can you match on part of a classification import field?

[mak4]

I currently have a custom list of 2000+ product detail pages that were removed from a site with multiple language variations. To answer questions about behaviors of visitors that had been to those pages (ie. changes in site search behavior or conversions), I want to create a visitor segment.

I wanted to use a classification on the page dimension, but that would require multiple variations to capture all the language variations. I don't believe that regex is supported. Is there a way with classification importer to have the key match on part of the string or would it have to include each variation? Would an alternative be to create a subclassification to eliminate the language variations or is there another option?

[MandyGeorge]

You can't use regex with the importer but you can use regex in the classification rule builder. It's pretty straightforward to make the rules and set it all up. Here's some documentation on it: 

 https://experienceleague.adobe.com/en/docs/analytics/components/classifications/classifications-rule...

If you don't want to use the rule builder, you can always use the importer, but you would have to have the full dimension value for that, you can't match just part of the string.

[bjoern__koth]

As @MandyGeorge mentions, regexes are supported and if you want to extract regex groups, just use the normal dollar syntax e.g., $1

[Harveer_SinghGi1]

Hi @mak4 ,

As @bjoern__koth and @MandyGeorge have mentioned you can use regex in Rule Based Classifications. The only issue with using Rule Based Classifications is that you can lookup only last 6 months of data. If there are some values that you need to classify and were never seen in last 6 months, the Rule Based Classifications will miss them.

So you can opt for a mixed approach for classifying all values,

1. Use regex in Rule Based Classification to classify all the unique keys received in last 6 months
2. Use Classification imported to classify any keys missed by the Rule Based Classification i.e. the keys that were never seen in the last 6 months but they were recorded sometime in the period before. For this you can get an export of these unclassified keys (which ideally should be a small list for pages dimension depending upon what kind of traffic you get on all pages) and use an offline method like regex with Python or some advanced excel to prepare the final classification file which can be input to Adobe.

Step 2 above is optional depending upon you want to classify the keys older than 6 months or not.

Cheers!

[Jennifer_Dungan · Accepted solution]

Just be careful with the mixed approach... I've seen people do this, but they had issues with timing... sometimes the importer ran first, other times the rule builder ran first....

If your rules depend on the importer data, then this may not be a safe approach... but if they are just both running on the same data, but don't actually rely on a value to be set to process the next value, you should be okay.

[mak4]

After looking further into the suggestions and your warning, we decided to sample data from the pages with the highest historical traffic to prove out the business case before building out a new rule or doing anything more complicated. 

Thank you @Jennifer_Dungan , @Harveer_SinghGi1 , @MandyGeorge . All the options with the potential pitfalls is exactly what I needed to make recommendations coming out of that analysis.