Segments value limits when using contains any of

[kostjohn87]

Hi,

Im trying to isolate about 10k values in a segment using the contains any of filter/logic. Essentially if any of these 10k values are in the hit, visit or visitor only show me corresponding behavior for this segment. I uploaded using the segments api and everything saves and works fine there. But when I try to pull the segment in on workspace Im getting the dreaded contact support alert. In Report builder it just keeps queuing and never processed(at least in the 2 hours I had it running) and Data Warehouse it processed 1 day in 2 hours(request had 30 day range in it, doesn't seem right it would take 60 hours for this, projecting out based off the 1 day processing time). So that led me to inquire about the best practices or limits that might be involved with the segments.  So is there a limit on how many you can isolate? Is this a

performance issue. Any help would be greatly appreciated.

[settytb1 · Accepted solution]

Hi kostjohn87,

A segment is not meant to be a limitless list of conditions.  For example in your case the best use case would be to setup classifications on that variable.  Setup a column of like Auth GUID type or whatever.  Then just do a vlookup in excel with your list of ID's.  If the guid meets your list set the type to be X.  Then build a segment based on the Auth Guid type = X.  This makes the segment much simpler, and frankly easier to manage with classifications bulk export/import capabilities. 

I hope that helps,

Seth

[Gigazelle]

100% agreed with Seth here. The best practice would be to create a classification based on the variable storing the ID's you're segmenting on. Everything you want to include in the segment gets the same classification value. Using the classification uploader should be pretty quick if you have the list of ID's already.

Once you've uploaded the classification file and it's processed, you create a segment defined as 'visitors with the ID classification value of X'. Just that one segment definition would be all that's needed, and won't run into performance issues in any way.

Segments start choking once you hit around 50-100 rules, so if you have a large amount of different values, using a classification is the best way to handle it.