Friday, January 16, 2009

Nine Choices on the Road to BI Solution Centers

Nine Choices on the Road to BI Solution Centers

December 2008

Forrester Research says BI Solution Centers offer a business-governed, solutions-focused edge over BI competency centers. Here are nine key considerations you'll face when building your BISC.

By Boris Evelson and James Kobielus, Forrester Research

BI is no longer just about back-office reporting. As BI solutions increasingly permeate the enterprise and span a wide range of applications, analytics-driven organizations recognize BI as a key corporate asset and a do-or-die platform. In today's turbulent and increasingly commoditized economy, enterprises must make better and faster decisions to stay competitive — and often just to keep their heads above water.

As BI grows more pervasive, complex, feature-rich, and mission-critical, it also becomes harder to implement effectively. Many information and knowledge management professionals question whether they architect, implement, and manage their BI initiatives properly. Doing so requires sound BI and performance management best practices — and an awareness of the myriad ways it can all go wrong.

Forrester's ongoing research compiles a litany of worst practices often committed, deliberately or inadvertently, by even the smartest, most experienced information and knowledge management professionals. Common deficiencies in many enterprise BI environments often manifest themselves at the application level, but the root causes of the problems go much deeper. The chief symptoms of suboptimal BI management practices include:

The lack of a single trustworthy view of all relevant information. Many organizations strive for a single unified view of disparate transactional data and commit themselves to the long-range goal of consolidating it all into an all-encompassing enterprise data warehouse (EDW). In practice, though, the goal of an uber-EDW is a moving target. EDW projects are frequently the victims of "scope creep," due to constantly changing requirements, relentless growth in the range of operational-data sources, and stubborn resource bottlenecks within IT. Insufficient focus on data quality and master data management (MDM) only adds to a lack of trust. Even data in a comprehensive EDW may be viewed as untrustworthy or, in a worst case scenario, incorrect. As a result, BI application users resort to old fashioned methods to collect and analyze data such as running their own SQL queries and bringing data into spreadsheets for analysis.

BI applications too complex and confusing to use effectively. Crafting sophisticated BI applications for power users is important, but designing them for casual business users is far trickier. Even the most user-friendly, point-and-click BI applications often require users to slog through a daunting range of user interfaces, features, reports, metrics, dimensions, and hierarchies. Also, BI is just a subset of the surfeit of productivity tools that information workers must juggle just to perform their basic responsibilities. As a result, most BI end users have barely tapped the productivity potential of the tools at their disposal and often run back to IT to help them create new reports, queries, and dashboards.

BI applications too rigid to address even minor changes. Our modern world moves at lightning speed, but BI solutions are often too rigid to keep up with the changes. One simple change to a single source data element can result in a few changes to extract, transform, load (ETL) and data cleansing jobs, which may turn into several data model changes in operational data store (ODS), data warehouse (DW), and data marts; this in turn affects dozens of metrics and measures that could be referenced in hundreds of queries, reports, and dashboards.

As these problems illustrate, the typical BI environment is far from realizing its potential as a strategic business asset. Many organizations have responded by developing BI support centers or BI competency centers, but a BI Solution Center (BISC) offers a more business- and solutions-focused advance on these concepts. This article, which is based on the Forrester Report "Implementing Your Business Intelligence Solutions Center," details nine choices you'll need to make on the road to building an effective BISC.

BI Solutions Centers Cultivate BI Best Practices
Implementing BI technology is easy (relatively) — but getting value from those technology investments is the truly hard part. Recognizing that sound BI management practices are often the missing ingredient, many companies have begun to transform their project-based BI support groups into a more strategic function: the BI solutions center. Though the BISC has great promise, it is no silver bullet. Enterprises with more successful BI implementations often implement some form of BISC practices, but there are a wide range of BISC implementation options, and not all of them are appropriate for every scenario. Why? Because there are many different approaches, organizational structures, and modus operandi for BISCs, each with its own pros and cons.

Forrester defines a BISC as:
A permanent, cross-functional organizational structure responsible for governance and processes necessary to deliver or facilitate delivery of successful BI solutions, as well as being an institutional steward, protector, and forum for BI best practices.
How does a BISC differ from such kindred concepts as the BI competency center (BICC) and BI center of excellence (BI COE)? Though it has the same core IT-centric functions (such as building OLAP cubes, deploying data warehouses, and writing ETL scripts) as a BICC or BI COE, the BISC differs in its business-led governance and solutions focus, as explained below.

The BISC, like a sharp business suit, must be cut, trimmed, and tailored to the contours of each organization. Each BISC must, at the very least, fit an organization's specific structure, people, business processes, technology, and especially the BI, data warehousing, and other analytics-relevant infrastructures.

The intersections of these four dimensions — process, people, data, and technology — create multiple BISC scenarios and approaches that information and knowledge management pros must consider when developing a BISC most relevant to support your BI efforts. Detailed below are nine scenarios and approaches you must consider when implementing your BISC.

Consideration 1: Strategic Or Operational Objectives?
Some organizations deploy BISCs that are purely strategic or advisory in nature. In those organizations BISC accepts the role of being a BI champion, providing subject matter experts, and overseeing BI standards, methodologies, and a repository of best practices. When these BISCs take on more operational duties they become responsible for tasks like the BI project management office (PMO), training, and vendor management. And in an ultimate operational manifestation of BISC, it can also carry the full spectrum of delivering BI solutions — BI solutions-as-a-service.

Consideration 2: In-house or Outsourced?
Enterprises deploying BI will need help from experienced consultants and systems integrators (SIs). This expertise is critical because BI is very much an art and will remain that for the foreseeable future, since it involves engineering a complex set of systems and data to address the changing imperatives of business organizations.

All successful, complete, scalable, "industrial strength" BI solutions require customization, application of best practices, and a significant systems integration effort. Because true best practices do not evolve from implementing two or three BI applications, internal resources with experience in dozens of successful BI implementations are difficult to find. A knowledge of best practices and lessons learned needs to be accumulated across hundreds of BI implementations — a privilege reserved for full-time systems integrators specializing in BI. As a result, most of the more successful BISC organizations include both internal and external staff.

Consideration 3: Virtual Or Physical?
Organizations have a choice of leaving their BISC staff within their lines of business (LOBs) or functional departments, or moving them to a centralized physical BISC organization. Since members within virtual BISC organizational structure have other management or hands-on responsibilities, they may lack BI focus and have to juggle conflicting priorities. Therefore, this type of a structure is typically more appropriate to BISCs that are strategic and advisory in nature. On the other hand, a physical, dedicated, and centralized organization is often more appropriate to fully operational BISCs. However, these tend to become just another "cost center" — as any centralized function carries with it the burden of process, methodology, and organizational structure. This implies bureaucracy, red tape, and a lack of flexibility. While such a structure is a must for certain IT functions, like infrastructure, security, and many others, it could be a BI showstopper. The first time IT cannot respond quickly or efficiently enough to a new requirement, a typical BI user will run back to spreadsheets to build a homegrown model, run the analysis, and get the job done. Information and knowledge management pros must determine which BISC structure — virtual or physical — would be most effective within their organizational culture.

Consideration 4: Operational or Analytical in Scope?
A BISC for some may focus on addressing the front-end access, presentation, delivery, and visualization requirements of analytic applications. Alternately, others may encompass a wider scope including data warehousing; data integration; data quality; master data management (MDM); and many other analytics-relevant infrastructures, processes, and tools.

Information and knowledge management pros can draw the scope of your BISC narrowly or broadly, and that line may depend greatly on how your company staffs, funds, and organizes these diverse IT groups. How far should a BISC go "upstream" to operational applications to draw that line? For example, is a database trigger implemented in an operational application for changed data capture (CDC) that feeds a DW part of the analytical or operational realm of your BISC's responsibilities? Is the data mart that calculates customer or product profitability and feeds these numbers to downstream operational applications an analytical or an operational data store? Such scope needs to be very well defined and managed to avoid the very real BISC "scope creep."

Consideration 5: Support IT only or All Stakeholders?
In especially large, heterogeneous, and siloed organizations, corporate culture and other realities may not make a centralized strategic or operational BISC a practical proposition. However, even in such an environment, it's still possible and often beneficial to centralize BI infrastructure (servers, DW, ETL, and BI tools) and let each individual line of business and functional department manage its own prioritization and BI application development, while leveraging the centralized BI infrastructure. Developers are the ultimate customers of these more narrowly scoped BISCs, and in several real life examples Forrester found that this is a practical limit of how much responsibility a BISC can take on without running into "turf battles." Information and knowledge management pros must determine whether their organizational culture is ready to support BISC beyond BI infrastructure in scope.

Consideration 6: Type of Funding Model?
BISC can be treated as a corporate cost center, and all departments across the enterprise can use and benefit from BISC services. The difficulty here is that this approach carries a stigma of "just another IT department/cost center." Furthermore, departments that are not yet set up to take advantage of the BISC will push back on carrying part of the cost burden. A cost allocation model based on the actual usage of BISC services can be fairer, but detailed, activity-based cost allocation models can be tricky to set up, implement, and manage.

Consideration 7: Narrow or Broad Scope?
Though a BICC-ish BISC is certainly possible, it's not preferred. Forrester recommends business leadership and business-led governance orientation, not a technology-centric focus, for the BISC. The same road map principles that apply to the best practices of implementing BI apply to the BISC: strategy first, architecture next, technology last. In its ultimate breadth of scope, BISC could encompass as many as 20 major components, roles, and responsibilities (see Figure 6 in Forrester's complete report), so it's very important to start small and increase the scope slowly.

Consideration 8: Performance Measurement Approach?
BISC stakeholders require transparent measurements of the success of the BISC program in order to support ongoing momentum and funding. BISC leaders must establish a clear set of BISC performance metrics and clearly communicate them on a periodic basis. Some BISC performance metrics are obvious and easy to calculate. Examples include number of BI applications delivered and maintained by BISC, number of BI users, number of reports in production and the usage patterns of these reports, reduced BI support staff, and reduced BI software and maintenance costs. Other metrics could be trickier to calculate and monitor such as improved information accuracy and turnaround time on BI requests.
Most leading BI products come with pre-built applications to monitor and analyze at least some of these metrics. If such out-of-the-box applications are not available from your preferred BI vendor, or if you are using multiple BI platforms, a centralized BI metrics management solution can be architected by using products from vendors such as Appfluent Technology and Teleran Technologies.

Consideration 9: Isolated or Aligned With Other Solution Centers?
No BI environment is an island from the rest of the data management infrastructure. Just as BI applications touch, depend on, and overlap with many related processes and technologies, BISCs cannot exist in isolation from other competency centers, solutions centers, or centers of excellence. Federation between the BISC and other data management competency centers is a best practice. Many such competency centers have existed in organizations for years, though they may not be recognized as distinct disciplines or organizations. Essentially, any group that defines, approves, and/or enforces standard practices for new projects or initiatives in any of these areas is a competency center.

To realize the full return on investment from BI, your organization's BISC should engage with all or most of the interdependent competency centers.
To succeed in your specific organizational environment, your BISC must have clear lines of demarcation, cooperation, and integration with all these other relevant initiatives. Failing to define a clear charter with appropriate collaboration, communication, and change management processes between complementary efforts can be a fatal pitfall in your BISC initiative.

Download the Complete Report
This article is based on the Forrester Report "Implementing Your Business Intelligence Solutions Center." The 16-page report includes charts, diagrams and seven recommendations not included in this article. Click here to download the free report.