Satisfaction survey over WhatsApp: how to do it + CSAT in practice

A satisfaction survey by email gets a single-digit reply rate — the customer doesn't even open it. A satisfaction survey over WhatsApp plays in a different league: it arrives on the channel where the customer already talks to you, gets answered with one tap, and can even fire automatically at the end of every interaction. This guide shows the formats that work, ready-made questions, how to calculate CSAT, and how to automate it without turning into spam.

Why WhatsApp (and not email or a form)

• A reply where the conversation already happened — the customer rates the support on the same screen where they were helped, with the context still fresh.
• Minimal friction — replying "9" takes two seconds; opening a form link with 12 fields, nobody does.
• A real-time signal — a low score arrives instantly, giving you a chance to turn the customer around before they become a detractor (or a 1-star review on Google).

CSAT, NPS, CES: which one to use?

CSAT (Customer Satisfaction Score) measures satisfaction with a specific interaction: "On a scale of 1 to 5, how was today's support?" It's the ideal metric for post-support on WhatsApp.

How to calculate CSAT: divide the number of satisfied responses (scores of 4 and 5) by the total number of responses and multiply by 100.

Example: 80 responses, 62 with a score of 4 or 5 → CSAT = 62 ÷ 80 × 100 = 77.5%. Market benchmark: above 80% is good; below 70% calls for investigation.

NPS (Net Promoter Score) measures brand loyalty: "On a scale of 0 to 10, how likely are you to recommend the company?" — promoters (9-10) minus detractors (0-6). Use it quarterly, for the relationship as a whole.

CES (Customer Effort Score) measures effort: "Was it easy to solve your problem?" Useful for processes (returns, technical support).

Rule of thumb: CSAT after every interaction, NPS every so often. Don't mix the two in the same message.

The 3 survey formats on WhatsApp

1. A simple question with a numeric scale (works on any WhatsApp):

"Before we wrap up: on a scale of 1 to 5, what score would you give today's support? Your answer helps us improve 🙏"

2. Interactive buttons (via the official API): the message arrives with buttons "😃 Great / 😐 Okay / 😞 Bad" — one tap, response logged, maximum reply rate.

3. A link to a form (when you need more detail): send the link after a score question, and only to whoever agrees ("Can you answer 3 quick questions? It takes 1 minute"). A cold, direct form is the format with the lowest reply rate.

Ready-made questions to copy

• Post-support (CSAT): "Was your request resolved? On a scale of 1 to 5, how would you rate the support from {{atendente}}?"
• Post-purchase: "Did everything arrive okay, {{nome}}? On a scale of 1 to 5, how was your buying experience with us?"
• Low-score follow-up: "Oh no, we're really sorry! What went wrong? I want to fix this personally today."
• High-score follow-up: "So glad you liked it! 💚 If you can refer us or leave a Google review, it helps a ton: {{link}}"
• NPS (quarterly): "On a scale of 0 to 10, how likely are you to recommend {{empresa}} to a friend? Just reply with the number."

When to send it (timing decides the reply rate)

• Right after closing the interaction — minutes, not days. With the context still fresh, the reply comes.
• Post-purchase/delivery: on the day of confirmed receipt.
• Business hours — a survey at 11pm is a spam report waiting to happen.
• A capped frequency: at most one survey per customer per period (e.g., 30 days), even if they open several conversations.

Automating it: a survey that fires on its own

Done by hand, the survey lasts a week — then the team forgets. The sustainable format is automatic, and here's how it works on a CRM like CRPRO:

1. The interaction is closed on the platform → a trigger fires the survey (a message with a scale or buttons).
2. The response gets logged on the contact — score, agent, date — with no manual spreadsheet.
3. A low score opens an alert for the manager to act instantly; a high score can trigger a Google review invitation.
4. The dashboard consolidates CSAT by agent, by queue, and by period — the metric becomes management, not curiosity.

In our support benchmark with real companies, the average manual response time is 15 hours — and slow support is exactly what tanks satisfaction scores the most. Measuring CSAT without attacking response time is spinning your wheels: with AI replying in 24 seconds and multi-agent support organizing the queue, the score climbs before the survey even asks.

Frequently Asked Questions

Is a satisfaction survey over WhatsApp allowed? Yes — it's a legitimate conversation with someone who's already your customer and just interacted with you. Keep the frequency low and offer an opt-out, as with any recurring communication.

Which scale should I use: 1-5 or 0-10? 1-5 (or 3 buttons) for post-support CSAT — simpler, more replies. 0-10 is the standard NPS scale. Don't invent your own scales: you lose comparability.

What counts as a good CSAT? Above 80% is healthy in most industries; 90%+ is excellent. More important than the absolute number is the month-over-month trend and the comparison between agents/queues.

Do I need the official API to run a survey? To send it manually, no. For interactive buttons and automatic post-support sends at scale, yes — through a platform connected to the official API.

What if the customer replies with text instead of a number? It happens — and it's gold: "it took a while but the guy was attentive" says more than a 3. A good flow logs the score when there's a number and routes the text for review.