We are currently using Amplitude for our Production, Staging and UAT environments. Our growth team are keen to get our marketing sites set up on Amplitude so we can track usage and journeys end to end.
I wasn’t sure whether we would get that desired outcome if we were to set up the marketing site/s in a new project? Is there a way with them segmented by project to still do some analysis across project to understand our users journey from the marketing site, through our app? Or is there a way to add it to the same project without damaging the integrity of it?
I hope that makes sense! Any guidance / thoughts would be incredibly helpful.
Best answer by Saish RedkarView original
From an implementation perspective, setting up your marketing site in a new project makes more sense and will be easier.
Just to confirm again, do you have access to the Amplitude's Portfolio add-on? The portfolio view is a paid add on available to customers on the Growth or Enterprise plan.
You can then tie the usage of a given user id across your production and marketing projects using this portfolio view. From what I know, except user lookup and event streams, you are able to use all the major charting features inside a cross project view.
As in the case of piping the new marketing data to existing production project, you are right in pointing out the risk of damaging the existing data integrity. Some major check list items involved are -
If you have the capacity and event volume to backfill your existing production data in case the marketing data ingestion goes rogue, then I would recommend testing out the integrations in small batches in new test projects.
Hope this helps.
Thanks Saish, this has been really helpful!
We don’t currently have access to the Portfolio add-on.
We are looking at piping it into our Production Project, just with some clear naming conventions, are there any other key steps you would recommend to make this a tenable option? Thanks!
Thanks for clarifying.
If some event names are common between the 2 data streams, I would introduce a source_system event property to all events if possible. This can help you understand and debug the source data in cases of 2 parallel event streams.