How do you get customers to answer a survey?

Ask soon, keep it short and act on every low score

Customers answer a customer satisfaction survey when it comes soon, takes a minute, reaches them in a channel they answer, and leads somewhere when the score is low.

In this guide, the survey moment is the event the survey asks about: a delivery, a closed support case, a renewal or a visit. The four conditions are:

• Soon: the survey comes shortly after the survey moment, while the customer still remembers it, and only once the moment is finished: the order has arrived, the case is closed.
• Short: one score question, the CSAT or NPS question with its scale, and one why question, the open question that follows: "What is the main reason for your score?" The answer to the why question is the reason, kept in the customer's own words.
• A channel they answer: a link in an email is easy to ignore. A short call, or a message in a channel the customer already uses, is harder to miss. Whatever the channel, a second attempt matters.
• It leads somewhere: every low score (NPS 0 to 6, or CSAT 1 or 2) gets a team follow-up, in which a person on your team contacts the customer with the reason in hand.

Then read the results in the right order. A survey that few customers answer gives you the score of the few who chose to answer. So the first number to check is not the score but the response rate: customers who gave a score divided by customers asked. The second is who answered. The score comes third.

The example month in this guide shows why. It is invented to show the method; it is not DRING data or industry data.

Why do so few customers answer a satisfaction survey?

Because a survey link is easy to ignore, and nothing visibly happens after it. The customer has to stop, open the link and fill in a form, with no sign that anyone will read the answer.

In this guide, an email link is a survey sent as a link in an email. A survey call is a short call that asks the score question and the why question.

The customers who do answer an email link are not a random sample of your customers. They tend to be the very pleased and the very annoyed: one wants to say thank you, the other wants to complain. The customers in the middle, satisfied but not delighted, have the least reason to click, so they stay silent.

Past surveys count too. A customer who gave a low score last time and heard nothing has little reason to answer the next survey.

With a response rate of 10%, the score describes 10 of every 100 customers you asked. It cannot tell you what the other 90 think.

How do you find out who your score is missing?

Run a split test on your own customers. Divide one month's customers at random into two halves. Ask one half in your current channel and the other in a new one, with the same questions. Then compare the response rate and the mix of Promoters, Passives and Detractors.

The example uses NPS: "How likely are you to recommend [company] to a friend or colleague?", on a scale of 0 to 10. Customers who answer 9 or 10 are Promoters, 7 or 8 Passives, and 0 to 6 Detractors. NPS is the share of Promoters minus the share of Detractors.

Example, not real data. An online retailer of furniture and home goods delivered 1,000 orders in one month. The team split those customers at random into two halves of 500. Both halves got the same NPS question and the same why question, the day after delivery.

• Half A, email link: sent the day after delivery, and one reminder email two days later.
• Half B, survey call: the day after delivery, up to two attempts, inside calling hours. The first attempt reached 200 customers and the second attempt 90 more, 290 in all. Of these 290, 250 gave a score and 40 said they had no time. The other 210 were never reached.

For each half, the response table counts who gave a score and who also gave a reason.

What the response numbers show:

• Response rate: the email link got a score from 50 of 500 customers (10%), the survey call from 250 of 500 (50%). On the email side, 450 customers gave no response. On the call side, 210 were never reached and 40 said no.
• The second attempt: 90 of the 290 reached customers (about 31%) were reached only on the second attempt. Our guide on when to call again, and when to stop covers when to place a second attempt and which hours to call in.
• Reasons: by email, the why question was an optional comment box, and 22 of the 50 customers who gave a score filled it in. On the call, 236 of 250 gave a reason.

Both halves got a second attempt (a reminder email for the email link, a second call for the survey call), and your own test should do the same.

Who answered matters as much as how many. The score mix table splits the 50 email answers and the 250 call answers into Promoters, Passives and Detractors.

What the score mix table shows:

• The middle: Passives were 6 of 50 email answers (12%) and 85 of 250 call answers (34%). In this example, the middle rarely clicked the link.
• Detractors: 20 of 50 email answers (40%) and 55 of 250 call answers (22%). Counted in customers, the call still found more of them: 55 against 20. That is 35 more unhappy customers who can get a team follow-up.
• Promoters: 24 of 50 (48%) and 110 of 250 (44%), nearly the same share.
• NPS: +8 from 50 answers, +22 from 250 answers. Neither number is the true score. The email score describes 50 of the 500 customers asked (10%), the call score 250 of 500 (50%). Do not read the 14-point gap as "calls make customers happier": part of it can come from the channel itself, as the next section explains.

Your own numbers will differ. The split test shows how many customers, and which ones, your current channel leaves out. It does not tell you which score is right. From then on, report the response rate next to the score, for each channel and each survey moment.

Will customers give a higher score on a call?

Some will. Saying a low number out loud feels harder than clicking it on a screen.

You cannot remove that effect, but you can keep it the same on every call:

• Use the same wording every time, and read the scale the same way.
• Accept any number without comment.
• Never ask the customer to reconsider a score.

Then compare call scores with call scores over time, never with scores from another channel. If you change channel, run both channels for a month and start a new trend line.

When should you send a customer satisfaction survey?

Soon after the survey moment, and only once it is finished. Ask while the customer still remembers it, never while a case is still open.

Here are example timings to adapt, not rules. The example month used the Delivery row.

Two rules apply to every survey moment:

• Ask only after the moment is finished. A delivery survey waits until the order has arrived. A support survey waits until the case is closed.
• Ask the same customer at most once in 30 days. A second attempt is part of the same survey, not a new one. The 30 days are an example; pick your own number and keep it.

What is NPS, and how is it different from CSAT?

NPS (Net Promoter Score) asks how likely the customer is to recommend you, on a scale of 0 to 10. CSAT (customer satisfaction score) asks how satisfied the customer was with one thing, on a scale of 1 to 5.

Many teams also ask NPS right after a survey moment, as the retailer in the example month did. That works, as long as you pick one score for each survey moment and keep it, so that each trend line stays comparable.

Whichever score you use, ask one score question and one why question. Add a third question only if someone will act on its answer.

What does a survey call sound like?

Short and plain: the agent names the survey moment, asks for a minute, asks the score question and the why question, and closes. Here is one with a customer who gives a low score, the day after delivery.

• Agent: "Hello [name], this is the AI assistant of [company]. Your order was delivered yesterday. Do you have a minute for two short questions about it?"
• Customer: "Okay, go ahead."
• Agent: "From 0 to 10, how likely are you to recommend [company] to a friend or colleague?"
• Customer: "A five, I'd say."
• Agent: "Thank you. What is the main reason for your score?"
• Customer: "One of the chairs arrived with a scratch, and nobody answered my email about it."
• Agent: "I'm sorry about that. Would you like someone from our customer service team to call you about the chair?"
• Customer: "Yes. Tomorrow morning is best."
• Agent: "Noted: our customer service team will call you tomorrow morning about the scratched chair. Thank you for your time. Goodbye."
• Customer: "Bye."

A score of 5 is a low score (a Detractor), so the agent offers a team follow-up and does not comment on the number.

What should happen after a low score?

A team follow-up, soon, with the reason in hand. An example rule to adapt: every low score gets a team follow-up within one working day, and the person who calls reads the reason before dialing.

Track how many low scores got a team follow-up, next to the score and the response rate. Customers who see that a survey leads somewhere have a reason to answer the next one.

Where does DRING come in?

DRING runs survey calls after an order, a support case or a renewal. The trigger event adds the contact to the calling list, and the call goes out within the window you configure after the event.

The agent asks your satisfaction or NPS question, then asks why, and keeps the reason in the customer's own words. Each survey has its own question set in your words, and the agent asks it the same way every time.

The score and reason go to the customer's CRM record with the transcript, and into Score, where CSAT and NPS are tracked over time. A low score can pass the call to your team right away, or open a follow-up task for them.

Survey calls stay inside the hours each country allows and retry by outcome. Reach runs them. After the call, an SMS can send a thank-you, a coupon or a support link.

For customers who prefer messages, Score can also ask by WhatsApp. Keep each channel on its own trend line.

Every DRING agent runs 1,000 to 10,000 simulated conversations, built for your company, before the first real call.

To see what survey calls change for your customers, run the split test from this guide on one survey moment. To hear a DRING call from the customer's side, request a callback. DRING calls you in two minutes, says who is calling and asks which survey moment you want to start with.