A client came to us last year with a support dashboard showing everything green: response time down, ticket count stable, team busy. Two months later their biggest account churned, citing support. The dashboard never flinched — because none of its numbers measured what that customer experienced: three handoffs, two reopened tickets, and a chatbot that closed conversations it hadn’t resolved.
That’s the trap with customer support metrics. The easy numbers (volume, speed, busyness) are the ones tools show by default, and they’re the least predictive of whether customers stay. The metrics that predict churn — first contact resolution, CSAT-adjusted deflection, reopen rates — take slightly more setup, so most small teams never see them.
This guide covers the 8 KPIs worth that setup, the target ranges we use across small-business support deployments at Builts AI, and the vanity metrics to delete from your dashboard.
What Are Customer Support KPIs?
Customer support KPIs are the small set of numbers that tell you whether support is working: satisfaction (CSAT), speed (first response time), completeness (first contact resolution), efficiency (handle time and cost per ticket), and load (volume trend and deflection). A useful dashboard tracks 6-8 weekly. More than that and nobody reads it; fewer and problems hide in the gaps.
The test for any metric: does it change a decision? Each KPI below maps to a specific action when it moves. If a number on your current dashboard doesn’t, it’s decoration.
What’s a Good Customer Satisfaction KPI?
Target 80%+ CSAT on a simple post-resolution survey, with 4s and 5s on a 1-5 scale counted as satisfied. That’s the single most useful customer satisfaction KPI for a small team — it’s per-interaction, fast-moving, and directly tied to what support did this week. Below 75% signals a process problem worth diagnosing; above 90% usually signals a skewed sample rather than perfection.
How you ask matters as much as the score. One question, immediately after resolution, in the same channel the conversation happened. Every added question cuts response rates, and low response rates are how teams end up celebrating a 95% CSAT built on 3% of customers. Track the response rate next to the score, always.
The 8 Metrics Small Teams Should Actually Track
Here’s the full dashboard, with the targets we set across SMB deployments and what to do when each number moves the wrong way.
| # | Metric | Target (small team) | When it slips, look at |
|---|---|---|---|
| 1 | CSAT | 80%+ (with 30%+ response rate) | Read the 1-2 ratings weekly |
| 2 | First Response Time | < 1 hr business hours | After-hours coverage, triage |
| 3 | First Contact Resolution | 70%+ | Agent access to data, KB gaps |
| 4 | Average Handle Time | Stable or falling | Only after FCR is healthy |
| 5 | Ticket Volume Trend | Falling per customer | Top 5 topics — fix root causes |
| 6 | CSAT-Adjusted Deflection | 40-60% | Bot training, escape hatches |
| 7 | Cost per Ticket | $25-35 human / $2-8 automated | Blend improving monthly |
| 8 | NPS | Quarterly only | Relationship, not tickets |
1. CSAT — asked the right way
Covered above. The one addition: read every 1 and 2 rating in a weekly 15-minute review. Low scores are a free root-cause feed, and in small volumes each one is statistically loud.
2. First Response Time (FRT)
Time from customer message to first human or automated substantive reply. Target under 1 hour in business hours. The number most teams get wrong is the after-hours half: a message at 9pm answered at 9am is a 12-hour FRT, whatever your daytime average says. Automated acknowledgment with a real ETA — or automated answers for the repetitive share — is the cheapest fix in support.
3. First Contact Resolution (FCR)
The share of tickets resolved without a follow-up exchange or reopen. Target 70%+. FCR is the metric that predicted our client’s churn story: their reopen rate was climbing for a quarter while response time looked great. Low FCR almost always traces to agents lacking data access (order systems, account history) or a knowledge base that answers the easy questions and skips the hard ones.
4. Average Handle Time (AHT)
Minutes of agent work per resolved ticket. Track it, but optimize it after FCR is healthy — squeezing handle time first teaches agents to close tickets fast rather than completely, which trades AHT for reopens. Healthy pattern: AHT falls as routing and macros absorb the mechanical work.
5. Ticket Volume Trend — by topic, per customer
Raw ticket count is a vanity metric; volume per customer by topic is a KPI. Tag every ticket against your top 15 topics and watch the trend. A topic growing faster than your customer base is a product or documentation bug generating tickets — fixing it beats answering it. This tagging is also the foundation for everything in our cost-reduction playbook.
6. CSAT-Adjusted Deflection Rate
Tickets resolved without a human where the customer rated the outcome 4+, divided by total ticket attempts. Raw deflection rewards bots that close conversations customers didn’t consider resolved; the CSAT adjustment catches it. Target 40-60% adjusted for most small businesses — the full math and benchmarks by industry are in our ticket deflection guide.
7. Cost per Ticket
Fully loaded support cost divided by resolved tickets. Typical small-business numbers: $25-35 per human-handled ticket (a $4,500/month loaded rep handling ~150 tickets), $2-8 per automated resolution once per-resolution AI fees are counted. The blended number is the one that belongs on the dashboard — it’s how you know automation is changing your economics rather than adding a line item.
8. NPS — quarterly, not per ticket
Net Promoter Score measures relationship loyalty, and relationships move slowly. Survey it quarterly at most, separately from support interactions. Teams that bolt NPS onto every ticket get interaction feedback mislabeled as loyalty data — and train customers to ignore surveys entirely. On a weekly support dashboard, CSAT does the work; NPS belongs in the quarterly business review.
How Do You Build the Dashboard?
Start with wherever your tickets live — every mainstream helpdesk (Zendesk, Freshdesk, Intercom, HubSpot) exposes CSAT, FRT, FCR, and volume reporting natively, and a shared spreadsheet fed weekly covers the rest at small-team scale. The assembly rules that matter more than the tool:
- One page. Eight numbers, current value vs target vs last month. If it needs scrolling, it needs cutting.
- Weekly cadence. A 15-minute Monday review: which numbers moved, and the 1-2 CSAT ratings from last week.
- Segment the two that lie when averaged. FRT and CSAT by channel and by business-hours/after-hours. Averages across segments are where “customers complain but data says we’re fine” comes from.
- Automate the collection. Manually assembled dashboards die within a quarter. Piping helpdesk metrics to a sheet is an hour of workflow automation that keeps the habit alive.
What Metrics Should You Stop Tracking?
Vanity metrics don’t just waste dashboard space — they reward behavior that makes support worse. The four we remove from almost every dashboard we inherit:
- Total conversation count. Rises with growth and with product failures alike. Says nothing without per-customer and per-topic context.
- Agent utilization. Rewards looking busy. A team at 95% utilization has no slack for the complex tickets that decide retention.
- Raw deflection rate. Without the CSAT adjustment, it’s a measure of how effectively your bot closes tickets — not how often it helps.
- Per-ticket NPS. Wrong instrument for the question, and it suppresses response rates on the survey that matters (CSAT).
The pattern behind all four: any metric your team can improve while customers get a worse experience will, eventually, be improved exactly that way.
How Do You Set Targets Without Copying Competitors?
Set targets from your own baseline, not industry benchmark tables. Measure four weeks of current performance, then set each target as a realistic improvement on your number — cut FRT in half, lift FCR ten points — and re-baseline quarterly. Benchmark tables average across industries, ticket mixes, and team sizes that don’t resemble yours; your own trend line is the only comparison that’s always fair.
The exception is direction: whatever your baseline, CSAT belongs above 75% and FCR above 60% before anything else gets optimized. Below those floors, the problem isn’t targets — it’s process, and usually one of the fixable patterns above.
Want the Dashboard Built for You?
Book a free automation audit and we’ll pull your last 200 tickets, tag them, baseline all 8 metrics, and set up the dashboard with automated weekly collection — plus flag which metric is costing you the most and the fastest fix for it.
