Use generic success events and create calculated metrics for specific conversions using an eVar? (e.g. 1 eVar tied to 1 form fill completion event and utilize calculated metrics to get total events by form version)
Create a unique success event per set variable? (e.g. 1 eVar tied to multiple form completion events, one for each version of the form)
A mix of both?
What are the advantages and disadvantages you find in the method you use?
I know for example we have to create a unique calculated metric for each attribution model we'd like to use, rather than being able to set that easily in the workspace table or use the Attribution IQ tool. Also, calculated metrics are not available in Data Warehouse so we need to use segmentation and do extra cleanup on the file. But, unique success events are more work on the implementation side. Any others?
I don't use Data Warehouse much, so my answer won't be 100% applicable to you.
I prefer "Use generic success events and create calculated metrics for specific conversions using an eVar".
In my experience, an AA user is more likely to want to know the total count of a success event rather than a unique count.
If a unique count is needed, a Calculated Metric can solve for that.
I don't particular use option 2. It's not trivial to calculate a total count from a unique count, if the total count is ever needed. Heck, I don't even think that's possible! So I'd reserve this only for those specific cases where a total count is illogical, e.g. the classic case of counting purchases where a user could reload the Thank You page.
For option 3, I've found that duplicate success events (e.g. one non-unique, one unique) can lead to more confusion, especially when new AA users get started.