Spendata uses classic AI mechanisms to automatically family and map spending. We use a combination of heuristics trained on spend data along with knowledge bases derived from existing spend cubes and other sources. This approach is analogous to the methods used by chess-playing programs on bespoke devices such as game consoles. Such programs have memorized the standard openings of the game — their knowledge base — supplementing that knowledge with heuristics to evaluate potential moves.
Spend analysis vendors don't want to talk about it, but automatic mapping always requires manual corrections and enhancements. There are lots of reasons why, but the simplest reason is that the way you use a vendor is not necessarily the way that another company uses them. For example, should Shell Oil be mapped to:
Another reason is that automatic mapping makes lots of mistakes. "95% accurate" sounds terrific until you realize that it means 1 in 20 mappings are wrong.
That's why you'll find full-service spend analysis vendors always quoting the same "4 to 6 weeks" to produce a spend cube. If automatic mapping were the panacea some try to pretend it is, then they'd be able to produce your cube in minutes, and your cost should be minimal. Neither is the case. They have to roll up their sleeves and fix all the errors, and then map all the items that the automaton didn't. Ironically, most of the advantage that the automation creates accrues to the vendor, since it reduces the effort required to build a cube and therefore inflates margins.
Spendata gives you full access to both our automated tools and our manual mapping tools. Both are easy to use. After automation is finished, you have many options, including refining the mapping yourself, or contracting with Spendata Services or a third party to do so. Or, you can map just those portions of the cube that appear to offer sourcing opportunities, and skip the rest. What is the point, after all, of mapping vendors with tiny spend that you don't intend to source?
The problem is, almost all spend analysis vendors perform their mapping offline, with tools and procedures often having nothing to do with the spend analysis application itself. Usually only the vendor can perform them, or they’re too confusing to be mastered. Either way, the end result is the same: a permanent, expensive dependence on the vendor to map new transactions or to correct errors in old mappings.
With Spendata, you can remove this dependence, and then you can use Spendata’s segmentation power to take the system far beyond just a display of spend transactions.
Whether you choose to perform your own segmentation with Spendata, or hire a services provider to do it for you, you’ll always have the tools at your disposal to change mappings, to add new columns and mappings, or to build entirely new datasets. You’ll never be dependent on a third party to maintain your spend cubes, and you’ll always have the capability to make instant changes as required. You can shed your third-party services provider at any time. But regardless of whether you solicit help or not, there’s no more waiting for the next “refresh” in order to see your changes; just make the change and you’re done.