Running NPS and feedback surveys on WhatsApp
There is a hard constraint here that nobody mentions until build day: the standard 0–10 NPS scale has eleven options, and a WhatsApp list message allows ten. The workarounds are fine. Discovering it late is not.
Short answer. Reply buttons cap at 3. List messages cap at 10 rows across all sections combined. So a 0–10 NPS scale does not fit in one interactive message. Use a 5-point scale, split across two messages, or use a WhatsApp Flow. Then ask the open-text follow-up, which is where the value actually is.
The limits that shape the survey
| Element | Limit |
|---|---|
| Reply buttons per message | 3 (labels 20 chars) |
| List rows across all sections | 10 (titles 24 chars, descriptions 72) |
| List sections | 10 — grouping only, not extra capacity |
Eleven does not fit into ten. Your options:
- Drop to a 5-point scale. Fits a list comfortably, and for most businesses the extra granularity of 0–10 was never used in a decision anyway.
- Three buttons. Good / Fine / Not good. Highest completion rate of any format, because it is one tap with no list to open.
- Two messages. First ask 0–5 or 6–10, then the exact number. Two taps, preserves true NPS scoring, costs one extra message inside a free window.
- A WhatsApp Flow. A proper in-chat form when you genuinely need the eleven-point scale for comparability with historical data.
Recommendation. Unless you are benchmarking against an existing NPS series, use three buttons. The completion rate difference between one tap and open-a-list-then-tap is large, and a score nobody answers is worth less than a coarser score everyone does.
Category and cost
A survey about a specific purchase or visit the customer made is utility: ~AED 0.058, no marketing opt-in, not exposed to the frequency cap. Attach an incentive — "answer and get 10% off" — and it becomes marketing, at ~AED 0.183 and with opt-in required.
And remember that since 9 April 2025, a template submitted as utility that Meta judges to be marketing is approved as marketing rather than rejected. So the incentive does not just cost more; it can silently reclassify a template you thought was utility. Check what category your survey template actually received.
The customer's tap opens a free 24-hour window, so the follow-up question and the entire conversation after it cost nothing.
The follow-up is the product
The score is a number for a dashboard. The sentence after it is the thing that changes your business. Because the tap has opened a free window, asking costs nothing — so always ask, and ask specifically:
- Not "any other comments?" — that gets silence.
- To a promoter: "What should we not change?"
- To a passive: "What would have made it a 10?" This is the highest-yield question in the whole survey.
- To a detractor: "What went wrong?" — and then a human, within minutes.
Route detractors to a person, immediately
The operational advantage of surveying on WhatsApp over email is that a bad score arrives in a channel where you can respond in the same thread, in the same minute, for free. Most survey tools deliver a low score to a report nobody reads until Monday.
Set the rule explicitly: a detractor response assigns to a named human with an SLA in minutes. A recovered detractor is worth more than a promoter, and this is the moment where the recovery is cheap.
Timing and frequency
- Ask once, close to the event. Same day for a service visit, 2–3 days after delivery so the product has been used.
- One reminder at most, and only if they never opened.
- Suppress recent responders. Someone surveyed last month should not be surveyed again this month, regardless of how many transactions they had.
- Never survey during an open complaint. Obvious, and a frequent embarrassment.
Language matters more than wording
In the UAE, send the survey in the language the customer has been conversing in. A survey in the wrong language depresses response and skews the sample toward whoever happens to be comfortable in English, which is exactly the bias that makes feedback programmes misleading. Our measured 11.7% reply rate on a UAE account is a useful reference for what engagement looks like when the language matches.
Adjoltz builds these flows for UAE brands on Meta's official Cloud API — correct template categories, the buttons, and the inbox behind them. Messages at Meta's rate with zero markup, from $149/month.
Frequently asked questions
Can I run a 0-10 NPS survey on WhatsApp?
Not in a single interactive message. Reply buttons cap at three and list messages cap at ten rows across all sections combined, so an eleven-point scale does not fit. Use a five-point scale, use three buttons, split the question across two messages, or build it as a WhatsApp Flow if you need the eleven-point scale for comparability with historical data.
Is a feedback survey marketing or utility on WhatsApp?
Utility, if it asks about a specific purchase or visit the customer made. It becomes marketing the moment you attach an incentive such as a discount for responding, which triples the cost and requires opt-in. Since 9 April 2025 a template submitted as utility that Meta judges to be marketing is approved as marketing rather than rejected.
What is the best question to ask after an NPS score?
What would have made it a 10, asked to passives. It produces the most actionable answers of anything in the survey. Ask promoters what you should not change and detractors what went wrong. The customer's tap has already opened a free 24-hour window, so the follow-up costs nothing.
How quickly should I respond to a bad survey score?
In minutes, in the same thread. The advantage of surveying on WhatsApp is that a detractor is reachable immediately and for free, whereas most survey tools deliver a low score to a report nobody opens until the following week.