Customer Service KPI Dashboard Example

Customer Service Performance Dashboard

On the first Monday in January, every blender, air fryer and espresso machine unwrapped over the holidays has a new owner, and some of those owners have a missing part, a blinking light and a phone. By 10am the queue reads 40 minutes, and the team lead is telling agents to take details and call people back. Each of those callbacks is a second contact the center will pay for, with a customer who is already annoyed.

Customers are less satisfied than they were. The American Customer Satisfaction Index fell to 76.1 in the second quarter of 2026, a decline that ACSI said had been surpassed only once this century, during the COVID-19 supply shortages. The same August 2026 release said customer complaints were at record levels.

The people answering are hard to keep. The Bureau of Labor Statistics counts about 2.7 million customer service representatives in 2025, with a median wage of $21.53 an hour in May 2025, and projects the occupation to shrink 5% between 2025 and 2035. It still expects about 289,500 openings a year, mostly to replace people who move to other jobs or leave the workforce. A support center is always training someone.

This example follows an invented company: a US maker of small kitchen appliances, sold through its own website and big-box retailers, with support split between two sites. Its monthly figures are modeled.

What is a Customer Service KPI Dashboard?

Overview

A customer service KPI dashboard shows whether customers who contact you get through quickly, get their problem fixed on that contact, how long the rest wait for an answer, how they felt about it, and what each contact cost to handle.

It starts in the queue. Service level is the share of calls and chats answered within a set number of seconds. It's the first number a contact center manager checks, because it shows what most customers felt before anyone said hello.

Then the contact itself. First contact resolution asks whether the customer had to come back about the same problem within seven days. Anything not fixed becomes a case, and average case resolution time measures how long those cases stay open. Overdue service requests counts the ones that have already blown the date the agent promised.

Then the verdict. Customer satisfaction (CSAT) comes from the short survey after a contact closes. Complaint rate counts the customers who were unhappy enough to escalate, ask for a supervisor or write in, per 1,000 contacts. Cost per contact sits underneath and tells you what the whole operation costs per conversation handled.

Who benefits from this dashboard?

1. VP of Customer Care: With both sites on identical definitions, a gap between them is a real gap. That's the starting point for every headcount and training decision.
2. Site managers: Service level and overdue service requests tell them by Monday lunchtime whether the week's schedule is short.
3. Workforce planning: When service level drops while contact volume is flat, the schedule missed. When it drops with volume, the forecast did.
4. Team leads: First contact resolution by team shows who is using the authority to fix things on the call and who is still logging callbacks.
5. Finance and product: Cost per contact explains the support budget. Repeat contacts about the same fault are often the first sign of a product problem, weeks before returns data catches up.

Cost per contact goes up when first contact resolution improves. Every repeat call you prevent drops out of the denominator, but the agents and the phone system cost the same. Show total monthly support spend next to it, or the site doing the right thing will look like the expensive one in the budget review.

Dashboard Example Overview

The company sells blenders, air fryers and espresso machines across the US. Support runs from two sites: an in-house center in Columbus, Ohio with about 40 agents, and a site in Tucson, Arizona, opened in 2023, with about 34. Together they handle between 39,000 and 63,000 phone, chat and email contacts a month, with the peak in January. Both sites report the same seven KPIs, split by site:

• First Contact Resolution: Contacts with no repeat contact from the same customer about the same issue within seven days, divided by contacts handled. Measured from the contact history, not from a box the agent ticks.
• Service Level: Calls and chats answered within 30 seconds, divided by calls and chats offered, leaving out those abandoned in the first 10 seconds. ICMI's 2015 guide to contact center metrics lists 80/20, 80/30 and 90/60 as moderate service level objectives. This center uses 80/30.
• Average Case Resolution Time: Average elapsed hours from a follow-up case being opened to it being closed. Only cases not fixed on the first contact count, so it measures the wait for everyone else.
• Overdue Service Requests: Open cases at month end that are past the date promised to the customer. The promise here is three business days.
• Complaint Rate: Formal complaints (supervisor requests, written complaints and public escalations) divided by contacts handled, times 1,000.
• Customer Satisfaction (CSAT): Survey responses scoring 4 or 5 on a five-point scale, divided by responses. The same one-question survey goes out after every closed contact at both sites.
• Cost per Contact: Total support center cost (agent pay and benefits, supervisors, telephony and software, facilities) divided by contacts handled.

Through late 2024 the two sites look like twins. In October 2024 first contact resolution is 70.4% in Columbus and 69.8% in Tucson, service level is 80.4% and 79.3%, and CSAT is 80.2% and 79.6%. The January 2025 holiday wave hits them equally: resolution drops to 68.1% and 67.2%, and the sites end the month with 326 and 329 overdue service requests.

At the end of March 2025, both sites get the same change. Agents can ship a replacement part or issue a credit of up to $150 during the contact, without waiting for a supervisor, and the agent desktop shows the parts diagram for the model number the customer reads out. The aim is simple: fewer contacts that end with "we'll call you back".

Columbus responds within weeks. First contact resolution reaches 74.0% in April, 79.3% in July and 80.9% in October 2025. Overdue service requests fall from 135 in March to 46 in September, the complaint rate from 7.2 to 4.5 per 1,000 contacts, and CSAT is 86.0%.

Tucson doesn't move. Its first contact resolution is 70.6% in April, 69.0% in July and 69.5% in October. On the center-wide line, resolution climbs from 70.6% in March to 74.5% in July, which reads like a change that half worked. Only the site split shows what happened: it worked fully in one place and not at all in the other.

The board also shows why. Tucson's service level, 79.7% in March, falls to 64.4% in May and 53.0% in July. The site lost 9 of its 34 agents that spring and summer. With customers waiting, team leads told agents to keep calls short and log anything complicated for a callback, and new hires didn't get the $150 authority until their 90th day. Average case resolution time rose from 57.9 hours in March to 75.2 in July, and overdue service requests from 135 to 305. The complaint rate reached 11.2 per 1,000 and CSAT fell to 73.2%.

The 2025 holidays make the gap impossible to miss. In December 2025 Tucson answers only 46.7% of contacts within 30 seconds and ends the month with 554 overdue service requests and a CSAT of 71.3%. Columbus, in the same month, has a service level of 74.2% and 95 overdue requests.

So Tucson gets its own fixes over the winter. Two hiring classes start, in November 2025 and January 2026. New agents get the $150 authority after a two-week certification instead of 90 days. Columbus takes Tucson's overflow chats for eight weeks while the new agents ramp up. By February 2026 Tucson's resolution is 70.8% and its service level 64.7%. By April it's 74.6% and 77.5%, and by September 2026 it's 78.8% and 81.0%, with 53 overdue requests and CSAT at 84.8%.

Two results weren't what the company wanted. Columbus's cost per contact rose from $9.83 in October 2024 to $10.58 in October 2025, though its monthly spend barely changed ($250,564 against $253,047). Contacts fell from 25,491 to 23,927 because the repeat calls stopped, so the same payroll was spread over fewer of them, and from April to July 2026 the figure sat just above APQC's $11.00 median. And Tucson, despite the recovery, still trails: 78.8% first contact resolution in September 2026 against Columbus's 82.1%, which is what a younger team looks like before it has seen every fault twice.

For the center as a whole, September 2026 against October 2024: first contact resolution 80.6% from 70.1%, service level 82.3% from 79.9%, complaint rate 4.7 per 1,000 from 7.9, CSAT 86.0% from 79.9%, and overdue service requests 90 from 289. Cost per contact is $10.86, up from $9.89 and still under APQC's median.

List of Customer Service KPIs

Service level and cost per contact have a published figure behind their targets: an ICMI example objective and an APQC median. First contact resolution has APQC medians by channel for context, though APQC doesn't publish how it confirms a contact was resolved. The others are this example's starting points, so replace them with your own once you have a year of history.

Why Customer Service Dashboards Matter

A contact center has more built-in trade-offs than most departments, and nearly every support number can be improved by quietly making another one worse. Push handle time down and first contact resolution follows it, because the hard problems get logged for a callback. Cut staff to bring cost down and service level drops, then complaints rise a month later. Close cases early to clear the overdue list and the same customers call back angrier, which shows up in the complaint rate.

Blending sites hides all of that. A center-wide average gives every manager a reason to feel fine: the good site lifts the line, the struggling site drags it down, and the total creeps forward. In this example the total said the new authority was working at half strength for most of 2025. The split said one site had fixed its problem and the other had a staffing problem that no policy could fix.

Turnover makes the split more important. A center that is always hiring always has a share of agents in their first few months, and those agents resolve less on the first contact. The rules that hold authority back from new hires decide how much less, which is why Tucson swapped its 90-day wait for a two-week certification.

So look at both sites in the first week after New Year. If one site's overdue service requests are climbing while it answers fewer than half its contacts within 30 seconds, it doesn't need a new script. It needs agents in seats, and the authority to fix the problem while the customer is still on the line.

For the internal version of this board, where the customers are employees and the tickets are laptops and passwords, see the IT dashboard example. For the sales side of the same holiday peak, see the e-commerce dashboard example.