On Beyond Accounts Payable
Spend analysis works best when all the sources of spend are consolidated into one consistent view. BIQ can be used very successfully to integrate information from different accounting systems (see Practical Solutions for Multiple Accounting Systems). And, as we argue elsewhere (see Suite Silliness and Where's the Analysis?), integration of a spend analysis with the ERP system or with a sourcing suite is either ill-advised or illusory. Spend analysis needs to sit independently on top of all sources of spend data, mapping and otherwise altering them so that solid spend visibility is achieved.
But what about other data in the enterprise?
With BIQ, you can build as many datasets as you like (see One Spend Cube Is Never Enough). For example:
Invoice-level AnalysisSome leading theoreticians in the spend analysis space landed 10 years ago at the Mitchell Madison Group. They derived two different ways to analyze spend: (1) top-down, through the A/P system, and (2) bottom-up, through invoice-level
The invoice-level analysis approach was first employed during an MMG engagement with a large commercial bank. This bank had a burgeoning business in a foreign country, but costs were spiraling out of control. The analysis team entered several months' worth of invoice-level detail into a database, and then poked at it with desktop tools. What they found was opportunity: errors in invoicing, failure to comply with contract pricing, and so on. Armed with these data, the team went back to the bank's suppliers and extracted rebates that paid for the study many times over.
That was then; this is now. Invoices are in much better shape these days, and many suppliers can supply them electronically. And, instead of the primitive tools applied by the original analysis team, tools like BIQ can be brought to bear on the problem with far greater efficacy. In fact, individual BIQ cubes — by commodity — are often appropriate, since some commodities have their own interesting flavor of data, requiring different cuts at it. Commercial print data, for example, is an enormously rich area for analysis, since there are so many variables in a job (paper, ink, cuts, folds, pre- and post-processing, press type, and so on).
Purchasing cards can be a rich source of transaction detail. It's fascinating data, but analyzing it and classifying it is a chore — and one certainly can't afford to dedicate a 6-figure system to the task. Building a BIQ P-card dataset, though, is inexpensive and easy.
Insurance Claims Data
Another fruitful area for exploration is insurance claims data. BIQ has already analyzed large datasets from claims systems, and with its new 64-bit server capability, datasets of 100M+ transactions can be supported with reasonable response times.
BIQ customers are analyzing non-spend as well as spend data. These applications include:
- Call center data analysis. Most call center software systems are helpless to deal with the enormous volume of transactional data generated, other than through the production of static reports and summaries. And, call center system data are not integrated together with sales data, as is critical, for example, in a telemarketing shop. Such integration is where high value occurs — and integration of data from disparate sources is BIQ's strong point (see Practical Solutions for Multiple Accounting Systems). So BIQ is being used not only to analyze the raw call center data, but also to enrich it with sales information.
- TV ratings analysis. Several customers are restructuring Nielsen data using BIQ's analysis framework, with great results. Nielsen data are too large for spreadsheets, but too complex for relational analysis. Inside BIQ's Viewer, relationships become easy to spot, and analyses become easy to do.
- HR data analysis. BIQ is being used to correlate HR data between multiple plant locations to establish company-wide pay grades and define consistent bonus and benefits policies. BIQ's ability to populate analysis spreadsheets with HR data, and to combine those data with survey results and other input, results in high-value insights and actionable information.