This continues a series of guest columns from practitioners and bloggers I respect. The category - The Real Deal describes them well.
Quick - name the person who has been involved with more HR software startups in the last 3 decades than any other. If you guessed Dave Duffield, you are wrong. It is Naomi Bloom, who has helped more vendors with designing HR process/data models and a few corporate clients as well. She is the Managing Partner of Bloom and Wallace and her contributions to HR technology have been recognized by IHRIM and numerous industry publications.
Here she challenges us to not just look at contemporary SOA implementations as technology plays. Great insight - not just for the HR domain.
"In all of the discussion about Web services and their enablement and orchestration via the highly touted SOA middleware, there has been far too little attention paid to the importance of basing those Web services on fresh, correct and sufficiently granular object models of the subject domains. Many of the so-called IT analysts have failed to make the tight connection between getting the models right and getting the expected business benefits from SOA/Web services. Even with the clean slate that Workday has, if they aren't able to get their enterprise and, initially, human resource management (HRM) models right, it won't matter that they may have great enabling technology. But getting those models right will not only require that the business analysts doing the modeling work have deep and repeated experience developing object models for the specific domains but also that the vendors are willing to create significant data and process discontinuities with their installed base and with conventional ways of thinking about these domains, discontinuities that the vendors are loath to discuss.
In HRM, about which domain I have the deepest knowledge, it's certainly time to make a fresh start. The design of most established HRM software never envisioned the integrated and durable use of contingent workers, the extent to which all HRM processes are linked strategically via competencies, the fact that a single employee might be working part-time in each of two quite separate positions within the same enterprise and need to be treated separately for such purposes as performance and scheduling but treated as a single person for purposes of taxation, or the extent to which organizations would have multiple and concurrent views of themselves for such different purposes as servicing clients and developing new products. All of these considerations and many more lead to quite different object models of the HRM domain than the ones upon which much current HRM software is built (or, in the case of some, than the absence of proper models upon which the software was built). Unless the shiny new Web services and SOA versions of older software and of all new software are based upon a very contemporary, complete, global and correct understanding of the HRM domain, we will have the same muddle under the covers that adds cost to today's implementations, operations, and service delivery.
But new underlying models will likely present an historical data conversion challenge, but that's a tale for another day. Suffice it to say that when the underlying data design changes anywhere in HRM software, there is at best a highly automated but subject to manual review data conversion process and at worst a ton of brute force remapping and review. So when Oracle says that their next generation Fusion applications will be based on EBS, they are telling you that PeopleSoft customers are going to have a substantial data conversion challenge in addition to whatever else it takes to get to Fusion. And when SAP tells you that their SOA by Design new products for the middle market are similarly a fresh start, they are telling you the same thing. But if we're going to pay the price in discontinuities, let's at least be confident that the shiny new Web services are based on domain models that will be durably correct once we get through the pain of these discontinuities and new implementations.
Are we ready to step up to the real heavy lifting, the fresh modeling of our domains, that's needed to make a success of Web services and SOA? Are we prepared to undertake new implementations and discontinuities with historical data to get the real benefits of these fresh models? Our major software vendors clearly think that we aren't ready or they wouldn't be so reluctant to discuss these challenges. Perhaps these challenges will be impetus for BPO that Y2K was to in-house software refreshes. But however we get there, we aren't going to have the software we need to support modern organizations unless we're willing to make a painful and difficult break with the past. And we haven't even begun to explain these points to our managements."
Naomi can be emailed at [email protected]
