How do you investigate user behavior beyond Mixpanel analytics to understand the why behind actions?

·Mar 12, 2026 02:38 AM·

Mixpanel is great for seeing what users did, but I often run into moments where the harder question is understanding why. For example: • users drop off during onboarding • people abandon checkout • someone cancels a subscription • a feature suddenly stops getting used You can see the event clearly in Mixpanel, but figuring out the reason usually means emailing users or scheduling interviews later. I’m experimenting with a different approach: Triggering a short AI-moderated voice interview when a product event happens (signup, cancel, drop-off etc) so you capture feedback while the experience is still fresh. Curious how others investigate these moments today? Do you usually: • send surveys • email users • schedule interviews • just rely on analytics?

3 comments

· Sorted by Oldest

Vlad S.
·
·
this is a very complicated thing you're touching on, I'll just give you some thoughts from the top, but there's probably a book to be written on the topic, and as many opinions as you have people in here you're correct that analytics alone will not tell you why users are doing what they're doing, they don't even tell you if they're all doing it for the same reason or not; what you can do for this is run A/B tests, you build a hypothesis as to why the users are doing the thing you're looking at, give half of your users a change that's supported by that hypothesis, and see if that changes your metrics... if it does, you're likely right about your hypothesis, if not, you're likely wrong qualitative data is just as important, if not more important than quantitative data, but it's a lot harder to obtain and interpret; for one, if you manage to get a hold of people that tell you why they're doing what they're doing, you can start to create hypotheses, which you can go test. this is great when it works, and pretty disappointing when it doesn't for example, if you're looking to see why users churn, you're not likely to get anyone to respond to your emails, surveys or phone calls after they've churned; they want nothing to do with you, you need to figure that out before they churn in the other direction, if you can find some superusers, they're usually excited to talk to you and give you constructive criticism about your product, sometimes giving you brilliant ideas about workarounds they are doing that you could easily eliminate or integrate anaytics will be very helpful in finding users to reach out to, either before they're about to churn or as you're looking to talk to super users
Vlad S.
·
·
from a personal standpoint, I absolutely hate it when software interrupts me from doing my thing to ask me talk to it and give it feedback; I would much rather have a feedback button that's always visible and easy to use, so that when I'm encountering an issue and want to complain to someone, I can do it
Mahesh K.
·
·
I totally agree with Vlad. I think the question almost becomes how can you help the user who is stuck before they get stuck which is not just about the product or UI but through proactive conversation.

Vlad S.
·
·
this is a very complicated thing you're touching on, I'll just give you some thoughts from the top, but there's probably a book to be written on the topic, and as many opinions as you have people in here you're correct that analytics alone will not tell you why users are doing what they're doing, they don't even tell you if they're all doing it for the same reason or not; what you can do for this is run A/B tests, you build a hypothesis as to why the users are doing the thing you're looking at, give half of your users a change that's supported by that hypothesis, and see if that changes your metrics... if it does, you're likely right about your hypothesis, if not, you're likely wrong qualitative data is just as important, if not more important than quantitative data, but it's a lot harder to obtain and interpret; for one, if you manage to get a hold of people that tell you why they're doing what they're doing, you can start to create hypotheses, which you can go test. this is great when it works, and pretty disappointing when it doesn't for example, if you're looking to see why users churn, you're not likely to get anyone to respond to your emails, surveys or phone calls after they've churned; they want nothing to do with you, you need to figure that out before they churn in the other direction, if you can find some superusers, they're usually excited to talk to you and give you constructive criticism about your product, sometimes giving you brilliant ideas about workarounds they are doing that you could easily eliminate or integrate anaytics will be very helpful in finding users to reach out to, either before they're about to churn or as you're looking to talk to super users
Vlad S.
·
·
from a personal standpoint, I absolutely hate it when software interrupts me from doing my thing to ask me talk to it and give it feedback; I would much rather have a feedback button that's always visible and easy to use, so that when I'm encountering an issue and want to complain to someone, I can do it
Mahesh K.
·
·
I totally agree with Vlad. I think the question almost becomes how can you help the user who is stuck before they get stuck which is not just about the product or UI but through proactive conversation.