The premise seems cool: personalized surveys generated by AI, so that you can gather more specific data.
But after seeing the examples on the website, this seems like a waste of time. All of the questions asked are bog standard generic survey questions. And none of them really dug into anything personalized enough to get valuable feedback.
But more importantly, I don't think you could really utilize this at scale for anything quantitative. The whole point of "rate this on a scale of 1-5" questions are that you can learn about the distribution of answers. For example, is the average user happy? Is the product confusing?
But when you personalize each question, and therefore ask each user something different, you can't get a true average because the question itself changed. And similarly, you can't find long tails or measure the standard deviation (at least without introducing a lot of potential error). So at best, even if the "personalization" just picks from a list of pre-written questions, you've still reduced your sample size significantly.
1Flow.ai is in-app survey platform powered by AI. It uses AI to automatically generate fully customized survey questions and even multiple choices given a business / user research goal. Such as understanding funnel drop off reasons, or why users cancel their subscriptions.
Is AI coming to take our jobs? Or is it helping us becoming more efficient? What's your take.
But after seeing the examples on the website, this seems like a waste of time. All of the questions asked are bog standard generic survey questions. And none of them really dug into anything personalized enough to get valuable feedback.
But more importantly, I don't think you could really utilize this at scale for anything quantitative. The whole point of "rate this on a scale of 1-5" questions are that you can learn about the distribution of answers. For example, is the average user happy? Is the product confusing?
But when you personalize each question, and therefore ask each user something different, you can't get a true average because the question itself changed. And similarly, you can't find long tails or measure the standard deviation (at least without introducing a lot of potential error). So at best, even if the "personalization" just picks from a list of pre-written questions, you've still reduced your sample size significantly.