A little too much effort was put into making it pretty, and not enough effort into making it practical for analysts.
It feels like it tried to be the best of both worlds, but it is below average with either…. average dashboarding abilities in tool, and less than average ability to do analysis.
Getting rid of Ad Hoc, and rather than building off-of and enhancing the platform most clients use for business analysis, is a mistake.
It feels like it is built with touchscreens in mind... Analysts don't work like that. Real Data Scientists would laugh at you outright. We need speed, keyboard shortcuts, and to be able to construct data sets quickly, rather than pretty drag-n-drops.
It's slow, in that everything takes a lot of clicks to get through. You can't even effectively copy-paste (it works sporadically). I can make a trended chart in half the time in Excel, with less work on my part.
Often times, I need to do my analysis in other tools, such as excel, R, Python, etc. 200 rows doesn't cut it.
Now you may be thinking - use Data Warehouse for your big pulls... ahhhh but there's the rub...you lose 100% of your agility... Need it by day? One pull. Wait 20min. Need it by day by device to chase a hunch? Second pull, wait 20min. Warehouse only works if you’re confident you know all of your inputs in advance, and then you can’t put segments on individual columns, so if I need a data set like that, I’m out of luck.
It is better at doing dashboards in-tool than previous tools, but lacks polish like Tableau or PowerBI
If you do more than one breakdown, workspaces is virtually unreadable, and unusable for the uninitiated, because you can't quite make out what is broken down by what, (because light grey on light grey is a fantastic color scheme for an iphone, but I have to do actual work, not just look good).
When breaking down, it puts too much effort on the user. If I need to see Day>Device>Department, I have to go in and manually break down each Device by Department, and then go back, click the “5” and turn it to “25” on every single totals row (which as stated above, looks exactly like a non-total row).
Ad Hoc is pretty much the only tool to be able to do medium size pulls effectively, be able to visualize it, and be able to construct data sets, and you are decommissioning it - and it's pretty much the go-to for analysts beyond novice level - Not a good move in my opinion.