Most ecommerce businesses track the metrics they can measure easily: total orders, revenue, average order value. Customer service metrics sit in a separate system and rarely make it into the same report. That separation is expensive. The measures that predict whether customers come back are in the support data, not just the sales data.
This guide covers 12 customer service metrics, explains exactly what each one measures, what good performance looks like, and how each connects to the revenue outcomes that matter for ecommerce growth.
Why customer service metrics connect to revenue
Every customer who contacts your support team has made a decision to put effort into the relationship with your brand. How that interaction goes determines three things: whether they complete the current purchase, whether they return for a second one, and whether they recommend you to someone else. Those three outcomes map directly onto conversion rate, repeat purchase rate and acquisition cost.
The causal chain is not complicated: slower responses produce lower satisfaction; lower satisfaction produces fewer repeat purchases; fewer repeat purchases require more acquisition spend to replace lost revenue. But the chain only becomes visible when you are tracking the right metrics.
How the metrics build on each other
leads to
Deflection
Less tickets needed
drives
Reference benchmarks
Benchmarks should be treated as reference points, not targets. A 70% FCR rate at a store selling bespoke furniture is a different achievement from 70% FCR at a store selling phone cases. Use the table below to identify outliers in your own data, not to set goals relative to averages that may not reflect your category.
| Metric |
Baseline |
Strong |
Red flag |
| First Response Time (FRT) |
Live chat: under 1 minute |
Email: under 4 hours |
Phone: under 30 seconds |
| First Contact Resolution (FCR) |
Above 70% |
Above 80% is strong |
Below 60% indicates a process problem |
| Customer Satisfaction (CSAT) |
4.0+ out of 5 |
4.5+ is strong for ecommerce |
Below 3.8 requires immediate investigation |
| Net Promoter Score (NPS) |
Above 30 is good |
Above 50 is strong |
Below 20 warrants investigation |
| Customer Effort Score (CES) |
Below 3 (on a 7-point scale) |
Lower is better |
Above 5 signals serious friction |
| Ticket Deflection Rate |
Above 20% with self-service alone |
40%+ with a chatbot |
Below 10% means self-service is failing |
| Resolution Rate (AI) |
40-60% for well-configured ecommerce AI |
Above 60% is strong |
Below 30% suggests knowledge gaps |
| Repeat Purchase Rate |
Varies by category; 25-40% is typical for general merchandise |
Track trend, not absolute |
Declining rate is a retention signal |
The 12 metrics
1
First Response Time (FRT)
Speed
Formula / How to calculate
Time between the customer’s first message and the first agent or AI reply.
What good looks like
Live chat: under 1 minute. Email: under 4 hours. The specific expectation varies by channel.
How it predicts growth
Speed of first response sets the tone of the interaction before content. A customer who waits 20 minutes for a first reply on live chat has already lost confidence in the experience. Stores with consistently fast first responses report higher CSAT even when resolution takes longer.
Primary levers
AI chatbot for first-contact resolution on common questions; pre-written responses for predictable enquiry types; staffing matched to peak enquiry hours rather than average hours.
2
First Contact Resolution Rate (FCR)
Resolution quality
Formula / How to calculate
(Tickets fully resolved in one interaction / total tickets) x 100.
What good looks like
Above 70% is a reasonable baseline for ecommerce. Above 80% is strong.
How it predicts growth
Every ticket that requires a follow-up contact costs double the support resource and creates a second opportunity for the customer to become dissatisfied. FCR is the single most direct measure of support efficiency and is strongly correlated with CSAT in post-interaction surveys.
Primary levers
AI that resolves complete enquiries rather than deflecting to another channel; comprehensive knowledge base so agents have the information to resolve on first contact; clear escalation paths that do not require the customer to repeat context.
3
Customer Satisfaction Score (CSAT)
Customer experience
Formula / How to calculate
Post-interaction survey, typically “How satisfied were you?” on a 1-5 or 1-10 scale. Calculate: (responses rated 4-5 / total responses) x 100.
What good looks like
4.0+ out of 5 is a reasonable baseline. 4.5+ is strong. Below 3.8 requires investigation.
How it predicts growth
CSAT directly measures whether the support interaction left the customer more or less likely to return. A pattern of low CSAT scores on a specific product or query type reveals a problem in the customer experience that is reducing repeat purchase rates. Track CSAT by query type, not just as an overall number.
Primary levers
Resolve on first contact; match tone and speed to the channel; follow up on unresolved issues proactively rather than waiting for the customer to chase.
4
Net Promoter Score (NPS)
Loyalty and advocacy
Formula / How to calculate
Periodic survey: “How likely are you to recommend us to a friend?” (0-10). NPS = % Promoters (9-10) minus % Detractors (0-6).
What good looks like
Above 30 is positive. Above 50 is strong for ecommerce. Scores below 0 indicate more detractors than promoters.
How it predicts growth
NPS is a leading indicator of word-of-mouth growth. A rising NPS correlates with organic acquisition through referrals and reviews; a falling NPS predicts churn and negative sentiment before it shows up in revenue data. Run NPS quarterly rather than continuously; it is a strategic signal, not an operational one.
Primary levers
Identify Detractors and contact them directly within 48 hours of the survey. Their complaints are your product and experience roadmap. Promoters are candidates for review requests and referral programmes.
5
Customer Effort Score (CES)
Friction
Formula / How to calculate
Post-interaction survey: “How easy was it to get your issue resolved?” on a 7-point scale (1 = very difficult, 7 = very easy). Lower scores are better.
What good looks like
Below 3 (on a 7-point scale) indicates low friction. Above 5 signals the customer worked harder than they expected to.
How it predicts growth
Effort is a stronger predictor of disloyalty than delight. Research published in the Harvard Business Review found that reducing customer effort consistently reduced churn in service contexts. In ecommerce, high-effort experiences (needing to call, email then call, repeat context to multiple agents) push customers to competitors on the next purchase.
Primary levers
Reduce channel-switching by resolving issues on the first channel the customer chose. AI that carries context across a conversation so the customer does not repeat themselves. Proactive updates (shipping delays, stock issues) that answer questions before customers have to ask.
6
Ticket Deflection Rate
Self-service efficiency
Formula / How to calculate
(Tickets not created because the customer found the answer themselves / total potential tickets) x 100. Estimated by comparing volume before and after self-service implementation, or by measuring knowledge-base views against contacts.
What good looks like
20-30% deflection through knowledge base alone; 40-60% with a well-configured AI chatbot handling first-contact questions.
How it predicts growth
Each deflected ticket is support cost avoided. For a team receiving 500 tickets per month at £8 fully-loaded cost per ticket, raising deflection from 10% to 40% reduces monthly support cost by £1,200 and frees agent capacity for high-value interactions that AI cannot handle.
Primary levers
Identify the top 20 questions driving ticket volume. Build specific knowledge-base articles for each. Implement AI that actively uses those articles to resolve incoming questions rather than directing customers to search for answers themselves.
7
AI Resolution Rate
AI effectiveness
Formula / How to calculate
(Customer enquiries fully resolved by AI without human intervention / total enquiries handled by AI) x 100.
What good looks like
40-60% is typical for well-configured ecommerce AI. Above 60% is strong. Below 30% suggests knowledge gaps or AI reaching its limits on your specific question types.
How it predicts growth
AI resolution rate directly measures the proportion of your support volume that does not require a person. A rising AI resolution rate means your support team handles a smaller proportion of the total contact volume, which increases their capacity for complex cases without additional headcount.
Primary levers
Feed the AI with accurate product data, current policies, and FAQs built from real question data. Identify the most common reasons for human escalation and address them with specific knowledge base content or AI training. Review AI responses that led to negative CSAT and use them to identify where the AI fails.
8
Average Handle Time (AHT)
Operational efficiency
Formula / How to calculate
Total time spent on tickets (including hold, resolution and wrap-up) / number of tickets handled.
What good looks like
Context-dependent. Track trend over time rather than comparing against a generic benchmark.
How it predicts growth
AHT measures agent efficiency. Declining AHT with stable or improving CSAT indicates agents are getting more effective, not cutting corners. Rising AHT with stable CSAT may indicate more complex enquiries or inadequate tools. Rising AHT with declining CSAT typically indicates process breakdown.
Primary levers
AI that handles the first layer of common questions; agent-assist tools that suggest responses for remaining contacts; clear workflow documentation so agents do not have to reconstruct process on each ticket.
9
Repeat Purchase Rate
Retention
Formula / How to calculate
(Customers who bought more than once in a defined period / total customers in that period) x 100.
What good looks like
Varies significantly by category. Consumables see higher rates (40-60%+); considered-purchase items lower (15-30%). Track the trend in your own data.
How it predicts growth
Repeat purchase rate is the downstream outcome most directly affected by the quality of the overall customer experience. A customer who received a fast, accurate, low-effort resolution to a post-purchase problem is more likely to return than one who had to chase a resolution. Track repeat purchase rate for customers who had a support interaction versus those who did not; the gap reveals the cost of poor support.
Primary levers
Post-purchase email sequence that maintains contact between purchases; loyalty programme that provides a reason to return; CSAT follow-up on any interaction rated below 4 to recover the relationship before the next purchase cycle.
Formula / How to calculate
Ecommerce churn is typically measured as customers who were active in a period and did not purchase again within a defined window. Formula: (customers who did not repurchase within window / total active customers in prior period) x 100.
What good looks like
Target a downward trend. The absolute number depends on category and purchase frequency.
How it predicts growth
Churn is the revenue already lost before it shows in the accounts. A customer who churns after a poor support experience does not send a complaint; they simply do not come back. High churn rates in cohorts that had support interactions (compared with cohorts that did not) indicate that support quality is a churn driver.
Primary levers
Identify churn predictors in your support data: Are customers who filed returns more likely to churn? Customers who had tickets unresolved after 48 hours? Use these signals to trigger proactive outreach before the churn window closes.
11
Support Cost Per Ticket
Operational efficiency
Formula / How to calculate
(Total support cost in period: staff, platform, infrastructure) / total tickets handled in that period.
What good looks like
Varies by channel and team. Track as a trend; look for cost per ticket falling as volume grows, which indicates the model is scaling.
How it predicts growth
Cost per ticket measures whether your support operation is scaling efficiently. If cost per ticket is flat or rising as volume grows, you are adding resources linearly. AI and self-service tools should reduce cost per ticket as volume increases by handling a growing proportion of contacts without proportional headcount growth.
Primary levers
Increase ticket deflection rate and AI resolution rate. Each deflected ticket and each AI-resolved ticket reduces the denominator cost without a proportional increase in the numerator.
12
Customer Lifetime Value (CLTV) by support segment
Strategic
Formula / How to calculate
Standard CLTV: average order value x average purchase frequency x customer lifespan. Segment by: customers with high CSAT support experiences vs customers with low CSAT experiences vs customers with no support contacts.
What good looks like
Customers who had positive support experiences should show higher CLTV than the average. If they do not, the support experience is not differentiating.
How it predicts growth
CLTV by support segment is the measurement that closes the loop between customer service investment and revenue. It shows whether the quality of support interactions actually affects whether customers stay and spend more. If CLTV is equivalent across high-CSAT and low-CSAT cohorts, your support quality is not the primary growth driver. If there is a meaningful gap, every improvement to CSAT is directly increasing CLTV.
Primary levers
Measure the gap. If CLTV is 30% higher for customers who received high-CSAT support interactions, that figure becomes the business case for every support quality investment.
Building a dashboard that shows what matters
Sample monthly dashboard view
FRT
28 min
-12%
vs last month
CSAT
4.3/5
+0.2
50 responses
FCR
71%
+4pp
past 30 days
NPS
41
+6
quarterly survey
Deflection
38%
+9pp
vs pre-chatbot
Repeat buy
29%
-2pp
last 60 days
Repeat purchase rate (flagged in red) signals a retention investigation is needed, even while other metrics are improving.
A useful support dashboard shows the operational metrics (FRT, FCR) alongside the customer experience metrics (CSAT, CES) and the business outcome metrics (repeat purchase rate, CLTV). Running them in separate reports means the connection between cause and effect is invisible.
Review operational metrics weekly (they change fast and respond quickly to changes in staffing and tools). Review customer experience metrics monthly (they require enough responses to be statistically meaningful). Review business outcome metrics quarterly (they reflect changes that compound over time and take longer to appear in the data).
The metric that changes fastest when you add AI support like AeroChat
Of the 12 metrics above, ticket deflection rate and AI resolution rate are the ones that respond most immediately to adding an AI chatbot. First response time also improves significantly when AI handles the first contact, since the AI responds in seconds regardless of staffing.
CSAT requires more time to show change because it depends on the AI actually resolving questions accurately, not just responding quickly. An AI that acknowledges questions immediately but sends customers to a human anyway will improve FRT without moving CSAT. The quality of the knowledge base and the accuracy of the AI’s product and policy data determine whether resolution quality improves alongside speed.
For ecommerce teams looking to implement AI support that moves these metrics, these Shopify AI chatbot tools show which platforms connect to product and order data — the difference between an AI that resolves questions and one that only responds.
See how AeroChat affects your support metrics
AeroChat connects to your Shopify product and order data and resolves customer questions autonomously. The 7-day trial gives you enough data to measure the change in FRT, deflection rate and resolution rate before committing to a plan.
Start the 7-day trial
