Customer Service Metrics: 8 KPIs Small Teams Should Track

The customer service metrics that matter for a small team fall into four groups: speed (first response time and resolution time), quality (first contact resolution and reopen rate), the customer's view (CSAT and CES) and workload (backlog and contacts per order). Track these eight in a one-page weekly scorecard, use medians rather than averages for times, and always read speed and quality together, because a team can get faster simply by closing cases too early.

You do not need a helpdesk system to start. A shared inbox and a simple log are enough for every metric below. For the survey-based ones, the CSAT and CES calculator turns raw answers into scores in seconds.

The four questions behind the metrics

Question	Metrics	Why it matters
How fast do we react?	First response time, resolution time	Waiting is the most visible part of service
Do we solve it properly?	First contact resolution, reopen rate	Unsolved problems come back as more work
How does the customer feel about it?	CSAT, CES	Speed and quality only count if customers notice
How much work is coming in?	Backlog, contacts per order	Shows whether the team or the business needs fixing

Pick at least one metric from each row. Any single metric on its own can be improved in ways that make service worse.

1. First response time

First response time = time of first human reply − time the request arrived

Report two numbers:

• Median first response time, so a few extreme cases do not distort the picture.
• Share answered within target, for example "88% answered within 4 business hours". This is the number that matches what customers experience against your promise.

Decide whether you count business hours or calendar hours and stick to it. An email that arrives on Friday evening and is answered on Monday morning is slow in calendar hours but fast in business hours. Automatic "we have received your message" replies do not count as a response; they only confirm receipt.

2. Resolution time

Resolution time = time the issue was resolved − time the request arrived

"Resolved" needs a definition: the customer has the answer, the replacement has shipped, the refund has been issued. Many teams pause the clock while they are waiting for the customer to reply, so that a customer who answers after a week does not make the team look slow. If you do that, say so in the report.

Again, use the median, and also watch the slowest cases. The oldest 10% of resolved issues often reveal process problems: approvals that take days, a supplier who does not respond, a question nobody feels responsible for.

3. First contact resolution (FCR)

FCR = issues resolved in the first contact ÷ all resolved issues × 100

FCR is the best single quality metric for service, because an issue solved at the first attempt saves the customer time and saves you a second or third contact.

The hard part is deciding what counts. Two practical methods:

• Agent tag: the person closing the case marks whether it was solved without further back and forth. Quick, but optimistic.
• Repeat-contact test: an issue counts as solved at first contact if the same customer did not get in touch about the same topic within seven days. More honest, but it needs a log with customer and topic.

Example: of 120 resolved issues in a week, 84 needed no further contact within seven days. FCR = 84 ÷ 120 = 70%.

4. Reopen rate

Reopen rate = resolved issues that were reopened ÷ all resolved issues × 100

A customer who replies "this still doesn't work" to a closed case reopens it. A rising reopen rate is the early warning that speed targets are pushing people to close cases before they are really done. Read it next to resolution time: if resolution time falls and reopen rate climbs, the improvement is not real.

5. Backlog and ticket age

Backlog = open requests at a fixed time, for example every Friday at 4 pm.

Add the age of the oldest open request and the number of requests older than your resolution target. Backlog is the metric to watch during busy periods: if it grows week after week, response and resolution times will follow, and so will complaints.

6. Contacts per order (contact rate)

Contact rate = customer service contacts ÷ orders (or active customers) × 100

This measures how many questions and problems your business creates. Fewer contacts per order usually means clearer product pages, better delivery information and fewer mistakes, not worse service. Group the contacts by reason ("where is my order?", "how do I…?", "wrong item") and the top reasons become a to-do list for the whole business, not just the service team.

7. CSAT (Customer Satisfaction Score)

CSAT = ratings of 4 and 5 ÷ all ratings × 100, on a 1–5 scale, asked right after an issue is resolved.

CSAT tells you whether the customer was happy with the outcome and the way it was handled. Our guide on how to calculate CSAT explains the top-2-box and average methods and the biases to watch, and the CSAT calculator does the arithmetic.

8. CES (Customer Effort Score)

Average CES = sum of ratings ÷ number of responses, usually on a 1–7 agreement scale ("[Company] made it easy to resolve my issue").

CES points at friction: having to repeat information, switch channels or chase an answer. It is the most useful companion to FCR. If you are unsure whether to ask about satisfaction or effort, see NPS vs CSAT vs CES. The same calculator has a CES mode.

NPS is deliberately not on this list. It measures the overall relationship, which service influences but does not own. The split is explained in customer success vs customer service.

Worked example: a weekly scorecard

A small online shop with two people handling email and phone keeps this scorecard. The figures are hypothetical.

Metric	Week 1	Week 2	Target
Requests received	140	150	–
Median first response time	3.1 business hours	1.8 business hours	–
Answered within 4 business hours	76%	91%	85%
Median resolution time	1.4 days	0.9 days	–
First contact resolution	71%	62%	70%
Reopen rate	5%	11%	under 6%
Backlog (Friday 4 pm)	18	12	under 15
Contacts per 100 orders	9.3	9.6	–
CSAT (responses)	87% (54)	81% (58)	–

At first glance, week 2 looks like a big improvement: answers are faster, cases close sooner and the backlog is down. The quality rows tell a different story. FCR fell, reopen rate doubled and CSAT dipped. After a team meeting about response times, the team had started sending quick first answers that did not fully solve the problem, and customers had to write again.

The fix was not to drop the speed target but to pair it with a quality rule: the first reply must either solve the issue or say exactly what happens next and when. A difference of six points in CSAT with around 55 responses per week can be chance, so the team watches the next few weeks before drawing conclusions from CSAT alone. FCR and reopen rate, based on every case rather than a sample, are the stronger signals here.

How to measure without a helpdesk

If requests arrive in a shared inbox, by phone and through social media, keep a simple log, one row per request:

Column	Example
Date and time received	Oct 12, 9:14
Channel	Email
Customer	Customer ID or email
Reason (fixed list)	Delivery status
First reply (date and time)	Oct 12, 10:02
Resolved (date and time)	Oct 12, 15:30
Contacts needed	1
Reopened?	No
CSAT score (if answered)	5

With this, every metric above is a formula: response and resolution times are subtractions, FCR is a count of rows with one contact, and contact reasons come from a pivot table. For complaints, the complaint log template adds the fields you need for escalation and follow-up.

Setting targets that help

• Measure four weeks before setting targets. Your baseline tells you what is realistic.
• Set targets per channel. Chat, phone and email have different expectations. One blended target hides problems in the slowest channel.
• Promise what you can keep. If you state "we reply within one business day", measure the share of requests that meet it. A promise kept builds trust; a fast average with frequent misses does not.
• Use survey scores for the team, not for individuals. CSAT from a few dozen answers per person swings too much to judge anyone fairly, and it invites people to ask customers for good ratings.

Common mistakes

• Tracking speed only. Fast, incomplete answers create more contacts and lower satisfaction.
• Averages instead of medians. One case that took three weeks can double an average response time.
• Counting automatic replies as first responses. It makes the metric meaningless.
• Changing definitions without saying so. Switching from calendar to business hours can halve your response time on paper.
• Ignoring why customers contact you. The cheapest service request is the one that never needs to be sent.
• Surveying only happy cases. Send CSAT after every resolved case, or at least a random sample, not just after the ones that went well.

Your next step

Start the log above today, or export the last four weeks from your inbox or helpdesk. Calculate median first response time, median resolution time, FCR and reopen rate, and send a one-question CSAT survey after every resolved case. After four weeks, enter the survey answers in the CSAT and CES calculator, set one target per row of the scorecard and review it every Friday.