Analytics
1.9K members online now
Understand information in your reports and troubleshoot reporting issues such as self-referrals, (not set) data, and inaccurate information
 
Guide Me
star_border
Reply

Cohort analysis shows week 0 not 100%

Visitor ✭ ✭ ✭
# 1
Visitor ✭ ✭ ✭

I simply want to see 'user retention' in cohort analysis for all traffic from China. Before applying China segmentation, all week 0 shows 100% which make sense. After applying China segmentation, I expected all week 0 to still show 100% but turn out they are 80-90% in some rows. What does this mean? How does GA calculate the user retention after applying the segmentation?

cohort.jpgweek 0 not 100%

1 Expert replyverified_user

Re: Cohort analysis shows week 0 not 100%

Rising Star
# 2
Rising Star
While I am still very fuzzy on the cohort reporting to me it looks like your site does not have a full twelve weeks of data that includes vistors from China. One thing I read that sticks out to me was this statement "Remember, a cohort is just a segment based on a date. So when you apply a standard GA segment you’re introducing another layer of segmentation." in your case you are only looking at users from China that fall within the specified 12 week period since the only Cohort type we can currently use is Acquistion date I am assuming that the date a user from China first visited your site had zero sessions during the week of Feb21-Feb27, and only 82.35% of the cohorts in the following week had any visitors from China meaning you had several days with no vistors from China and the same for the third week of Mar 6 - Mar 12. but the last three weeks you had at least 1 vistor from China each of those days. Again this is a very fuzzy subject to me and it is my best if not totally wrong understanding of Cohorts.

Re: Cohort analysis shows week 0 not 100%

Rising Star
# 3
Rising Star
I also found this little snipit of information that might shed some light
"Because the Cohort Analysis Report is user-based, if you apply Segments based on sessions, you can get unexpected results that do not include 100% of users on Day 0 as you would expect.
For example, if a Segment is based on a condition like Location: Country exactly matches United States, which is session-scoped, then users whose Day-0 sessions originated in countries other than the United States are not included in the Day-0 column when you apply the Segment."

Re: Cohort analysis shows week 0 not 100%

Visitor ✭ ✭ ✭
# 4
Visitor ✭ ✭ ✭
Hey Brian, thanks for the info. I'm still not clear what does this mean exactly though. Does it simply mean if I have a total of 100 users and 10 users are not from US, I would see 90% on week one after applying US segmentation? or it means it will show day-0 less than 100% only when there are users who on week 0 is not from US but then have traffic from US on later weeks?

What even more confusing to me is after applying segmentation, the number of users is more than when I don't apply. Does this mean it switch to count users based on session? Note that I'm using the user id feature - https://support.google.com/analytics/answer/3123662?hl=en so may be that's related?

Re: Cohort analysis shows week 0 not 100%

Rising Star
# 5
Rising Star
The cohort report is user based when you created your segment what are you filtering by Sessions or Users. If possible can you do a screen capture of your segment and include it. There are lots of varibles here and it would help if it was narrowed down.

Cohort analysis shows week 0 not 100%

Visitor ✭ ✭ ✭
# 6
Visitor ✭ ✭ ✭

Hi Sarun, Wanted to understand if you got clarity on this ?

In the above case - would you say that the Week 1 retention is 33.93% or is it lower/more if we consider folks who may or may not be part of the segments?