By David Dorf on May 25, 2011
Today Oracle Retail announced availability of a new product called Oracle Retail Merchandising Analytics, the first of several BI applications planned for the retail industry. To further describe the product, I've asked Mark Lawrence, the brains behind ORMA, to explain the strategy and why this approach is different than what came before.
It's probably safe to say that those reading this blog are all too aware of retail's "data rich but information poor" reputation, and that today's competitive pressures are forcing the industry to compete on analytics. You can't improve on something if you don't measure it and monitor it, right?
After spending many years building a homegrown Enterprise Data Warehouse (EDW) at Circuit City (eh-hem, great BI was unfortunately not enough to save the company), I was hired by Oracle to lead the creation of a next-generation BI solution for retail. One that would leverage the full Oracle BI/DW technology stack, storage-to-scorecard, yet not necessarily require that full stack. One that would be optimized for Oracle's retail apps, but designed to integrate with non-Oracle data sources as well. One that would not only address retail enterprise needs, but those of the full corporate enterprise. One that was modularized so that it could serve as a retailer's EDW or that could augment an existing EDW with one or more specialized data marts, perhaps enabling a next-gen EDW via incremental data mart implementations. One that could surface BI, properly-filtered, to the right people, at the right time, using the right delivery method whether it be mobile, dashboards, or objects embedded in a planning or operational app. One that I would have wanted to employ at Circuit City, had it been available then (reminds me of my former dream of the "BI guy" saving the company and retiring early on stock options...).
So, Oracle Retail Analytics, with the first of five planned modules just launched last month, embodies all of those things. That first module, Oracle Retail Merchandising Analytics (ORMA,) is now Generally Available, is built on Oracle database 11gR2 and includes packaged integration using Oracle Data Integrator 11g with Oracle's merchandising product family, expansive Oracle BI 11g metadata and reporting, and a data model that is based on Oracle BI Applications 11g to enable cross-domain, retail + ERP/CRM analytics.
Each module is "plug-and-play" in that it includes packaged integration with the associated Oracle Retail applications, fully physicalized data model, and Oracle BI metadata and reporting. What I really like about the strategy is the ability to choose among 5+ retail BI modules and 25+ ERP/CRM BI modules to meet the unique needs of your particular retail enterprise, yet deploy that selection on a consistent and cohesive framework, and do so incrementally if desired.
Want to combine, say, Merchandising with Customer, Loyalty, Finance and HR to turn data from Retail, Siebel, EBS and Peoplesoft into information to drive business decisions? Want to, say, compare labor costs (HR) with sales per employee (merchandising)? Perhaps you have these Oracle apps and want to include supply chain BI coverage but don't own Oracle's supply chain apps? Oracle Retail Analytics is designed to also accept data from non-Oracle sources yet preserve the majority of packaged ETL transformations (ETL tends to consume 60-80% of the effort that goes into developing a BI/DW solution, and we want to pass as much of that value along as we can regardless of data source).
What also really excites me are the possibilities when
running Oracle Retail Analytics on Exadata. While we've baked-in plenty of features to
enable optimization of both loads and queries on Exadata, we've been careful to
ensure great performance and scalability regardless of chosen platform (Exadata is optional). Since we've had the good fortune of being
able to design from the ground-up using the very best and latest Oracle tech,
at times we've felt like kids in a candy store. Designing "from the ground-up" has also enabled some features that
otherwise would be difficult to design in a performant manner, like
"as-is/as-was" reporting for the product and organization dimensions
- allowing users to account for changes to these dimensions when assessing
historical performance. So, as items are
reclassified, or stores open, close, or move to new regions, reporting is done
based on the dimensions as they were, and/or as they are.
Using Oracle BI 11g, Oracle Retail Analytics enables more than just viewing reports. It enables deep analysis including data mining (detailed, transaction-level data is retained) and in-context and embedded actions - so we have the ability to initiate an action right from a dashboard or report. These actions can include things like triggering a workflow to order more stock, or kicking off a promotion based on events or metric thresholds being crossed. Or, they can be simple things like notifying people of key information, guiding someone to do further analysis. We call this 'Closed Loop Analytics' - because it enables closing the loop between insight and action, and Oracle Retail Analytics is designed with this capability in mind.
If you're at Crosstalk in June, attend my session to learn more. --Mark