It’s the 95% perspiration necessary for the 5% insight [NY Times]. With spend analysis, there are two ways to do the sweaty part of the job:

1. get a spend analysis tool that can easily map and segment your data, or
2. spend your nights and weekends hacking on spreadsheets.

Isn’t (a) standard fare with any spend analysis system? Regrettably, no.

You’ll find that vendors in the spend analysis space fall into two buckets: (1) purveyors of analysis tools, whether sophisticated or not, that allow users to load, map, and segment their own data; and (2) service companies that perform data manipulation and mapping offline, then load the resulting dataset into a hosted BI tool with pre-formatted dashboards and reports.

The class 2 product isn’t really a product at all — it’s a mapping service and a BI license. Vendors in this class can only be compared on the quality of their mapping (difficult to measure) and the insightfulness of their reports and pre-formatted dashboards (equally difficult to measure, but likely inadequate for serious investigation). Any custom mapping or segmentation has to be done offline, in a spreadsheet.

Let There Be Cubes

To gain a handle on what’s really going on, an analyst will want to create many cubes (mini databases, if you will) — one for each commodity, one for Pcard spend, AP spending, IT spending, one for each subsidiary (or portfolio company), one for all entities or subset of entities, and so on. She’ll want to manipulate each cube, mapping it, creating new dimensions and views, adding new data and measures. You get the idea. Cubes will come and go, but there will be many of them, and they’ll be changing all the time.

Real freedom to manipulate data — the key to gaining insight — stops at the pre-formatted dashboards in a class 2 solution. For some this may be all that’s required. But the serious analyst will want to do much more.