Your survey is not working and you already know it
NPS gives you a number. It does not give you a reason. Here is why forms fail and what to do about it.
2 min read
Somebody in your company owns a survey. It goes out every quarter. The response rate is 8%.
The results come back. Score went from 7.1 to 6.8.
Everyone stares at the slide. Nobody knows why.
Forms cannot ask "why"
That is the whole problem in one line.
A form asks what you thought of when you wrote it. It cannot notice that an answer was strange and ask about it. It cannot hear hesitation.
So you get:
- "How satisfied are you, 1 to 10?" → 6
- "Any other comments?" → (empty)
Six. Great. Six what? Six because of the price, six because your app was slow last Tuesday, six because they are just a six kind of person?
That comment box was your one chance and 90% of people skip it. Fair enough, it is work.
Why people do not fill it in
Not laziness. Three real reasons.
- It is boring. 14 questions with radio buttons is a chore, and it looks like a chore before they even start
- It feels pointless. They filled one in last year and nothing visibly changed
- It is not the right shape. Their actual opinion does not fit in "strongly agree to strongly disagree"
The irony is that most people are quite happy to tell you what they think. They just do not want to do admin about it.
Conversations get more, with less
Weird thing about talking. People say more in three minutes of conversation than in fourteen boxes.
Because a conversation can do the one thing a form never can. Follow up.
"It was fine I guess." "What would have made it better than fine?" "Honestly the setup took a week and nobody warned me."
There it is. That was never going to show up in a rating.
But you still need numbers
Yes. Obviously. You cannot take 200 transcripts to a board meeting.
So do both. Have the conversation, then score it against whatever you actually care about. Same criteria every time, so 200 chats become a chart, and the chart has the quotes sitting underneath it.
That is the bit that used to be impossible without a research team. It is not impossible anymore.
Try this instead of your next survey
Pick one question you have been guessing at for months.
Ask 30 customers, properly, in a real conversation. Score it. See what comes back.
I think you will be a bit annoyed at how obvious the answer was.
ColloAI does this. Free with your own AI key, about five minutes to set up your first one.
Read next
- Why customers cancel, and how to actually find outYour churn reason dropdown is lying to you. Here is what people actually mean when they click "too expensive".
- Exit interview questions that get honest answersMost exit interviews collect polite nothing. Here are the questions that get the real reason, and why people hold back.
- How to screen 200 applicants without losing your mindNobody reads 200 CVs properly. Here is what actually happens, and a way out that does not involve hiring the fastest applicant.