Sequential Segment Head Scratch - Pages seen in different visits

[sviala]

Hello everyone!

I have been trying for some time now to build a sequential segment that would accurately show the volume of visitors that saw page A in one visit then returned to page A in  different visit. The goal being the show the visitor return rate to a page/group of page.

Now, if I stick to what is explained in the Adobe Analytics documentation for Segments on pages 30 and 31 (https://marketing.adobe.com/resources/help/en_US/analytics/segment/analytics_segment.pdf), I should only have to build a Visitor scoped segment, with 2 visits containers connected by a 'THEN' operator.

SEGMENT:

I get 85K visitors for the past 90 days with this segment.

Now my issue is, if I add the condition "AFTER 1 VISIT" next to the THEN operator, I get a lesser number of visitors (about 40K visitors). Although logically this should give me the same result of 85K visitors if my understanding is correct, since the original segment should have looked for visitors that saw the homepage twice in separate visits (so after 1 visit or more).

Additionally, I continued testing and found that if I switch the clause to "WITHIN 1 VISIT" I manage to get data too, for a total of 67K visitors. But wouldn't "WITHIN 1 VISIT" means that the visitors saw the same page twice in the same visit, which goes against what the original segment above should define? 

[Urs_Boller]

might be because your pageview is within a visit container. that means, the condition „after 1 visit“ starts to look in visit 3 and later.

i would change the sequence to „within 1 visit“ and you should get what you want

[sviala]

Thanks for pitching in!

I am not convinced I should use the "within 1 visit" clause as

1) I am still not sure what it is actually pulling,

and

2) I am also interested in capturing visitors that returned to the said page 2, 3, 4 or 5+ visits after initially seeing it. Not necessarily visitors that returned to it immediately in their next visit.

Plus, the "THEN After 1 visit" condition gives the same result whether the page name field is within visits or hits containers.

[Urs_Boller · Accepted solution]

let me try to explain what it does (might be difficult without a picture...)

your segment {(visit „pv“) then „after 1 visit“ (visit „pv“)}

it looks for visitors who visited the page. if found, the „after“ starts at the end of the visit container (since you put the page within a visit container). that means the condition „after 1 visit“ is actually skipping the next visit. the second visit container must be visit 3 or later.

by changing to „within 1 visit“ it looks only to the following visit - but this is not what you are looking for. remark to your last sentence: it doesn‘t matter if you have a visit or hit container afterwards, since by the time condition it looks for the whole next visit (regardless if you add a hit or visit condition) and this must result in the same.

if it doesn‘t matter in which visit the second pageview follows, just change the time condition to a simple „then“, making you‘re segment look like:

{(visit „pv“) then (visit “pv“)}

since you‘re first pageview is within a visit container, it starts to look for the second visit container right at the end of the first. and i think that is exactly what you‘re looking for. remark: you could skip the second visit container if you want, it will not change the result...

one more idea: if you skip the visit containers, you can use your first time condition and you should get the same results:

{„pv“ then „after 1 visit“ „pv“}

explanation is that here the time condition starts at the end of the first pageview, skipping one visit end (the current) and the looks for the second pageview.

i hope that helps

[sviala]

I agree the segment I should go with is {[visit, pv] then [visit, pv]}

Now I think the logic as to why {[visit, pv] then [visit, pv]} and {[visit, pv] then "after 1 visit" [visit, pv]} are not giving the same results is a bit different that what you suggested. I think the issue is with the "after 1 visit" as you mentioned but not for the same reason you described.

One of the hints the issue is not the 'skipping 1 visit' logic is the following:

I also would think that, logically, {[hit, pv] then "after 1 visit" [hit, pv]} would work the same way as {[visit, pv] then [visit, pv]} and it was one of the first tests I conducted. However, the segment {[hit, pv] then "after 1 visit" [hit, pv]} gives the same result as the segment {[visit, pv] then "after 1 visit" [visit, pv]}, which is less visitors than what the segment {[visit, pv] then [visit, pv]} gives (cf. screenshot below).

My theory is that the then "after 1 visit" condition might only filter for the visitors that exactly returned after 1 visit, not those that returned after 2, 3, 4, 5 visits. I think this is a loophole that some Adobe Staff people are also not aware of as I saw them recommending to apply the sequential segment {[hit, pv] then "after 1 visit" [hit, pv]} on other forum thread to achieve the same results I was trying to do.

When the right way to go about it is actually to do {[visit, pv] then [visit, pv]}

Ex:

[Urs_Boller]

thanks for the feedback and for additional testing. if you‘re right, this relly looks strange and should be investigated by adobe.

maybe someone of those people is following the thread and can further test it  

[Pablo_Childe]

Hi you mention your goal being:

"The goal being the show the visitor return rate to a page/group of page."

What if you just created segement

Single page

Page =X based by visit.

or for multiple Pages

Pages(s)=X or Y by visit(sequence isnt important as you want to account for both types)

Then pull a return frequency report apply your segment and see the frequency tiers of user types.

In this manner you account for both types of visits to a page(s)(primary or via another page). You can also then adjust date range to align to campaigns if desired.

regards

[Gigazelle]

So the reason why {[visit, pv] then [visit, pv]} and {[visit, pv] then "after 1 visit" [visit, pv]} are different is because the visit containers are both pulling in the full visit and continuing to look for sequential segment matches at the same time.

If you have the following segment:

Visit container where page equals A then visit container where page equals B

It will first scan for all hits where page equals A, then once it does, it will look for any hits that subsequently define page equaling B. Those two pages can be in the same visit, and if they are, their visit containers are redundant, as they'll both pull in the visit they were defined in. Page A and page B can either be in the same visits or in different visits with this segment.

If you have the following segment:

Visit container where page equals A, then after 1 visit, visit container where page equals B

That forces the segment to look for page A and page B in separate visits. This is why you would see a significant decrease with this segment than with the one above.

[sviala]

Hi Gigazelle

Kind of answering this late as I did not see further answers had been submitted.

While I understand the point you are making, this goes against what is explained in the official Adobe Analytics documentation (cf.https://marketing.adobe.com/resources/help/en_US/analytics/segment/analytics_segment.pdf , pg. 30/31):

As you can see it is explained clearly that {[visit, pv=A] then [visit, pv=B]} is the way to go to get the segment to look for page A and page B in separate visits.

Addtionally, it is mentioned later that mixed scope containers would also work for my purpose so {[visit, pv=A] then [hit, pv=B]} . I have tested it personally and do indeed get the same result as the other segment defined above.

Colleagues at my office have escalated this to AA representatives further up in the hierarchy that have confirmed that the "after 1 visit" is a legacy sequential logic that would only pull the visitors that saw page A in one visit, and page B exactly in the next visit (so it would not capture the visitors that saw page A and then after 2/3/4/x amount of visits saw page B).
Side note "after 1 week:" and any other timeframe-based sequential condition seem to be working fine however (so "after 1 week" would indeed look for the condition that happened after 1+ week, not exactly after 1 week - as it should).