Blog Best practice

10 customer service KPIs worth measuring in a small company

FRT, resolution time, CSAT, backlog and other KPIs: what they measure, how to read them and how to avoid misleading targets.

10 customer service KPIs worth measuring in a small company
In this article
  1. Why measure?
  2. The "I think" problem
  3. The benefits of metrics
  4. The 10 essential KPIs
  5. 1. First response time (FRT)
  6. 2. Average resolution time (ART)
  7. 3. Customer satisfaction score (CSAT)
  8. 4. Net promoter score (NPS)
  9. 5. Ticket volume
  10. 6. First contact resolution (FCR)
  11. 7. Ticket backlog
  12. 8. Agent utilisation
  13. 9. Cost per ticket
  14. 10. Self-service success rate
  15. Dashboard: which KPIs to display
  16. Operational view (daily)
  17. Tactical view (weekly)
  18. Strategic view (monthly)
  19. Common mistakes
  20. 1. Measuring too much
  21. 2. Numbers that say nothing
  22. 3. Optimising the metric, not the result
  23. 4. Not acting on the data
  24. 5. Unfair comparisons
  25. How to start
  26. Week 1: Choose 5 KPIs
  27. Week 2: Set up tracking
  28. Week 3: your starting line
  29. Week 4: the targets
  30. From month two: watch and act
  31. Conclusion

Customer service metrics should support operating decisions, not simply fill a dashboard. They need to show where waiting time accumulates, how consistent service quality is and whether the team is absorbing demand without compromising the customer experience.

This guide covers ten KPIs that are useful to SMEs, with formulas, interpretation criteria and the limits to understand before turning them into targets.

One thing about the numbers. You will not find industry benchmarks on this page, because the ones in circulation almost never state how they were collected, and a number with no method is not a reference. Where we give thresholds they are examples of targets you might set, to be replaced with your own as soon as you have two months of measurements. The comparison that counts is with yourself.

Why measure?

The "I think" problem

Without data:

  • "Customers seem satisfied" → But how many have stopped buying?
  • "We reply quickly" → How quickly? 1 hour? 1 day?
  • "The team works well" → Who performs best? Who needs support?

The benefits of metrics

With data:

  • Clear goals: "Cut FRT from 4 hours to 2 hours"
  • Informed decisions: "We need one more agent on Mondays"
  • Visible trends: "Satisfaction has been falling for 3 months"
  • a sum you can check: "since we changed tools, first response time has come down by this much", with the figure in front of it

The 10 essential KPIs

1. First response time (FRT)

What it is: Time between the ticket being opened and the agent's first reply.

Formula:

FRT = First reply timestamp - Ticket creation timestamp

Why it matters: The customer wants to know that someone is working on their problem. A fast reply (even one that does not solve anything yet) reduces anxiety.

Indicative targets:

ChannelCautious targetAmbitious target
Emailwithin a few working hourswithin the hour
Socialsame daywithin the hour
Chata few minuteswhile the person is still connected

These are not industry figures: they are commitments you choose and then have to keep. State only what you can meet — a missed promise costs more than silence.

How to bring it down: automatic acknowledgement, automatic assignment, alerts on stalled tickets, shifts built on your peaks. The seven levers in full are in how to reduce response times.


2. Average resolution time (ART)

What it is: Average time from ticket creation to full resolution.

Formula:

ART = Σ(Resolution time of all tickets) / Number of resolved tickets

Why it matters: It measures how fast you solve problems, not just how fast you reply.

Indicative targets:

ComplexityGoodExcellent

How to read it: the average on its own misleads, because one three-week case disappears inside a hundred one-hour ones. Look at the worst tail too, the ten per cent of tickets that take longest: that is where angry customers come from.

How to bring it down: a knowledge base the agent actually opens, canned responses, written escalation rules, training on the cases that keep coming back.


3. Customer satisfaction score (CSAT)

What it is: Percentage of customers satisfied after an interaction.

How it is measured: After the ticket is closed, ask:

"How satisfied are you with the support you received?" ⭐ Very dissatisfied → ⭐⭐⭐⭐⭐ Very satisfied

Formula:

CSAT = (4-5 star replies / Total replies) × 100

Why it matters: Direct feedback from the customer. You know whether the problem was resolved satisfactorily.

How to read it: the absolute value depends on how and when you ask the question, so it is not comparable with another company's. What counts is the direction: if it falls two months running, something has changed and the negative ratings will tell you what.

How to improve it:

  • Close the loop (is the problem really solved?)
  • An empathetic tone
  • Proactive follow-up
  • Analyse negative feedback

4. Net promoter score (NPS)

What it is: It measures customer loyalty and how likely they are to recommend you.

How it is measured:

"How likely are you to recommend Company to a friend or colleague?" Scale 0-10

Categories:

  • 0-6: Detractors
  • 7-8: Passives
  • 9-10: Promoters

Formula:

NPS = % Promoters - % Detractors

Where it comes from: the question, the 0-10 scale and the split between detractors, passives and promoters are the definition given by Fred Reichheld with Bain & Company in 2003. The link between NPS and growth is the claim the indicator was proposed on, not a fact settled once and for all: treat it as a thermometer for the relationship, not as a revenue forecast.

How to read it: a negative value means detractors outnumber promoters, and that much is a fact; the rest of the scale has no universal thresholds. Watch how it moves over time, with the question asked the same way every time.

Note: NPS measures the overall relationship, not a single conversation. Ask it at regular intervals, or after a meaningful moment in the relationship.


5. Ticket volume

What it is: Number of tickets received in a period.

Related metrics:

  • Tickets/day
  • Tickets/week
  • Tickets per channel (email, social, chat)
  • Tickets per category (billing, technical, shipping)

Why it matters:

  • Resource planning: do you need more agents?
  • Spotting problems: a spike of tickets on one topic?
  • measuring the effect of the knowledge base: are the repeat questions falling?

How to use it:

  • Trend over time (is it rising?)
  • Week-on-week and month-on-month comparison
  • Correlation with events (product launch, promotions)

6. First contact resolution (FCR)

What it is: Percentage of tickets resolved at the first interaction.

Formula:

FCR = (Tickets resolved in 1 reply / Total tickets) × 100

Why it matters:

  • For the customer: the problem is solved straight away, with less frustration
  • For the company: less work, more efficiency

How to read it: it depends enormously on what you sell. A simple service closes almost everything on the first reply, a technical product does not, and comparing the two numbers means nothing. Measure yours and see whether it rises.

How to raise it:

  • Complete replies (anticipate the follow-up questions)
  • Well-made templates
  • A complete knowledge base for the agent
  • Training on common problems

7. Ticket backlog

What it is: Number of tickets open or waiting at a given moment.

Why it matters: A growing backlog means the team cannot keep up with the volume.

How to monitor it:

  • Backlog at the end of the day
  • Weekly trend
  • Backlog per agent
  • Backlog by priority (how many urgent ones are queued?)

Warning signs:

  • a backlog growing week after week
  • tickets sitting untouched for days
  • a workload badly spread across people

The automations that prevent the second are described in the five helpdesk automation workflows.


8. Agent utilisation

What it is: How much time the agent actually spends on tickets versus available time.

Formula:

Utilisation = (Time on tickets / Working time) × 100

Why it matters: a low value means you are paying for time nobody needs; a very high one means there is no room left for a peak, and that people are working at their limit. The right threshold is not a universal number: it depends on how much of the day your team spends in meetings, training and on the phone.

The time not spent on tickets is needed for:

  • Breaks
  • Training
  • Meetings
  • Handling sudden peaks

9. Cost per ticket

What it is: How much it costs to resolve a single ticket.

Formula:

Cost per ticket = Total support costs / Number of resolved tickets

Costs to include:

  • Agent salaries (pro rata)
  • Software (helpdesk, tools)
  • Training
  • Infrastructure

Why it matters:

  • It justifies investment (a tool pays for itself if the cost per ticket falls)
  • Comparison between channels (which one costs you more at the same volume?)
  • budget planning

To assess total cost alongside features and workflows, see the Uppo and Zendesk comparison.

Note: the absolute value varies enormously by sector. What matters is the trend (is it falling?).


10. Self-service success rate

What it is: Percentage of customers who solve the problem on their own using the knowledge base or FAQs.

How to measure it:

Self-service rate = Knowledge base visits with no follow-up ticket / Total knowledge base visits

Or:

Ticket deflection = (Tickets before the knowledge base - Tickets after the knowledge base) / Tickets before the knowledge base

Why it matters: Every ticket avoided is:

  • Time saved for the team
  • An immediate answer for the customer
  • Zero cost

How to improve it:

  • articles written on the questions you genuinely receive
  • a search that finds things even with different words
  • links to articles inside your automatic replies
  • the list of searches with no results, read every week: that is your writing list

How to build all of this is in knowledge base and self-service portal.


Dashboard: which KPIs to display

Operational view (daily)

For the team leader, every day:

KPIWhat to look at
Tickets opened todaythe number, and how it compares with yesterday
Average first response time todayagainst the target you stated
Current backlogwhether it is growing or falling
Tickets stalled past your chosen thresholdthis has to be zero

Tactical view (weekly)

For the manager, every week:

KPIWhat to look at
Ticket volumethe direction, not the value
Average resolution timethe average and the worst tail
First-reply resolutionswhether it is up on last month
CSATwhether it falls two weeks running
Workload per personwhether it is spread evenly

Strategic view (monthly)

For the leadership team, every month:

KPIWhat to look at
NPShow it moves, with the question asked the same way
Cost per ticketwhether it is falling
Share resolved without contactwhether it is rising
Volume per channelwhere it is growing

Common mistakes

1. Measuring too much

Fifty indicators amount to none: nobody reads them.

Start with five, and add more once those five are under control.

2. Numbers that say nothing

"We answered 500 tickets this month!"

Fine, but were the customers satisfied? Did you solve the problems? Volume on its own says nothing about quality.

3. Optimising the metric, not the result

If you reward agents on FRT alone, they will reply quickly with "We're looking into it" and solve nothing.

Always balance speed (FRT, ART) with quality (CSAT, FCR).

4. Not acting on the data

Collecting metrics without acting on them is pointless.

Every KPI off target should trigger an action:

  • Low CSAT → Analyse negative feedback → Training
  • High FRT → More agents at peak hours
  • Growing backlog → Prioritisation or hiring

5. Unfair comparisons

Do not compare your FRT with that of an e-commerce giant. Compare:

  • With yourself (last month, last quarter)
  • with similar companies by size and sector, and only if you know how they measured

How to start

Week 1: Choose 5 KPIs

To begin with:

  1. first response time, the speed the customer feels
  2. CSAT, the customer's verdict
  3. volume, the workload
  4. backlog, the state of the queue
  5. first-reply resolutions, the efficiency

Week 2: Set up tracking

Most helpdesks track automatically:

  • FRT, ART (automatic timestamps)
  • Volume (ticket count)
  • Backlog (open tickets)

For CSAT, switch on post-ticket surveys.

Week 3: your starting line

Collect a few weeks of data before setting any target:

  • What is your current FRT?
  • What is your current CSAT?

You cannot improve if you do not know where you start from.

Week 4: the targets

Set targets from your own numbers, not from an article: one step on from where you are today, rather than a leap towards somebody else's figure.

From month two: watch and act

  • one reading a week, on the same day
  • look for directions, not single spikes
  • every value off target has to produce a decision
  • and when something improves, tell the team

Conclusion

You do not need 50 complex metrics.

You need a few key metrics, tracked consistently, used to make decisions.

Start with 5 KPIs. Measure for a month. Then ask yourself:

  • Where are we strong?
  • Where can we improve?
  • What do we do tomorrow to improve?

Numbers on their own change nothing. Actions do.

Uppo

Turn the method into customer service that works.

Centralise conversations, assign clear ownership and keep response times and quality under control from one helpdesk.

Start your free trial Talk to the team