Best Practice for Event Implementation



Hoping to get someone from #AdobeStaff to chime in on this as well.

To those of you who have been using Adobe Analytics for a LONG time, I was curious to get your feedback on best practice for using more generic events for multiple interactions on a site.  For instance, you can view the details of a number of items on a site, like products, articles, recipes, coupons, etc.  Of course, we have the prodView event, but we also fire an event for viewing products in order to use those events against other objects not having access to product views.  Recently, we've been talking about having a separate event for each detail view we execute: view coupons, view articles, view recipes.  Would you recommend using a separate event for viewing the details of different things, or would you maintain a single event for each type of item you can view?




What if you create and evar like "Site View types" then have a state for each type of article view

view coupons

view articles

view recipes

The evar would then allow you to segment users by view types or allow you to create view chains to see people who view X then Y...

Rank articles or pages by view types, etc...

this saves using an events and gives you flexibility to add more types in future.




9 times out of 10 I would probably set up a different event for each of these view events. And then you can always have a calculated metric that sums them all up if needed.




OK.  Curious about the justification on that.  I'm trying to play devil's advocate around this issue and understanding the "why" as opposed to just what people do will be extremely helpful.





Dear Jeff,

If you use eVar, we do have few benefits.

It is a variable so we can set allocation to understand the impact of other actions. Example : If somebody viewed 'Articles' and then submitted Lead, you just need to select the metric against the eVar to understand the impact. So 100 Prod Views for 'Articles' and out of it 10 have submitted lead. Till eVar is replaced, it is persisted to capture all the events based on expiration.

Even in terms of reporting, it is easy since the line item is going to be unique. Moreover, we can sub relate them with other variables.

If you use events, yes, we can use calculated metric to understand the overall numbers. But when it comes to understand the impact, you need to use Segments. Example : You need to create a Sequential Segment to understand the visitors who viewed the product thereby adding it to the cart. Reporting will be little difficult.

I will recommend you to go on with the suggestion give by





I'm with Nick on this one. Many Analytics contracts come with up to 1000 events. If you've got them, don't be afraid to be liberal in using them. If you need aggregate numbers, you can use calculated metrics. You could also use a separate event for an aggregate metric if you want to guarantee deduplication on hits.