Customer Satisfaction Survey (CSAT): A Practical Guide for Calls and WhatsApp

A customer satisfaction survey (CSAT) is the simplest way to know whether a customer left a call or chat happy or annoyed. Many companies run a busy call center and a full WhatsApp inbox, yet when you ask "are our customers satisfied?" the answer is a feeling, not a measurement. This guide covers how to run a short CSAT survey after every call and chat, how to word the questions, when to send it, how to read results per agent and queue, and what to do with them.

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

Three common measures, each answering a different question:

CSAT: satisfaction with one interaction

CSAT (Customer Satisfaction Score) is a short question right after the interaction: "How would you rate our service today?" The customer picks from a scale, usually 1-5 stars. Because it is tied to a specific call or chat, you can trace it back to the agent, the queue and what happened. It is the best fit for service teams and call centers.

NPS: long-term loyalty

NPS (Net Promoter Score) asks "How likely are you to recommend us to a friend?" on a 0-10 scale. It covers the whole relationship: product, price, delivery and service. Useful for management and marketing, but it won't tell you whether a given agent handled a call well.

CES: how easy it was

CES (Customer Effort Score) asks "How easy was it to get your issue handled?" It exposes friction: too many transfers, repeated questions, a long IVR menu. Great when you want to fix processes rather than individual performance.

In practice: start with CSAT after every call and chat. Once it is stable, add an ease question (CES) to the same survey, and keep NPS for a separate periodic study.

One survey for calls and chats

A common mistake is one survey for the call center, a different one for WhatsApp, and separate reports for each. You can't compare them or see the full picture. Use one survey with the same questions and scale, whether the customer called, wrote on WhatsApp or used your website chat.

That is how the Klamna cloud call center works: in survey settings you create the survey, choose whether it applies to calls, chats or both, and write the intro and thank-you messages. After a call the customer rates the service in a short survey. In the omnichannel inbox the survey is sent automatically when the agent marks the conversation as resolved, as quick-reply buttons or a list, and the customer answers with a tap or by typing a number. You can pick the survey per WhatsApp channel and per website chat widget.

Why on resolve? Because that is when the experience has actually ended and the details are fresh. Chats closed by the bot alone, or with no assigned agent, don't get a survey, so ratings really reflect your team.

Wording your customer satisfaction survey: ready examples

Rule of thumb: one or two questions, three at most. Every extra question lowers completion. Keep options few and short so they show as WhatsApp buttons.

• CSAT: "How would you rate the service you received today?" Great / Good / Needs improvement. Or "Rate your experience from 1 to 5 stars."
• Resolution: "Was your issue resolved?" Yes / Partly / No. This separates a polite agent with a pending request from a case that is truly done.
• Effort (CES): "Was it easy to get what you needed?" Easy / OK / Hard.
• Comment (optional, chats): "Anything we could do better? Just reply with a message."

Sample intro: "Hi {customer name}, thanks for contacting {company}. Could you rate our service in two quick questions?" Thank-you: "Thanks for your time, {customer name}. Your feedback goes straight to our team."

Wording tips: use the customer's language; don't ask two things in one question; don't put the agent's name in the question (it invites polite ratings); keep the scale direction the same in every question.

When to send it, and when not to

• Right after resolution. The longer you wait, the more customers forget or rate the company as a whole.
• Once per conversation, not after every message or internal transfer.
• Not while the issue is still open. Wait until the agent actually closes it.
• On the official WhatsApp API the survey is a reply inside a conversation the customer started, so it goes out within the open customer-service window per Meta's rules; if the window has closed, it isn't sent. That is very different from WhatsApp marketing campaigns, which need approved templates and customer opt-in.
• If the customer replies with something else, Klamna cancels the survey and handles the message normally, reopening the chat for your team.

Reading results per agent and queue

The overall number on its own says little. The value is in the breakdown:

1. By channel: calls vs WhatsApp vs website chat. If one is consistently lower, look for the cause: waiting time, cold canned replies?
2. By agent: the survey report shows each agent's average for calls and for chats separately. Compare agents with themselves over weeks, not only with each other.
3. By team or queue: a complaints queue naturally scores lower than an enquiries queue; compare each with itself.
4. By call type: link results to wrap-up codes or chat labels to find which request types score low.
5. Written comments: read them all every week. One sentence can reveal a process problem, not a people problem.

Permissions matter: management sees everything, team leaders see only their team members, and regular agents don't see survey reports. Build your own views in the report builder and schedule them by e-mail.

Put CSAT next to other call center KPIs such as waiting time, abandoned calls and first-contact resolution. Low ratings with long waits point to staffing, not attitude.

Close the loop after a low score

A survey with no follow-up teaches customers their opinion doesn't matter:

1. Low score = follow-up. A team leader reviews the chat or listens to the call recording the same working day.
2. Contact the customer: "We saw your rating and want to understand what happened and fix it."
3. Open a ticket if the problem is still open, in your ticketing system, with an owner and a deadline.
4. Record the cause: agent, process, product or delivery. Each has a different owner.
5. Turn it into coaching: review the same call with the agent using quality scorecards, with evidence rather than impressions.
6. Thank happy customers too, and invite a review or referral without pressure.

Common mistakes

• Long surveys: ten questions after a two-minute call.
• Tying pay directly to ratings: agents start asking for "fives" or avoiding hard calls. Use CSAT for coaching, alongside other measures.
• Only reading the average: look at low ratings and comments on their own.
• Tiny samples: don't judge an agent on two ratings in a day.
• Changing questions every month: you lose comparability.
• Surveying unresolved issues: it irritates the customer.
• Ignoring comments: they are the most valuable part of the survey.

Getting started today

1. Write one CSAT question and one resolution question using the examples above.
2. Turn the survey on for calls and chats, and select it for each WhatsApp channel and website chat.
3. Agree with team leaders on daily review of low scores and a weekly report.
4. After a month, review by channel, queue and agent, and pick one improvement.
5. If you use a WhatsApp chatbot, compare chats handed to an agent after the bot with chats that started with an agent.

Klamna brings calls, WhatsApp, website chat, surveys, reports and quality into one system in Arabic and English. Build your package or start a free trial and pick the "Customer service center" template, which sets up the survey with a queue and canned replies. Want help designing it? Talk to our team.