Hello!
TL;DR:
I'm wondering if someone out there has knowledge or advice for how to see which attributes/events within datasets are going unused by using the UI or query service (API is not a great option for us). It's understood that there is likely not a one-click solution to this.
Full version:
I'm a new member of our organization's product team for Audience Manager, and I have been tasked with something that more experienced team members haven't yet encountered. We need to reduce the size of our profile store to avoid going over our contract (and also just for good housekeeping). This reduction in size will likely come from a combination of reducing TTL for events, removing unused events & attributes, and removing entire datasets if they aren't being used.
We feel confident that duplicate data is not the culprit, but rather unused data. In our earliest days of using AM, the goal was getting as much data ingested as possible to get our users off of legacy platforms. We're now more mindful of avoiding mass ingestions in favor of more thoughtful ingestions. But there exists a medium-to-high likelihood that we're ingested data that has gone unused.
My original plan was finding a way to view/export a list of published audiences and the respective attributes/events they contain. And then bump that list up against *all* of our attributes & events to see what is going unused. But I'm not seeing any tables in query service that would provide this. ChatGPT (which I realize isn't always accurate) says that audience logic is not queryable in query service,and that the only options are manually going into each audience in the UI and copy/paste the included attributes/events to get a list of what data is actually being used in audience builds. The only other option I've found is using an API combined with a scripting language to pull what I'm asking for. Working with API's, scripting, etc., is beyond the scope of our team's skillset. We have an internal engineering team that might be able to do this, but they currently have little-to-no available capacity.
So...I'm wondering if anyone else out there has worked on something similar, and found a solution via the UI or query service (the assumption being that it won't be a one-click solution).
TIA!
