Skip to main content

Hi everyone, 

I need some help with understanding how to segment funnels. I need to measure the impact on funnel conversion, when users perform certain events vs not.

So for example, my funnel is ‘viewed products > viewed basket > submitted the order from the basket”, and I want to see what’s the impact of users engaging with some filters on the product page. 

For this, I’ve created two segments, and was particular about setting the “any time - within 10 mins” the same time frame as the overall completion time of the funnel, to make sure I’m accurately ONLY including users who engage with the filter within the same journey, rather than a later session. 

 

 

Nevertheless, my confidence is low because the data surprises me, and when I explore the journey paths, I see some users engaging with the filters in the “didn’t use filter” segment and vice versa. 

Also, I am not quite sure how to get accurate trend data for this funnel, for example, weekly. Choosing: any time - within the same week seems like the option, BUT then how would I interpret the data, for users that may do both within the same period: it’s very plausible that users may go through this funnel multiple times per week, and only sometimes engage with the filters. 

 

I’ve read through the documentation below but I am still unsure. 

Many thanks in advance! 

Raul 

 

Documentation 

https://help.amplitude.com/hc/en-us/articles/231881448#inline-behavioral-cohorts

and similar topics 

 

 

Hi Raul!

The way you’re using “perform within 10 minutes” is only looking for users who do/don’t use the filters within 10 minutes of their very first session start. So it’s not really what you want.

I would suggest instead that you create two funnels: (1) with the filters event added in as the second step (2) with the filters event as an “exclusion”. Once you have that you can compare the two, but you’ll have to think carefully about which conversion step to compare (basket to order?) to make a fair comparison.

Hi Timothy, 

Thanks for your help. I took your advice and I think it looks good. 

I had tried this approach before, but the thing I was missing was filtering the funnel by only step 2 and 3, which is essentially what I want, because if I included at the whole funnel, it would include users who drop off because they’re not engaging with the filter. 

The CVR between the two funnels is now close, and engaging with the filters improves conversion, which answers my original question! :) 

 

Also, thanks for clarifying the “within 10 minutes of their very first session start” … sadly this confirms my suspicion that there’s no way to use Amplitude segmentation for this kind of analysis… essentially, what I’d expect Amplitude to enable is a way to segment by “within x time from the start of the funnel / first event in the funnel” or something like that. 

 

Thanks again! 

Raul 

You’re welcome!

You will always have a challenge with this kind of analysis - those that engage with a feature will nearly always be higher intent users. The best approach is to A/B test with/without the feature, if that’s viable in the case.

Reply


Terms & ConditionsCookie settings