Expand my Community achievements bar.

SOLVED

Struggling to understand how I can create segments (time-specific cohorts) to get correct data in reports

Avatar

Level 1

Dear friends,

Please help me if you know or stir me in the right direction here:

 

What I'm looking to do:

 

I'm looking to build a report for an ecommerce store that shows Purchasers (Unique Visitors who Placed an Order) by month.

 

At that I need to show these purchasers as several distinct segments:

 

Loyalty program members who signed up for a loyalty account within the past:

 

Segment 1: 30 days

Segment 2: 3 months

Segment 3: 6 months

Segment 4: year

 

Example:
I think the report should look somewhat like this:

--------------------------------------------------------------------------------
Month   | Segment 1 | Segment 2 | Segment 3 | Segment 4 |
--------------------------------------------------------------------------------
Week 1 |    XXXX      |    XXXX      |    XXXX      |    XXXX     |
--------------------------------------------------------------------------------
Week 2 |    XXXX      |    XXXX      |    XXXX      |    XXXX     |
--------------------------------------------------------------------------------

Week 3 |    XXXX      |    XXXX      |    XXXX      |    XXXX     |
--------------------------------------------------------------------------------

Week 4 |    XXXX      |    XXXX      |    XXXX      |    XXXX     |
--------------------------------------------------------------------------------

 

Question/Challenge:

 

I can't seem to get correct data. How do I create these segments?

 

I'm struggling to understand how I can create these segments to get correct data. The challenge is that these segments are essentially time-based user cohorts.

 

Ex:

- Segment 1 needs to include all people who created a loyalty account within the past 30 full days (yesterday minus 30 days) - let's say there are 100 people in this segment
- Segment 2 needs to include all people who created a loyalty account within the past 3 months (yesterday minus 90 days) - let's say we have 300 people in this segment
- Segment 3 needs to include all people who created a loyalty account within the past 6 months (yesterday minus 180 days) - let's say we have 1000 people in this segment
- Segment 4 needs to include all people who created a loyalty account within the past 1 year (yesterday minus 365 days) - let's say we have 2000 people in this segment


Here is how i define the segments in Adobe Segment Builder:


Include (Visitor):

Date Range: Last 30 rolling days
AND
Metric: Loyalty Signup Success


- Then I need to see how many of the people from each of these segments (cohorts) made a purchase for a specific week (Week 1).


What I expect to see (numbers made up for the sake of example):

--------------------------------------------------------------------------------
Month   | Segment 1 | Segment 2 | Segment 3 | Segment 4 |
--------------------------------------------------------------------------------
Week 1 |       10        |       30        |      100        |       200       |
--------------------------------------------------------------------------------


Isn't it logical to assume that since each segment is of a different size, it will show a different number of PURCHASERS (unique visitors who placed order) for the same time range?

 

Ex:
Segment 1 - out of total 100 people 10 made a purchase in week 1
Segment 2 - out of total 300 people 30 made a purchase in week 1
etc ...

 

However here is what I see in my report:

--------------------------------------------------------------------------------
Month   | Segment 1 | Segment 2 | Segment 3 | Segment 4 |
--------------------------------------------------------------------------------
Week 1 |       10        |       10        |      10          |        10       |
--------------------------------------------------------------------------------

 

Same numbers for every segment! Although the number of users in each segment is different it shows the same number of PURCHASERS (unique visitors who placed order) for week.

 

What am I doing wrong?

Topics

Topics help categorize Community content and increase your ability to discover relevant content.

1 Accepted Solution

Avatar

Correct answer by
Community Advisor and Adobe Champion

Hi,

 

I think you will need to create your segments like this:

 

VISITOR [

    HIT container [

        Loyalty Signup Success exists

        AND

        Last 30 Full Days

    ]

]

 

 

I would also make ALL your segments based on Days.. rather than using the pre-built of "Last 3 Full Months" or "Last 6 Full Months", because if you are looking at a report today, so including the first 3 weeks of November, the date ranges based on "full months" will only go up to the end of October.

 

So you can use the prebuilt "Last 30 Full Days" and "Last 90 Full Days", then create equivalent "Last 180 Full Days" and "Last 365 Full Days"....

 

Or even better, since you are looking at the data per week, you might want to use "Last 4 Full Weeks", "Last 12 Full Weeks", "Last 52 Full Weeks" and then create custom "Last 24 Full Weeks" (so that all the data is grouped into Weeks.

 

 

Now, note, that once you add date ranges into your segments, your table will expand to fit the largest range by your segments, and unfortunately, if you use Date Ranges as your breakdown, you don't get the correct values... something about the cross section of the Date Ranges makes things go a little wonky.

 

Your best bet might be to use the standard "Week" dimension as your breakdown, sort descending and limit the display to 5 rows.... Your totals will of course be the total for the entire 6 months or year, etc... I can't figure out a solution to avoid that.... 

View solution in original post

5 Replies

Avatar

Correct answer by
Community Advisor and Adobe Champion

Hi,

 

I think you will need to create your segments like this:

 

VISITOR [

    HIT container [

        Loyalty Signup Success exists

        AND

        Last 30 Full Days

    ]

]

 

 

I would also make ALL your segments based on Days.. rather than using the pre-built of "Last 3 Full Months" or "Last 6 Full Months", because if you are looking at a report today, so including the first 3 weeks of November, the date ranges based on "full months" will only go up to the end of October.

 

So you can use the prebuilt "Last 30 Full Days" and "Last 90 Full Days", then create equivalent "Last 180 Full Days" and "Last 365 Full Days"....

 

Or even better, since you are looking at the data per week, you might want to use "Last 4 Full Weeks", "Last 12 Full Weeks", "Last 52 Full Weeks" and then create custom "Last 24 Full Weeks" (so that all the data is grouped into Weeks.

 

 

Now, note, that once you add date ranges into your segments, your table will expand to fit the largest range by your segments, and unfortunately, if you use Date Ranges as your breakdown, you don't get the correct values... something about the cross section of the Date Ranges makes things go a little wonky.

 

Your best bet might be to use the standard "Week" dimension as your breakdown, sort descending and limit the display to 5 rows.... Your totals will of course be the total for the entire 6 months or year, etc... I can't figure out a solution to avoid that.... 

Avatar

Level 1

tank you very much for the awesome tips Jennifer! I will try this!   

Avatar

Community Advisor and Adobe Champion

Good luck!

 

Time based segments used to be a lot easier before they changed the logic...

 

Previously, Visitor segments looked at ALL time visitor data, then you could constrain it with a date range, but it wouldn't impact the panel date range... when they added the "relative date range logic" they changed the order in which the logic was run, and now the date ranges inside the segments take precedence...  

Avatar

Level 1

Thanks again, Jennifer! It looks like with the adjustments to the segment definition and the switch to using the standard Week dimension it's giving me more realistic numbers (I'll do some validation of specific purchasers from each segment to see their signup date):

 

AntonEn1_0-1732299458717.png

 

 

 

 

Avatar

Community Advisor and Adobe Champion

Looking good... and yeah, Date Ranges using Months vs Weeks vs Days can be such a simple thing, but it also so easy to overlook.

 

I definitely agree to test as much as you can... in case something unexpected pops up in the calculation.