Delivery and read counts tell you a campaign arrived. They do not tell you whether the customers who acted on it, booked the service or claimed the offer came away happy. WappBlaster can ask that question in the same chat, but only if you ask the right people, at the right moment, and do something with the answer.
Keep the WhatsApp CSAT survey out of the broadcast
The tempting shortcut is to add "Rate us 1 to 5" at the bottom of the offer itself. It fails twice. People who ignore the offer will not rate anything, so you hear only from the already engaged, and those who do answer are rating the promotion, not the experience that followed it. A Diwali sale message with a rating button mostly measures whether people liked the discount.
Satisfaction is about what happened after the click: the order, the delivery, the appointment, the reply from your staff. Ask after that has happened.
Choosing the sample for a CSAT survey
The useful population is narrower than the send list. Start from contacts who did something:
- Replied to the campaign and had a thread handled by your team.
- Clicked the call-to-action and completed an order or booking.
- Tapped a quick-reply button that started a bot flow.
From that group, take a random slice instead of everyone. Tag sampled contacts with the campaign name and a "csat-asked" tag carrying the month, and exclude anyone with that tag from the next draw. Segments built from tags and behaviour make this exclusion automatic rather than a spreadsheet exercise. Random matters: if agents choose whom to survey, they choose the friendly ones.
Sending the CSAT survey while the WhatsApp window is open
When a customer messages you, a 24-hour customer service window opens, and inside it you can send free-form messages, including interactive buttons and lists, without a template. That is the cleanest place for the question: right after an agent marks a thread resolved, or right after the bot confirms an order.
If the moment you care about comes days later, such as after a courier delivers, the window will have closed and the question needs an approved template. Budget for that template's approval and its Meta charge, and keep it plainly about the transaction the person just had.
Writing a WhatsApp survey question people answer
Reply buttons are capped at three, so a five-point scale needs a list message (up to ten rows) or a three-step scale on buttons. For most businesses three buttons work better on a phone:
Illustrative: "Thanks for ordering with us, Priya. How did it go?" with buttons Very good, Okay, Not good.
- One question only. A follow-up "anything to add?" is optional and only sent to Okay or Not good.
- Name the thing being rated: the delivery, the stylist visit, the support chat.
- Build it once in the no-code chatbot builder so the score is stored against the contact as a field, not lost in the chat text.
The first hour after a bad WhatsApp CSAT score
A "Not good" tap should hand the thread to a person immediately, with the full conversation visible, an internal note saying which campaign it came from, and a "detractor" tag. Assign it to a senior agent rather than back to whoever handled the original chat. The reply should ask what went wrong and offer something concrete: a callback, a replacement, a refund check. Mark the thread resolved only when the customer has responded.
Reading CSAT survey scores against the campaign
| Pattern you see | Likely cause | What to change |
|---|---|---|
| High read, low reply, decent CSAT | Offer weak, service fine | Rework the template, not operations |
| Strong replies, poor CSAT | Inbox overloaded after the send | Stagger sends or staff up for launch day |
| Poor CSAT tied to one city | Local fulfilment problem | Fix the branch before the next city send |
| Rising opt-outs near survey sends | Surveying too often | Shrink the sample, lengthen the gap |
Put CSAT beside the delivered, read, replied and opted-out figures in per-template campaign analytics so a template's score is never judged in isolation.
When a WhatsApp CSAT survey is the wrong choice
Skip it when the campaign produced almost no interactions, since a handful of scores is noise. Skip it for people who opted out or complained about frequency. And skip it if nobody is assigned to read the detractor queue; collecting complaints you never answer is worse than not asking. Because a survey template sent after the window closes is billed by Meta per delivered message and WappBlaster passes that on at cost on top of its ₹7999 annual fee ($200 for accounts outside India), a smaller sample is also the cheaper one.