Simple ways to measure customer satisfaction
Kazee team · · 6 min read
Ask the owner of a small business in Kenya whether customers are happy, and you will usually get a feeling. They remember last week's angry call and the thank-you message from a loyal client. A feeling is a fair place to start, but it cannot tell you whether this month was better than last month, or which part of the service lets people down.
You do not need a research team to do better. One short question, asked at the right moment and read with care, will tell you a lot. This guide covers three common scores, how to choose one, how to write the question, and what to do with the answers.
Decide what you want to find out
Start with the decision the number will help you make. A score nobody acts on is a waste of your customers' time.
- "Did we handle this case well?" You want feedback on one piece of service, soon after it happens.
- "Was it easy to get help?" You want to find the steps that make customers work too hard.
- "Would they stay with us and speak well of us?" You want a view of the whole relationship, not one case.
Pick one question for now.
Three common scores
Customer satisfaction score (CSAT) asks how satisfied someone was with their experience. IBM describes it as focused on a specific interaction, and gives the usual sum on a five-point scale: add up the ratings of 4 or 5, divide by the number of responses, and multiply by 100.
Customer effort score (CES) asks how easy or difficult an interaction was. Salesforce gives a common wording, "The company made it easy for me to handle my issue", and calculates the score as an average of the ratings. It is useful when you suspect customers are being passed around or asked to repeat themselves.
Net Promoter Score (NPS) asks how likely someone is to recommend you to a friend or colleague, on a scale of zero to ten. Bain & Company, which publishes the method, works it out as the percentage who score 9 or 10, minus the percentage who score 0 to 6. It says more about the relationship than about one case, so it is better asked now and then than after each ticket.
For most small support teams, CSAT is the right first step. It is simple, and a low rating points to one case you can look at.
Work it out the same way every time
Suppose you use a five-point scale. In one month, with made-up numbers, you get 50 ratings, and 40 of them are a 4 or a 5. Your CSAT is 40 divided by 50, times 100: 80%.
Scores are only comparable if they are worked out the same way. Tools can differ here. Zendesk's own help pages, for example, explain how its numeric scales map to good and bad: on a 1 to 5 scale, 4 and 5 count as good. Before you compare your number with last year's, or with anyone else's, check what counts as a good rating.
Published benchmarks mix different industries and scales, so your own trend tells you more.
Write a question people can answer
Pew Research Center, which runs surveys for a living, warns that even small wording differences can substantially change the answers. A few rules follow from that.
- Ask about one thing. "How happy were you with our speed and our staff?" is what Pew calls double-barrelled. A customer may love your staff and hate the wait.
- Keep the wording neutral. "How was our excellent service?" tells people the answer you want.
- Keep it the same. If you change the question, you can no longer compare this quarter with the last.
- Add one open question at most. "What could we have done better?" often tells you more than the number does.
- Ask soon. GOV.UK's service manual, written for government services, says to let people tell you what they think once they have finished.
Read the results honestly
The people who answer may not be typical. Pew's work on telephone polls found that a low response rate raises the risk of error but does not by itself make a result wrong. The key issue is whether people's views are linked to their decision to answer. The same worry applies to your survey. A customer with a bad experience may be keen to tell you, or may simply have given up on you.
Small numbers jump around. With 20 responses, one person changes your score by 5 points. Look at several months together before you call anything a trend.
A score shows where to look, not why. GOV.UK also suggests combining survey results with other sources, such as your helpdesk. Read the comments and open the tickets behind the low ratings.
Do not turn it into a league table. If an agent's pay or reputation rests on the score, you may find them asking customers for a good rating.
If you have no survey tool
You can start this week with a spreadsheet and a phone.
- Choose a small sample. Take five customers whose cases closed last week, and do not hand-pick the happy ones.
- Call or text them.
- Ask the same two questions. "How satisfied were you, from 1 to 5?" and "What could we have done better?"
- Write down the answers word for word.
- Look at the sheet every month.
Five calls a week will not give you a reliable percentage. It will give you honest comments, and those are often what a small team needs most.
Act on what you hear
- Follow up on every low rating. Ask a named person to contact the customer soon.
- Give each low rating a reason. Slow reply, wrong answer, fault not fixed, rude handling. Keep the labels few.
- Fix the most common reason first. Often the cause is a process, not a person.
- Review once a month. Fifteen minutes is enough: the score, the top reason, one thing to change.
Where Kazee fits
Kazee Desk is the support app in the Kazee suite, and it is Live. Email, SMS and live chat arrive as tickets in one queue. You switch on customer feedback in Settings, and write the message yourself. When a ticket closes, Desk sends the customer a link to a rating page with five choices, from Very bad to Amazing. A bad rating can reopen the ticket, alert the agent and department leaders, and add a tag. Reports show satisfaction by channel, priority and agent, next to volume, resolution time and first reply targets met.
It will not suit every need. The survey is tied to a closed ticket, so it cannot reach customers who never raised one, and Desk has no phone channel for customers who only call. It is a five-choice rating, not a general survey tool, so for an NPS or effort question you would need a separate form. Send yourself a test ticket on each channel you use, so you can see how the link arrives.
Whichever tool you use, the method is the same: ask one clear question, work it out the same way each time, and follow up on the low ratings. If you would like to see how this works in Kazee Desk, book a demo and we will show you around.
Sources
- IBM Think, What is CSAT and how to calculate it: definition and five-point calculation
- Salesforce, What’s your customer effort score: common question and calculation
- Bain & Company, Measuring your Net Promoter Score: question and calculation
- Zendesk help, Viewing your CSAT score and ratings: how numeric scales map to good and bad
- Pew Research Center, Writing survey questions: wording and double-barrelled questions
- Pew Research Center, Response rates in telephone surveys have resumed their decline (Courtney Kennedy and Hannah Hartig, 27 February 2019)
- GOV.UK Service Manual, Measuring user satisfaction (last updated 19 February 2021)