What is a good CSAT score?

There's no generally good number, because the outcome depends heavily on your scale, your timing and which part of your customers responds. More useful is whether your number rises or falls under an unchanged method, and whether the response rate is high enough to read anything from it.

What's the difference between CSAT and NPS?

CSAT measures satisfaction with one specific contact moment and is asked right after that contact. NPS measures how likely someone is to recommend your company, so it's about the relationship with your brand, not that one conversation. For steering a customer service team CSAT is more usable, because you can tie it to a topic and a conversation.

Should I ask for CSAT after every conversation?

At a low volume that's fine. At larger numbers it's wiser to survey a portion, so customers who contact you often don't get a survey every time. What you mainly want to avoid is agents choosing when the question goes out, because then you're measuring a selection instead of your work.

CSAT · Customer Satisfaction Score

What is CSAT and what does the score actually measure?

CSAT is the percentage of customers who rate a contact moment as positive. The score is easy to collect and easy to misread.

In short

CSAT, in full Customer Satisfaction Score, is the percentage of customers who rate a completed contact moment as positive. After a conversation you ask something like "How satisfied are you with this help?" on a scale of 1 to 5, and divide the number of 4 and 5 answers by the total number of answers. So CSAT measures the experience of one specific conversation, not satisfaction with your company as a whole.

Calculation

CSAT = (number of 4 and 5 scores / total answers) x 100

On a 1 to 5 scale, 4 and 5 count as positive. On a 1 to 10 scale, 8 through 10 count. What matters is that you fix the threshold and never move it again, because last year's CSAT with a different threshold is a different number.

When do you ask the question?

The moment determines the answer. Ask right after the conversation and you measure the experience with the agent and the speed. Ask three days later and you measure whether the problem was actually resolved. Both are useful, but they're different questions and the outcomes aren't comparable.

For most teams right after closing works best, because the conversation is still fresh and the response rate is higher. If you're specifically measuring whether a repair or return was handled well, a few days later makes more sense.

Why a high CSAT says nothing about the customers who don't respond

Customers who bother to give a score are rarely the average customer. People who are very satisfied or very angry fill it in more often. The large group who thought it was simply fine doesn't respond. So you mainly measure the extremes.

So always look at the response rate alongside the score. A CSAT of 94 percent from 40 answers out of 2,000 conversations says something different from 88 percent out of 900 answers. The second number is usable, the first is an impression.

What do you do with a low score?

A score without an explanation can't be fixed. So always put an open question behind the score, and read those answers weekly. That's where what went wrong is written, and it's the only thing you can actually change.

Also split the score by topic and by channel. An average across everything hides the fact that return questions score structurally lower than delivery time questions. That split points to where you should start.

CSAT, NPS or CES: which score measures what?

	CSAT	NPS and CES
Question	How satisfied are you with this help?	Would you recommend us (NPS) or how much effort did it take (CES)?
Concerns	This one conversation	Your brand (NPS) or the process (CES)
Moment	Right after closing	Periodic (NPS) or after an action (CES)
Steerable by the team	Yes, by topic and channel	Limited (NPS) to good (CES)

If you want to know where your customer service should improve, use CSAT or CES, because those attach to a conversation you can look up. NPS says something about your brand and barely moves from a better answer to a shipping question.

Where it goes wrong

Comparing the score with an industry average

Industry averages for CSAT circulate widely, but almost never with the measurement method attached. A score on a five-point scale right after a chat is not the same as a ten-point score three days after a return. So compare your number with your own number from last month, not with a figure from a report whose setup you don't know.

Judging agents on their personal CSAT

Whoever gets the hard conversations scores lower. Tie assessments to CSAT and the temptation appears to pass difficult conversations on or not send the survey. The number rises, the service doesn't.

Only looking at the average

A CSAT of 85 percent can mean nearly everyone is satisfied, or that 85 percent is enthusiastic and 15 percent is furious. Look at the distribution of scores, not just the final percentage.

What to do this week

Put an open question behind your CSAT question if there isn't one, and split the score by topic. On Friday, read the twenty lowest scores with their comments. Almost always one or two recurring causes come out, and those are more concrete than the percentage itself.

What this looks like in a system

In Cuego the CSAT question hangs on closing a conversation, in every channel. The score lands on the ticket and on the customer card, so at the next contact you can see this customer was dissatisfied last time. In the analytics you can split the score by topic, channel and agent, with the response rate shown alongside.

Analytics