8 Customer Satisfaction Metrics to Track (and How to Measure Them)

Measuring customer satisfaction is one of the clearest ways to tell whether your contact center is meeting its goals. Satisfied customers tend to stay longer, buy more, and recommend you, so these numbers connect directly to revenue and retention.

The challenge is that there are dozens of customer satisfaction metrics and KPIs out there, from first response time to customer health score to agent turnover. Not every team needs to track all of them. What you do need is a short list of metrics that reflect how customers actually feel about their experience, and a way to measure them consistently, ideally in real time.

In this post, I'll walk through the eight customer satisfaction metrics I think matter most for contact center and support teams, how to calculate each one, and how to connect them to the outcomes your business cares about, like customer retention.

What are customer satisfaction metrics?

Customer satisfaction metrics are measurements that show how happy customers are with your products, services, and support interactions. Some come straight from customer feedback, like CSAT and NPS surveys. Others come from operational data, like how long customers wait or whether their issue was resolved on the first contact.

The most useful set combines both. Survey scores tell you how customers say they feel, while operational metrics help explain why. Here's a quick reference for the eight metrics in this guide:

Why customer satisfaction metrics matter

The short answer: satisfied customers are easier and less costly to retain. Tracking satisfaction regularly helps you catch problems before they turn into lost accounts, and it shows your team where to focus.

They help you reduce churn

It doesn't take much for a customer to have a bad experience with a contact center. Long holds, a distracted agent, or getting bounced between departments all stick with people. When customers have plenty of alternatives, a few bad interactions can be enough for them to leave. Satisfaction metrics give you an early warning, often before a customer cancels.

They improve loyalty and retention

On the flip side, good support can help you keep customers for longer. Sometimes it's as simple as agents having previous interactions in front of them, so customers don't have to repeat themselves. Satisfied customers are also more open to upgrades and more likely to buy again, which is why satisfaction is closely tied to customer retention.

They drive word of mouth

Customers who have a negative experience may tell friends, post a review, or call you out on social media. Customers who have a great experience can become advocates, and word of mouth is one of the most credible and least expensive forms of marketing you have. Keeping existing customers happy also tends to cost less than acquiring new ones, so every point of satisfaction you protect has real financial value.

8 customer satisfaction metrics to track

We've covered why customer satisfaction matters. Now let's get into how to measure it with these eight metrics.

1. Customer satisfaction score (CSAT)

CSAT is the most common customer satisfaction metric, and for good reason. It gives you a direct read on how customers felt about a specific interaction, purchase, or experience.

How to calculate it: Ask customers to rate their satisfaction, usually on a 1 to 5 scale. Count the satisfied responses (typically 4s and 5s), divide by the total number of responses, and multiply by 100. So if 85 out of 100 respondents answer 4 or 5, your CSAT is 85%.

Most teams collect CSAT with a short survey right after an interaction, by phone, text, or email. In Dialpad Support for contact centers, you can set up CSAT surveys that run automatically after calls and add an open-ended question so customers can explain the score they gave. I'd also pair your scores with what customers actually say. With Custom Moments, you can track how often phrases like "price" or "technical problem" come up in calls, which can point to bigger issues worth passing to other teams.

Why it matters: CSAT is one of the simplest ways to see how satisfied customers are, and almost every company tracks it in some form. Just don't look at it in a vacuum. A high CSAT score next to very long call durations tells a different story, both financially and in terms of agent performance, than a high CSAT score on its own.

2. Net promoter score (NPS)

NPS shows whether your customers are likely to recommend you to others. It's an indicator of long-term loyalty rather than how one interaction went.

How to calculate it: NPS surveys ask one question: on a scale of 0 to 10, how likely are you to recommend this company to a friend or colleague? Respondents fall into three groups:

• Promoters (9 or 10)

• Passives (7 or 8)

• Detractors (0 to 6)

Subtract the percentage of detractors from the percentage of promoters to get a score between −100 and 100. The more responses you collect, the more confident you can be that the score reflects your customer base.

Why it matters: NPS puts a number on loyalty, which makes it useful for tracking how customer sentiment shifts over quarters. The downside is that a single question doesn't explain why someone gave the score they did, so follow up with detractors when you can and look at NPS alongside your other metrics. The difference between CSAT and NPS comes down to one interaction versus the overall relationship, which is why many teams track both.

3. Customer effort score (CES)

CES measures how easy or difficult it was for a customer to get what they needed, whether that's an answer to a question, a return, or a fix for a technical issue. In a contact center, it reflects how much work customers had to put in to reach someone and get their issue resolved.

How to calculate it: After an interaction, ask customers to rate a statement like "The company made it easy to handle my issue" on a scale such as 1 (strongly disagree) to 7 (strongly agree). Add up the ratings and divide by the number of responses to get your average score.

Why it matters: Customers who don't have to work hard are more likely to stay satisfied and loyal, while things like long waits and repeated transfers push effort up. CES helps you find those friction points so you can improve the customer experience. Staffing and training make a big difference here. Beyond being polite and attentive, the best agents anticipate what customers need before they have to ask. (I always try to hire people like that.)

4. First contact resolution (FCR)

FCR, also called first call resolution, tracks the share of issues your team resolves in the first interaction, with no callback, transfer to another queue, or follow-up ticket needed.

How to calculate it: Divide the number of issues resolved on the first contact by the total number of issues handled, then multiply by 100. Decide upfront how you'll confirm resolution, for example by checking whether the same customer contacts you again about the same issue within a set window, like seven days.

Why it matters: Few things frustrate customers more than having to call back about the same problem. FCR is closely tied to both CSAT and CES, so if those scores dip, FCR is one of the first places I'd look. Low FCR often points to gaps in agent knowledge, routing, or the information agents can see during a call.

5. Customer sentiment

Sentiment measures how customers feel during the conversation itself, based on what they say and how they say it. Unlike survey-based metrics, it doesn't depend on customers choosing to respond.

How to measure it: Track the share of interactions with positive, neutral, or negative sentiment over time, and look at how sentiment changes over the course of a call. In Dialpad, contact center sentiment analysis uses AI to flag sentiment from call transcripts in real time, so supervisors can see a conversation going sideways and step in while the customer is still on the line.

Why it matters: Sentiment helps fill in the story behind your other numbers. A call that ends in a resolution but had negative sentiment throughout is worth reviewing, because it may show up in your churn numbers later even if it never shows up in a survey.

6. Average speed of answer (ASA)

Waiting in call queues is nobody's favorite thing. If a customer is already unhappy, a long hold makes it worse. Average speed of answer tells you how long callers wait before an agent picks up. It typically starts counting after greetings and menu options finish.

How to calculate it: Divide the total wait time for answered calls in a period by the number of answered calls. Most teams track average speed of answer by interval, since a daily average can hide peak-hour spikes.

Why it matters: ASA is one of the clearest signals of whether you're staffed correctly. On my team, we watch wait times and call volume by time of day and day of week to spot patterns where we might be short-staffed or overstaffed. In Dialpad, queue performance reporting tracks wait times, call volumes, and service level achievement in real time, which makes those patterns easier to act on. If you take queries on chat or email too, measure those response times so you can balance agents across channels.

A well-planned IVR and automatic call distribution setup can also lower ASA by routing callers to the right place faster and answering simple questions, like business hours or return policies, without an agent. AI agents add another routing option: they can understand what callers need in natural conversation and send them to the right person without a menu. They can also resolve routine requests, like order status or appointment scheduling, on their own, which keeps your human agents available for the conversations that need them. If you have service level targets, for example answering a set percentage of calls within a set number of seconds, ASA feeds directly into your call center service level.

7. Abandon rate

Abandon rate measures the share of callers who hang up before reaching an agent, whether they're in the queue, listening to menu options, or sent to voicemail instead of a person.

How to calculate it: Divide the number of abandoned calls by the total number of inbound calls in a period, then multiply by 100. Many teams exclude very short abandons, such as hang-ups within the first few seconds, since those are often wrong numbers.

Why it matters: Every abandoned call is a missed chance to resolve an issue or make a sale. A rising abandon rate can mean you're understaffed, certain agents need support, or your menus are too long. Once you have the numbers, look at why customers are hanging up. Small changes can help: telling callers their place in the queue or a realistic wait time, offering a callback option, or using call routing to get callers to the right person from the start.

8. Customer churn rate

Churn rate is the percentage of customers you lose over a period. It's the outcome most of the other metrics on this list are trying to predict and prevent.

How to calculate it: Divide the number of customers lost during the period by the number of customers you had at the start, then multiply by 100. Leave out customers you gained during the period so new sign-ups don't hide losses.

Why it matters: Churn tells you whether satisfaction problems are turning into lost revenue. Tracking it next to CSAT, NPS, and sentiment helps you see which signals actually come before a customer leaves, which is where the next section comes in.

How to connect customer satisfaction metrics to retention

None of these metrics measures retention directly. CSAT, NPS, and CES tell you how customers feel; churn tells you what they did. The real value comes from connecting the two, so you can see which signals show up before a customer leaves. Here's how I'd approach it.

Measure at the right moments in the customer journey. Timing matters as much as the question you ask. A simple cadence might look like this:

Tie every score to a customer record. Scores sitting in a survey tool don't tell you much on their own. When CSAT and NPS responses attach to the customer's account in your CRM, you can compare satisfaction trends against renewals, upgrades, and cancellations.

Set alerts for low scores. A detractor score or a 1 out of 5 on CSAT is an opportunity to recover the relationship. Route those to a supervisor or account owner quickly so someone follows up while the experience is still fresh.

Segment your results. An average score can hide problems. Break results down by customer segment, account value, product, channel, and agent team, so a drop among your highest-value customers doesn't get buried in a healthy overall number.

Pair survey data with behavior. Look at satisfaction next to what customers actually do, like how often they contact support, whether issues get resolved the first time, and how sentiment trends across their conversations. Customers who report being happy but keep calling about the same problem may be at more risk than their scores suggest.

Why surveys alone miss most of the picture

One of the biggest challenges with survey-based metrics like CSAT is that many customers never fill them out. The ones who do tend to be either very happy or very frustrated, which can skew your results and leave you guessing about everyone in the middle.

That's where your everyday customer conversations come in. Every call already contains signals about how the customer feels, and AI can now read those signals at scale. Dialpad's AI CSAT detects customer satisfaction from call transcripts between agents and customers, so you get a score for eligible inbound and outbound calls instead of only the small share that end with a completed survey.

A score on its own only tells you so much, though. AI CSAT Explanations breaks each scored call down across seven categories: agent communication, agent empathy and professionalism, agent knowledge, agent speed and efficiency, completeness of resolution, customer sentiment, and operations. That way supervisors can see which part of a call pulled satisfaction up or down, and coach to it, instead of guessing from a single number.

This broader view can make a real difference. EF Education First used AI CSAT to infer satisfaction from all of its calls. As Mike Monteiro, Senior Manager of People Operations, put it: "Over the last three months with Dialpad, we've been getting over 97% in CSAT, from about 89% previously."

I'd still keep traditional surveys running. Direct feedback from customers is valuable, and pairing survey responses with AI CSAT gives you both what customers tell you and what their conversations show. Combined, they turn everyday interactions into customer intelligence you can act on.

How to measure satisfaction when AI agents handle conversations

More contact centers now use AI agents to handle routine requests like order status, appointment scheduling, and FAQs before a customer ever reaches a person. That changes what you need to measure.

The metric teams often reach for first is deflection or containment rate: how many conversations the AI agent handled without a human. It's useful, but on its own it can be misleading. A conversation the AI "contained" because the customer gave up isn't a win. I'd track satisfaction for AI-handled interactions with the same metrics you use for human agents, especially sentiment, CES, and whether the customer had to contact you again about the same issue.

Handoffs deserve their own attention too. When an AI agent escalates to a person, the customer shouldn't have to start over. Watch satisfaction on escalated conversations closely, because a clunky handoff can undo a good experience fast.

This is a lot easier when AI and human interactions happen in the same place. Dialpad Resolve, Dialpad's AI agents for voice and digital interactions, runs on the same platform as Dialpad Support for contact centers. When an AI agent hands a conversation to a human, the context from that conversation carries over, and you can review AI-handled and human-handled interactions in one platform.

Turn customer satisfaction metrics into better decisions

You don't need to track every customer satisfaction metric in existence. Start with the ones in this guide, look at them together rather than one at a time, and connect them to what customers actually do, like renewing, buying again, or leaving.

Whether you run a small business call center, a mid-sized support team, or an enterprise contact center, the goal is the same: understand your customers well enough to fix problems before they cost you. When calls, messages, AI agents, and analytics share one system, as they do on Dialpad's AI platform for customer experience, every interaction can feed the same view of how customers feel, and turn into insight your team can act on.