creating segments for daily visits of visitors and certain number of visits in a given time

[kr8tive]

hello

I am new to adobe analytics and I got stuck while trying to make new segments.

there is an event page with a daily check-in for a limited date. if you visit the page and log in, you get a certain prize every five times. e.g. 5th, 10th, 15th

I am trying to find out how often visitors participated in this event and how many people logged in 15th time.

I tried to make segments with

dimensions: logged in(prop), page name, URL, visit number, return frequency

metric: visits

I tried several versions but when I apply them and break them down, they don't seem right

the top bucket is the visitor 

<identical part>

logged in = logged in

and

page name = 

or

url = 

and these parts are where I tried different approaches

v1 and visit number = 5

v2 then within 1day visits exists

v3 hit bucket, then within 40 days visit number = 5 and return frequency = less than 1 day

what am I missing? thanks in advance

[Jennifer_Dungan · Accepted solution]

Honestly, looking at your criteria...

"Daily Check-In", "every 5 check-ins"...

I don't know if Adobe's Segment Builder can do this... I see what you were trying to with "Visit Number".. but those don't necessarily equate to one a day, since Visits have a 30 min timeout... Also, your logic would rely on the overall site visit to be completely sequential, which it may not be...  The user may have come to the site on visit number 5, but not hit the Check-In page... They may have come back later (Visit Number 6) and hit the check-in...

You also have to remember that Visit Number is an all time metric, the user may be on Visit 1049 when this promotion started.. so by your above logic, only completely new people to your site would be eligible.... 

Honestly, if I were trying to do this, I would get the data out of Adobe and use more advanced SQL to try and find the data...

I assume this promotion is already running, so it's too late to go back and add specific tracking.. or better yet, not run the promotion based on Web Analytics that people could be blocking or opted out of... Something like this should be coded and stored in a specific Database... Web Analytics is about trends, not absolute accuracy... 

[kr8tive]

thanks for your advice.

you are explanation helped me to realize that the visit number is not a proper dimension to use.

the promotion ran from this May to June and I was not involved back then. there was no specific tracking tag for the event and that's why I tried to analyze it with visits.

I tried with other combinations like visits = 5 and return frequency = less than 1 day (for sequential visits) but couldn't find the answer.

I guess it would be more logical to compare first visits and return visits. 

[Jennifer_Dungan]

Right, but Adobe's segmentation is less about calculating individuals, and more about general traffic / trends...

I think you will still have trouble trying to evaluate a correlation between visitor, visits (to the Check In page), and limit it to one check-in per day...

Analyzing the Raw Data feeds, though it will be a big job, would allow you to do a lot of that fancy logic and stitching....

Also, I hear you... the number of times people in my company have decided to run a contest then at the end of the contest come to me and ask me to pull data so they figure out a winner... 

Luckily, we have our raw data going into a data lake, so I usually pull from there so that we can layer in the criteria.

[kr8tive]

yeah, I will look into the raw data. I learned a lot from your answer. thanks. I hope I can become an Adobe master like you one day. until then take care.

[Jennifer_Dungan]

You're welcome and thank you...

And based on what I see above, you are on your way to becoming one... even if things don't work, the fact that you are trying different things (from different angles), and that you recognized that the data returned isn't quite right shows that you are thinking about the problem critically. You didn't just throw something together, and assume that the data returned was correct.