We also consolidated from multiple systems but we never had that many in parallel, we always were finished in time before adding/splitting the next set.
The previous setup was all SAP, but that was too slow and we couldn't get the change rate we wanted. That was over 10 years ago, and AX had different issues but at least nothing the end-users have to deal with.
Besides AX and 365 we have 3 more, as well as some legacy processes still using MSBI, but all of the data going in and data going out is captured (either in Databricks or in our integration systems) which means that reporting and things like invoicing, banking, purchasing etc doesn't really care much about the underlying ERP. The ERP core team of course does, but the days where most of the work was serving direct business day-to-day needs are long gone, it's mostly edge cases and keeping the ERP in line with business changes.
All the metrics are processed in spark and exposed to both Tableau (together with an SQL Warehouse that lets you access all the individual data as well), and Thanos. So if a technical team wants to do some deeper analysis side-by-side with technical data that's possible too (which is great since they tend to be more comfortable with that vs. tableau). It's pretty nice to have financial data, WSSI data, GTM data and service deployment data overlaid so you can see the direct impact of everything on everything. (even if the resolution for some data is not the same - zoom out enough and you get very actionable insights)
Perhaps one of the differences is what we are mainly in retail, warehousing and logistics, when we buy a product or services business we tend to ditch everything except the data and the people. Our data engineering team takes care of the capture and transformation and the respective specialised process teams (CRM, ERP, PIM, fulfilment etc) coordinate from the business, legal and operational perspective. There will be a cutoff date and depending on the size we'll either have had a realtime streaming period (CDC style mainly) or do a one-off snapshot of the business and do it in one go.
For our business users it means they won't need to go to a secondary system to get the 'other' data; instead, the data comes to them.
I'm not sure why this method has not failed yet, I suspect it's because some M&A strategy components outline the cost/benefit clear enough to not have us integrate something that is not integrate-able.