How to track customer service performance: metrics, tools, and cadence

Summary: Most ecommerce brands track customer service with four numbers: ticket volume, response time, CSAT, and some measure of cost. The brands that improve fastest track eight numbers, review them on a weekly cadence, and tie each metric to one action. This guide covers exactly what to track, how to pull it, and how to turn the data into changes that reduce cost and improve CSAT.

---

Why most support teams track the wrong things

Ticket volume tells you how busy your team is. It does not tell you whether your team is solving problems or creating new ones. A team with 1,000 tickets per month and a 30% re-contact rate is effectively handling 1,300 problems. The extra 300 are customers who did not get their issue resolved the first time.

CSAT tells you whether customers felt good about an interaction. It does not tell you whether the interaction was necessary. A team with 4.8/5 CSAT and 70% human-handled tickets is running an expensive, happy support operation. The 70% human-handled tickets could be AI-deflected.

Track metrics that connect to cost and to resolution quality, not just volume and sentiment.

This article is part of the ecommerce customer service hub.

---

The 8 metrics that matter

1. Ticket volume by category. Total volume is less useful than volume broken down by category: WISMO, returns, product questions, billing, complaints. Category breakdown tells you where to invest in deflection first. WISMO at 35% of volume is a clear directive: build WISMO deflection before anything else.

2. First response time (FRT) by channel. Average time from customer message to first reply, tracked separately for AI chat, human chat, email, WhatsApp, and Instagram. Benchmarks: under 10 seconds for AI chat, under 2 minutes for human chat, under 4 hours for email. One number averaged across channels hides poor performance on specific channels.

3. Deflection rate. Percentage of tickets resolved without a human agent. Formula: AI or self-serve resolutions divided by total support attempts, multiplied by 100. Target: 70 to 85% for AI-first ecommerce brands. Track deflection rate separately for each ticket category. A 50% overall deflection rate with 0% WISMO deflection means you have not yet connected order data to AI.

4. CSAT by ticket type. Overall CSAT masks where the problems are. An 85% overall CSAT with 60% CSAT on returns and 95% CSAT on product questions tells you the returns process needs work. Measure CSAT for each major ticket category, not just overall.

5. Re-contact rate. Percentage of customers contacting again within 72 hours of a resolved ticket. Target: under 10%. Above 15% on a specific ticket type signals that your resolution for that type is not actually resolving the issue.

6. Average resolution time (ART). Total time from first contact to full resolution, including all back-and-forth. Different from first response time. A ticket responded to in 2 minutes but taking 3 days to resolve has poor ART despite good FRT. Target: under 30 minutes for chat, under 24 hours for email.

7. Cost per ticket. Total support cost (platform + agent salaries + overhead) divided by total tickets resolved. This metric connects your support operation to the P&L. A cost per ticket above $15 for standard queries signals over-reliance on human handling for deflectable ticket types.

8. Knowledge base gap rate. Percentage of AI or self-serve searches that return no results. This is the leading indicator of deflection rate decline. If 20% of searches return no results this month, deflection rate will drop next month. Fill gaps before they affect the headline number.

---

Review cadence: what to check and when

Daily (only during high-volume events like BFCM):

• Open ticket queue depth
• Average response time in the last 4 hours
• Any SLA breaches

Weekly (30-minute review every Monday):

• Ticket volume vs prior week, by category
• FRT by channel vs target
• Deflection rate vs target
• CSAT score vs prior week
• Top 3 ticket types that reached humans (candidates for AI improvement)

Monthly (60-minute review, first Monday of the month):

• Cost per ticket trend (last 3 months)
• Re-contact rate by ticket type
• ART by channel
• Knowledge base gap rate and which terms are missing
• SLA hit rate by channel
• One action assigned per metric below target

Quarterly:

• Team capacity vs volume trend (do you need to hire or can AI absorb growth?)
• Platform ROI calculation (deflection savings vs platform cost)
• SLA review (are targets still appropriate for your volume?)

---

The weekly review format

A weekly support review should take 30 minutes maximum. The format:

1. Pull the 8 metrics from the previous 7 days.
2. Compare to the prior 7 days and to the 30-day average.
3. Identify the one metric furthest below target.
4. Assign one action to move that metric before the next review.
5. Note it and close.

Do not try to improve everything at once. One metric, one action, one week. Teams that improve systematically over 12 weeks outperform teams that hold monthly “deep dives” and never act between them.

Tropicfeel tracks deflection rate and CSAT by ticket category weekly. They identify gaps and update AI training accordingly. The result: 85% automation with CSAT above 90%. Read the Tropicfeel case study.

---

Building your tracking stack

Minimum viable stack:

• Helpdesk with built-in reporting: Gorgias, Zendesk, or Freshdesk. Each includes ticket volume, FRT, and CSAT dashboards out of the box.
• CSAT collection: built-in helpdesk CSAT survey (thumbs up/down after ticket close). Export weekly.
• AI deflection dashboard: if using Zipchat or a similar AI platform, use the built-in deflection analytics. This shows AI resolution rate, escalation rate, and AI confidence score per ticket type.
• Monthly tracking spreadsheet: one Google Sheet with 8 columns (the 8 metrics above) and one row per week. This is your trend view. The helpdesk dashboard shows current performance. The spreadsheet shows whether you are improving.

When to add more tools: Add a dedicated analytics platform (Klipfolio, Geckoboard) only when you have more than 5 agents and a team leader who reviews dashboards daily. Below that threshold, a weekly spreadsheet review is faster and requires no additional software cost.

---

Turning data into action

The metrics above have no value unless they drive decisions. Every metric below target should have one question attached: “What is the single most likely cause of this gap?”

FRT above target for email: cause is usually queue depth (too many tickets for available agents) or routing (tickets going to wrong queue). Fix: add routing rule or reduce human-handled tickets via AI.

Deflection rate below target: cause is usually missing knowledge base content or AI not connected to order data. Fix: identify top 5 categories not deflecting and add FAQ entries or connect order data.

Re-contact rate above 10%: cause is incomplete resolutions (agent closed ticket before customer’s issue was fully addressed) or incorrect AI answers. Fix: review a sample of re-contact tickets and identify the common resolution gap.

CSAT below 80% for a specific ticket type: cause is either resolution quality or resolution time. Review the last 20 low-CSAT tickets for that type. If agents are responding fast and correctly, the issue is the resolution itself (policy or product). If agents are slow or incorrect, it is a process issue.

Family Nation improved CSAT by tracking it per ticket category and identifying that returns CSAT was dragging down their overall score. Read the Family Nation story.

---

Related reading

• Ecommerce customer service hub
• Customer service KPIs: CSAT, FCR, NPS, and AHT explained
• Ticket deflection strategies for ecommerce
• Customer service response time: benchmarks and improvement
• Customer service automation playbook

---

See your metrics in real time

Book a demo to see Zipchat’s analytics dashboard showing deflection rate, CSAT, and AI confidence score by ticket type. Book a demo or start a free trial.

Return to the ecommerce customer service guide for the full cluster.