[tom_b]

Ugh, as a Linux BI hacker, I hate XLSX. I have to handle data imports from products that produce incorrect XLSX formats, apache POI blows up for the relatively small numbers of rows (<250K or so - I cheat by giving my JVM 16GB of memory to handled it and to be fair, I believe this is fixable but I have an ETL tool with POI embedded in a hacker-unfriendly way), but you are dead-on correct that you need some type of Excel output. Users love it. Particularly the kind of smart people who learn how to use the "Insert Function" wizard.

I actually rather like cubes, but find that simpler roll-up reports seem to satisfy most users. But I'm now in healthcare/research and remember in the finance world that cube slice-and-dice was more standard.

And I want to second the price range you mention - $20K to $30k per year is noise in the budget for groups that want an actual BI setup like you describe (not everybody needs hadoop). I have found that enterprises seem to have strong vendor preferences (e.g., they are OBIEE, Business Objects, etc shops and focus entirely on report devs with very narrow skillsets). So I am excited to see options pop up, but feel a bit of sympathy for a group that wants to crack the enterprise market nut.