We gained a solid undertsanding in discussing with some of the retail leaders that BI is widely used across departments and business functions. Thus retailers have adopted a strategic approach on driving fact-based decisions and business performance intelligence throughout the organization.
However, when looking at the typical BI application landscape in retail, that is still caracterized by application and data fragmentation. Enterprise systems and BI applications fragmentation is a major cause of data fragmentation – as matter of fact data integration vendors remained on the rise also during 2009, due to continuing requirements on consolidating internal datasets/application flows and to allow for more responsive supplier-retailer collaboration scenario.
Three interesting perspectives emerge from our discussions with the retailers:
• Resiliance to changing such a fragmented application landscape is also the effect of consolidated work practices that predominate in many instances. Think at Excel for example - spreadsheets provide the ultimate in flexible data manipulation and formatting. While this flexibility may be appealing to some individual users, it presents several challenges around data governance, process continuity and user access rights management.
• Talking about dashboard and reporting, most often retailers want to define their own set of KPIs, thus not much leveraging predefined KPIs that are included by several BI vendors in their solutions. Still, sales margin KPIs are among the most popular, including sell-through, sales and top categories at single store level. What is reported by the retailers as the more effort-intensive task is the design of KPIs – not their actual implementation.
• Albeit proof points exist for measuring the ROI of BI tools, most of the retailers we have been speaking to agree on the fact that it is pretty difficult to measure the benefit of BI, and quantify its payback and financial return over time.
Generally, gaining success with BI projects is an iterative process, as organizations learn how to make better use of the technology and also learn how their organization will benefit from a fact-based approach to decision-making. Therefore, the IT perspective is that BI benefit can be locked by the organization ability to react to their needs.
On the other hand, TCO reductions remain an important objective. In fact, we are seeing some retailer investigating into license, maintenance and development costs, and now searching for lower cost alternatives that may work well in their landscape.
The BI focus in retail is now shifting on delivering ad-hoc flexibility to users and performance – the latter not only from an analytic or reporting tool standpoint, but considering the end-to-end BI process, including datawarehouse and disparate databases, ETL and integration tools, statistical engines and data modeling tools.
The two objectives – ad-hoc flexibility and performance - are going into opposite directions - the higher the number of ad-hoc users, the bigger the challenge of predicting query loads to the platform from the IT perspective. Instantly, cloud computing comes to mind, given the ability to allocate computing capacity on-demand at a fraction of on-premise implementations' costs.
Certainly, providing an ad hoc query and analysis environment with the flexibility, data visualization needs and simplicity demanded by end users remains a challenge. What proves valuable to increase user acceptance and satisfaction in working with new BI tools is to train on the data in the user's application at the same time as training on the tool.
Is your organization requiring a more flexible BI solution (e.g. allowing more users to make queries easily, with minimal training, or even using natural language), or are you looking to consolidate to a few super users and make reports available to the organization true that single point of control? Where do you draw the line?
Whatever your approach, our syndicated research and advisory services can help guide you in the process.
Ivano Ortis, IDC Retail Insights









Tweet me!