Best Practice: Classifications vs. Separate Variables

[jeff_bloomer]

I'm so surprised there are no other questions out there like this.  So, here it goes:

I'm wanting to understand the best practice for capturing a group of related user preferences and then turning around and gathering the metrics for them.  Let's use the following as an example.  These are not the real world values.  Say we have a section on our site that lists a bunch of sports publications, and each have an indicator for what sports are covered within.  This does NOT work as a filter, it simply displays an icon or more representing the the sports genre represented within the publication.

sportsPreferences:

• baseball
• soccer
• football
• hockey
• basketball
• volleyball

Here's the scenario: When the user logs in, all of the preferences are "turned on" by default.  The user may then go to a page where they can modify the list of preferences so that the full list of publications are updated to display the specific sports preferences the user selected.  Important to note is that none of the displayed products are filtered out of the list.  They simply update to display only the specific sports genres where they are relevant to the options the user elected to keep.

From this scenario, the business wants to understand how many users are turning off at least one choice, and then the ranking of the items left on vs. the ranking of the preferences turned off.

My initial thought is to capture all choices as a single string of true/false values and then separating them by using a classification rule.  However, I'm wondering if that's the best way to go.  Has anyone out there had a similar situation where you might be able to recommend a better approach?

Thanks in advance!

[Andrey_Osadchuk]

jeff_bloomer try to use the s.products variable with an eVar and two events (one for turned off, another for turned on).

;sport-preferences;0;0;event1=1;evar1=baseball,;sport-preferences;0;0;event2=1;evar1=soccer,...

[jeff_bloomer]

Interesting concept, though, I'm already using the products variable for different data.  I'm leaning toward using a list variable at the moment.  Your suggesting is why I'm a big supporter of the linked suggestion below.  It would be a lot easier for us to pick and choose the variables and then decide what attributes we would like to use to gather the data we need.

Combine Traffic Variables (sProps) and Conversion Variables (eVars)

[Gigazelle]

You could use a list prop that provides all enabled values, then another list prop that provides all disabled values. It doesn't sound like you need persistence, so you wouldn't burn one of the three super-valuable list vars available.

[jeff_bloomer]

That's quite an intriguing approach.  Not sure yet how to adapt that to the reporting need, so I'll have to think it through.

[Gigazelle]

Were you able to figure out the reporting obstacles behind any of these solutions?

[jeff_bloomer · Accepted solution]

Actually, I ended up using a list variable along with an event firing on launch, and a different event when the user updated their specific preferences.  Then, I also used a merchandised eVar to capture specific values related to the product when it is viewed, added to cart, etc.  This looks like I will be able to get the reporting I need based on that setup and then configure the classifications to break them out accordingly.  I really just had to think about it a little bit.  I really appreciate the advice.  It helped me frame my thoughts and goals.  Thanks!