-
So I’m an associate performance director at large network media agency (we’ve also been in the news due to sheer amount of recent people cuts, and AI is one of the reasons). I stay as luckily i’m on the AI side and actively try to implement it throughout our agency.
We’ve tested a few tools — pipe dream (honestly not good quality) and windsor (good, but only raw data). So far we’ve found the most useful SegmentStream. It brings all data in our own BigQuery instance, connects to GA4, Hubspot, etc. applies attribution logic, good skills library, etc.
Honestly so far it’s too good to be true, because it’s free (for now?). It’s kind of earlyresearch preview, so I hope it won’t turn into some big $$$ after next few months of extensive usage.
The alternative approach would be getting raw data via Make/n8n and other tooling that has offiical data connectors, but then you’d need to use AI to somehow clean and unify this data to prepare for normal AI usage.
