community
cancel
Showing results for 
Search instead for 
Did you mean: 

Alteryx designer Discussions

Find answers, ask questions, and share expertise about Alteryx Designer.
Community v19.6

Looks aren't everything... But the latest Community refresh looks darn good!

Learn More

Pathing analysis into Tableau or some other visualization suite

Highlighted
Meteoroid

Hello Alteryx community,

 

I have a meaty one today. I'll try to give as detailed an explanation as to what I'm trying to accomplish, and I'm open to basically any solution. The core issue on the table is transforming complex pathing data into a format that can be repeatably translated into Tableau or some other appropriate visualization suite.

 

I have a set of digital click behavior, which I've extracted via SQL. Here's an example table:

 

TimeVisit IDHit NumberPageAction
12:59:55 AMA11Visit Start[NULL]
1:00:00 AMA12HomepageClick on Fun Page Link
1:00:05 AMA13Fun Page[NULL]
1:00:10 AMA14Fun PageClick on Funner Page Link
1:00:15 AMA15Funner Page[NULL]
1:00:20 AMA16Visit End[NULL]
1:59:55 AMA21Visit Start[NULL]
2:00:00 AMA22HomepageClick on Funner Page Link
2:00:05 AMA23Funner Page[NULL]
2:00:10 AMA24Funner PageClick on Filter: Funnest Stuff
2:00:15 AMA25Visit End[NULL]
2:59:55 AMA31Visit StartNotification
3:00:00 AMA32Funnest Page[NULL]
3:00:05 AMA33Visit End[NULL]

 

In this example, there are three visits.

  • Visit A1 records their launch event, then loads the homepage. Then, they click on a link that brings them to the Fun Page, and that page loads. Then, the click on a link that brings them to the Funner Page, and that page loads. Then, they exit the site.
  • Visit A2 records their launch event, then loads the homepage. Then, the click on a link that brings them to the Funner Page, and that page loads. Then, they filter the items on the page by those that are the Funnest. Then, they exit the site.
  • Visit A3 records their launch event which, unlike Visits A1 and A2. The visitor was prompted to begin the visit when they received a notification to their device, which then took them directly to the Funnest Page. After a couple of seconds, they exit the site.

The main complication is that the scale here is gigantic -- millions of visits per day. While one could hard-code the common ones, it would take a long time and wouldn't scale with changes to the site. It's also important to be able to identify common paths or sequences. Ultimately, this data will be converted into user journey maps, but I'd start with something as simple as a Sansky chart.

 

I'm wondering what, if anything, Alteryx can do to help push this towards a solution. 

Quasar
Quasar

This is a common use case that most web analytics platforms (Google Analytics, Adobe, etc) have built in.  To do it in Alteryx, you should start with the end visualization in mind - I've used a sequences sunburst in the past: https://bl.ocks.org/kerryrodden/766f8f6d31f645c39f488a0befa1e3c8

 

Given a target viz and layout, you just need to build a flow to turn your click stream data into that structure.  The version i linked to can use a CSV file where you just separate each node with a hyphen (see their notes on making the data file).  So in Alteryx, I would build a batch macro to take each visit at a time, and convert their series of steps into a single row in that format, then output of that macro step would go into the CSV output file.

 

As you note, you'll likely want to do some filtering or aggregation, but that can be an iterative process once you know what your data looks like and what you want to focus on.

Labels