoter Score (NPS)
Net Promoter Score (NPS) for support-specific interactions measures whether your support experience creates promoters or detractors. This is different from your overall product NPS:
- Above 50 is excellent
- Between 30 and 50 is good
- Below 30 indicates your support experience is actively harming customer loyalty
Survey a sample of customers monthly and segment results by issue type, channel, and whether AI or a human handled the conversation.
10. Knowledge Base Coverage
Knowledge Base Coverage tracks the percentage of incoming support topics that have a corresponding knowledge base article. Audit this quarterly by reviewing your top 100 ticket topics and checking article coverage. The target is 80% coverage of your top support topics. Each gap represents a category of questions your AI cannot answer accurately and your customers cannot self-serve. Closing coverage gaps is the single most effective way to improve both auto-resolution rates and customer satisfaction simultaneously.
Bringing It All Together
Key insight: The key to making these metrics useful is consistency and context. Track them monthly, review them as a team, and always pair efficiency metrics (FRT, resolution time, cost per resolution) with quality metrics (CSAT, FCR, NPS). Optimizing for speed without monitoring quality leads to fast but unhelpful support. Optimizing for quality without monitoring efficiency leads to excellent but unsustainably expensive support. The best teams hold both in balance.
Sources
Want to simplify your support workflow? Try Corebee free — flat-rate pricing, unlimited agents.