Spend Analysis Today

The state of spend analysis today, despite all the frenzied M&A activity between 2003 and 2005, is still a fractured and confusing one. There are about thirty (30) vendors selling solutions in the market-place, and only a handful are reselling or repackaging someone else’s solution. However, the big difference between today and a few years back is that there are only a couple of true stand-alone vendors left on the market, notably BIQ and Zycus, as most of the stand-alone vendors were swallowed up by the Big 6.

What makes it even more confusing is that, even though most of the vendors have their own solutions, it sounds like the vast majority are selling the same type of solution. Furthermore, based on all of the overlapping marketing, it appears that most of the vendors are trying to differentiate themselves based upon either the “intelligence” in their data classification algorithms, or the number of canned reports their application comes with – not on any obviously unique capabilities.

Although it is true that it’s impossible to do spend analysis without good data, and this requires your data to be as complete and as accurate as possible (at least 90%, but preferably as close to 99% as possible), which means that any good spend analysis solution needs good ETL (extract, transform, load) and automated classification capabilities, it’s also true that spend analysis is more than just classification and baseline reporting. Spend Analysis is about uncovering previously unknown savings opportunities. Such opportunities are not likely to be found with standard reports, since obvious opportunities are likely to have already been found and addressed by in-house analysts using basic SQL queries and simple reports. Thus, spend analysis must go beyond what a simple reporting engine can do or what your average analyst can do with SQL in order to be truly useful and find genuinely new savings opportunities.

In other words, creating an OLAP database on cleansed spend transactions will be a worthwhile effort the first time you do it, because you will be able to identify most of the obvious savings opportunities by way of variance and non-compliance. However, once you have addressed those “low hanging fruit” opportunities, there will be little residual value to the effort as it will simply report the success you have already achieved and fail to identify any new opportunities. In order to realize the true power of spend analysis, a user needs the ability to “play” with the OLAP database the same way she can currently “play” with the standard reports in Excel spreadsheets. It’s not about pivoting around the standard cube, but being able to create your own cube with your own data and your own dimensions and slice and dice those dimensions in any way you can dream up in your quest for that next savings opportunity.

When you find that opportunity, it’s about capturing the process used to derive the opportunity and re-applying it in an automated fashion to similar commodities and categories and to the same commodity and category again in the future to make sure that the identified improvements get implemented and stay implemented so that you realize the savings. This certainly requires that you have an instance of the application running a standard cube that is integrated with your contract management system and your procurement system to make sure you are continually buying on contract – but it also requires that you have the ability to build multiple cubes to address commodity-specific analyses and to address datasets that originate from sources other than the ERP system.

When you get right down to it, only two solutions on the market stand out – Zycus and BIQ. The two last independent players. Zycus stands out because, in addition to the advanced extraction, cleansing, aggregation, and enrichment capabilities that you will find in the other Big 4 players (Emptoris, Ketera, Procuri), it has built a first generation opportunity finder that goes beyond just pre-packaged standard reports to integrate variance analysis and market intelligence to automatically identify all of your “low hanging fruit” opportunities and included a pipeline-based workflow management process to attack and manage your initiatives.

BIQ (also available as part of Iasta Smart Analytics) stands out because it is the only product on the marketplace that truly gives the user the ability to “play” with the OLAP database. In BIQ, each user has the ability to define their own cube, either on any of the “standard” dimensions in the centralized data warehouse (which could be another spend analysis platform or an ERP system) or on any dimension they want to define using BIQ’s capability to define new dimensions in near real time. Plus, the user can re-order the dimensions of the cube for reporting purposes at any time, dynamically create their own reports, and even analyze multiple dimensions simultaneously using treemaps (based on Shneiderman diagrams) and multidimensional extract capabilities. Finally, there’s BIQ’s unique ability to allow users to define and re-define the classification rules dynamically using a very powerful rules engine, and their forthcoming meta-rollup capability (programmatic rollup of rolled-up data).

In short, spend analysis is about the analysis, and, currently, with the exception of BIQ, that’s a point that the current (leading) vendors are failing to grasp. Like Aberdeen, they’re Lost in the Trees. Now it’s true that BIQ does not come with built-in facilities that will automatically classify 95%-plus of your spend, but spend classification is fundamentally not that hard to do. The secret sauce to do that has been known by your leading consulting firms for years: map the vendors, map the GL codes, map the vendor and GL code combinations and then create exception based rules for whatever is left or whatever doesn’t map properly. This is something that can be done by a tactical procurement agent or accounting clerk in a short time frame in even the largest of Fortune 100 companies – at a very reasonable cost. (So why are you paying hundreds of thousands of dollars for technology to do it for you?)

There’s no excuse not to look at any tool that gives you better analysis capabilities when even a basic ETL tool can do what you need to do with a little bit of elbow grease up front. Especially since I’ve heard good arguments that automated classification does not really exist. After all, what are automated classifiers doing? They’re applying rules. Where did those rules come from? A human being. The only real difference between the big solutions with enhanced classification capabilities and the little solutions with basic classification capabilities is that the big solutions have rules that have been defined, or in the case of automatically derived rules, checked by experts based on years of doing manual spend analysis projects for their clients. (Furthermore, I haven’t seen a classifier yet that has not required heavy human intervention on the back end to correct mistakes – especially during the initial implementation. And, despite what the sales people would have you believe, this often takes just as much effort, if not more, than simply having a knowledgeable human define the rules in the first place.) The underlying technology, fundamentally speaking, is not that different. It’s true that some of the algorithms employed by the big players are a lot more advanced, but they are still based on rules and knowledge derived originally from a human. As I’ve said before, computers are not intelligent. They are just very good at doing the calculations they’ve been programmed to do.

Furthermore, even if you have a spend analysis tool already, there’s nothing stopping you from employing a dual-tool approach – a standard Big 4 (or Big 6 when you throw Ariba and CGI into the mix) solution to automatically extract, cleanse, classify, amalgamate, and track all of your spend data from your various data sources in a standard cube setup and then a BIQ (-like) solution that can be delivered on-demand to the members of your strategic sourcing team to help them find the next great savings opportunity. The standard solution will be able to automatically create all of the reports that finance and the executive team want to see while the BIQ (-like) solution will give the power users on your strategic sourcing team the tool they need to uncover the next level of savings opportunities. Plus, a BIQ (-like) solution is on-demand and relatively inexpensive (for example, BIQ only runs your average organization between $3K and $6K a month for the sourcing team), especially compared to the realizable savings it will identify. It’s certainly something that should be considered.

Up Next: So You Want To Do Spend Analysis (7 Starting Steps)

Note that this isn’t to say that suite-vendors like Ariba, Emptoris, Ketera, and Procuri (etc.) don’t have valuable solutions – they do (and I’ve even written about some of them on this blog in the past), just that, on their own, their spend analysis solutions don’t truly achieve the analysis necessary for your sourcing team to go beyond the low hanging fruit – which is the key to achieving year-over-year savings (even if these systems do work great the first year). From a finance perspective, they are pretty good – centralized cube, standard reports, automated feeds, automated classification, etc. etc. – they just don’t have everything the power hitters on your sourcing team need today!

I have an opinion. How ’bout you?