Most CX dashboards measure the wrong things. They count tickets, average reply times and CSAT — and stop there. None of those tell you whether your support function is making the brand money or costing it money.
This is the KPI stack we instrument on every retainer at custo.tech. Five categories, twelve metrics, and the numbers you should expect once they're tuned.
Why "tickets per agent" is a vanity metric
The most common KPI on CX dashboards is volume — tickets handled, response time, resolution time. These matter, but only as inputs. Volume is something you optimize away. A team handling fewer tickets because automation took the easy ones is doing better, not worse.
The metrics that matter answer a different question: does support make the brand more money than it costs?
The brands that grow CX as a function — not shrink it — are the ones running this question on their dashboard.
Category 1: Operational baseline
You need these to know if the basics are working. Don't celebrate them. They're table stakes.
First Response Time (FRT). Time from ticket creation to first agent (or AI Agent) reply. Bar: under 30 minutes during business hours on email, under 2 minutes on chat. Below those, customers churn.
Resolution Time. Time from ticket creation to resolved status. Bar varies by category — sizing question should resolve in under an hour, complex returns can take 3 days. Track by category, not in aggregate.
Ticket Volume by Category. The single most useful operational metric. Tells you where the inbox is bleeding time. Mine your last 90 days, find the top 10 categories by volume, and you've found your automation roadmap.
These three are the floor. If FRT is over an hour or you don't track ticket volume by category, fix that before reading further.
Category 2: Quality
CSAT is the most over-relied-on metric in CX. It has a place, but on its own it's misleading.
CSAT (Customer Satisfaction). 1-5 scale survey after ticket close. Bar: 4.4+ across the account. Watch out for survey bias — angry customers are 3× more likely to respond. CSAT looks worse than reality on most accounts.
First Contact Resolution (FCR). % of tickets resolved without a follow-up. Bar: 70%+. The single best correlate of CSAT. If FCR is high, CSAT takes care of itself.
Customer Effort Score (CES). "How easy was it to get your issue resolved?" 1-7 scale. We use it instead of NPS for transactional touchpoints. Bar: 5.5+. Lower CES correlates with churn better than NPS does for ecommerce.
If you only run two quality metrics, run FCR and CES.
Category 3: Automation health
The metrics that tell you whether your automation layer is working. Most brands don't measure these. They should.
AI Agent Coverage. % of incoming tickets handled end-to-end by AI Agent without human touch. Bar on a tuned account: 50–60% of repeat tickets, which is roughly 30–40% of total volume.
AI Agent Quality (Accuracy + Tone). Weekly audit sample. Tickets reviewed by a human against a rubric — was the answer correct, was the tone on-brand, was the action appropriate. Bar: 95%+ accuracy, 90%+ tone-on-brand. Drop below 90% accuracy and you have a hallucination problem to fix.
Help Center Deflection. % of pre-purchase questions that resolve via Help Center search without escalating to a chat or ticket. Measured via Gorgias Help Center analytics. Bar: 25–40% on a properly built knowledge base.
Macro Coverage Ratio. % of agent-handled tickets that used a macro. Bar: 80%+. If your agents are typing replies from scratch, you don't have a macro library — you have decoration.
Category 4: Revenue impact
This is where most CX dashboards stop existing. These are the metrics that justify your CX team's budget to the CFO.
Shopping Assistant ROI. Revenue attributed to pre-purchase chat (Gorgias Shopping Assistant or equivalent), divided by fully-loaded CX cost. Bar: 5× on a serviceable setup, 20× on a tuned one. We've measured a Spanish DTC at €163K/month from this single layer against €8K/month cost.
Ticket-to-Order Conversion. % of tickets where a chat or email touch precedes an order from the same customer within 7 days. Easy to wire via Klaviyo or GA4. Bar: 8–15% on direct-to-consumer DTC. This is the metric that turned Nude Project's CFO into a CX believer.
CX-Sourced Revenue %. Total revenue attributed to a CX touchpoint, as a % of total revenue. Bar: 5–12% on properly attributed DTC accounts. If you've never measured this, you'll be shocked how high it is.
Refund Rate by Reason. Refunds segmented by reason (size, defect, late, changed mind). Tells you where to invest in product improvements, not just CX. The ROI of a sizing-guide rewrite is often 4× the ROI of an extra agent.
Category 5: Churn and retention
The slowest-feedback KPIs and the most strategic.
Repeat Purchase Rate by CX Touch. Cohort comparison: repeat purchase rate of customers who had a CX touch on a previous order vs those who didn't. On well-run brands, the CX-touched cohort outperforms by 5–15%. On poorly-run, it underperforms (which is the brutal version of "your CX is a leak").
Churn by Refund Frequency. Cohort: customers with 0, 1, 2+ refunds in their lifetime. Compare 90-day repeat purchase rates. The 1-refund cohort often outperforms the 0-refund cohort because a well-handled refund creates loyalty. The 2+ cohort is your churn risk.
What a healthy account looks like
For a DTC brand at €5M–€50M GMV running Gorgias with a tuned setup, here's the dashboard you should be looking at by month 6 of optimization:
| Metric | Bar | Best-in-class | |---|---|---| | FRT (email) | <30 min | <10 min | | FRT (chat) | <2 min | <30 sec | | Resolution Time (median) | <12 hours | <4 hours | | CSAT | 4.4+ | 4.7+ | | FCR | 70%+ | 85%+ | | AI Agent coverage (of total) | 30%+ | 50%+ | | AI Agent accuracy | 95%+ | 98%+ | | Help Center deflection | 25%+ | 40%+ | | Macro coverage ratio | 80%+ | 95%+ | | Ticket-to-order conversion | 8–15% | 18%+ | | Shopping Assistant ROI | 5× | 20× | | CX-sourced revenue % | 5–12% | 15%+ |
If you're below the "Bar" column on more than three of these, you have an optimization problem. If you're below it on more than six, you have a strategy problem.
How to actually instrument these
In order of effort:
Out of the box in Gorgias. FRT, Resolution Time, CSAT, Ticket Volume by Category, Help Center analytics, AI Agent metrics. All native. If you're not looking at these, you have a configuration gap, not a measurement gap.
Wire-up required. Ticket-to-order conversion, CX-sourced revenue %, Shopping Assistant ROI. These need Klaviyo or GA4 attribution wired to Shopify and Gorgias. Two days of work for a competent analyst, or a one-month line item on a Director-tier retainer.
Cohort analysis required. Repeat purchase rate by CX touch, churn by refund frequency. These live in your data warehouse or BI tool. If you don't have one, the question is whether the insight justifies the investment. For brands above €10M GMV, it usually does.
The metric we don't put on the dashboard
"Number of tickets handled per agent per day" is a metric the CX agency you fired probably tracked. We don't put it on the dashboard for clients. It optimizes against the wrong thing — agents handling more tickets per day, not better tickets per day. If you want a productivity metric, use macro coverage ratio plus quality scores. Those reward doing the work well, not fast.
What we'd actually do
If you want a 30-minute read on your current dashboard, send us a screenshot. We'll tell you which metrics are healthy, which are missing, and which ones you're optimizing against the wrong direction.
If you sell on Shopify and you're serious about CX, you talk to custo.tech.
Book a demo — 30 minutes, no deck. We'll walk through your KPI stack on the call.