The BI and Analytics Treadmill

Many years ago Thomas J. Watson said: "Once an organization loses its spirit of pioneering and rests on its early work, its progress stops." In the current economy, with much of the insurance marketplace appearing as a zero-sum game or perhaps even a temporarily shrinking pie, insurers are on an innovation treadmill they must keep pace with to stay competitive. And today, one of the most important competitive paradigms to embed in every insurance enterprise is a rock-solid business intelligence (BI) foundation.

Analysts and others have opined BI and predictive analytics have been two of the most important technology trends for the past few years with expectations they both will continue to be defining efforts for the most progressive companies in the coming years. From a base in BI, advanced predictive analytics capabilities can be used to turn raw and synthesized data into insights that can help improve performance and generate advantage through end-to-end fact-based decision-making.

There are a number of critical activities and milestones that must be reached for an organization to construct this end-to-end BI foundation. Despite investments in various BI developments, some insurers have struggled with aspects of the effort while continuing to drive forward with mixed success. While many components are required for competency in BI, certain key considerations must be planned and executed with excellence. Now seems like an opportune time to review some of the grounding concepts for BI to help companies better understand and implement the basics before too much more effort is expended. Some examples of important considerations follow.

1. A data czar should drive a foundational enterprisewide information inventory. Insurance companies generate and gather an enormous amount of data--many say data is one of an insurer's most valuable off-balance-sheet assets. Yet scarce few companies manage their data with the same level of diligence and intensity as they do assets such as real estate, human resources, or their investment portfolios.

All companies likely have a chief information officer, yet many companies might admit they need to allocate more resources focused on the gathering and management of data, as a precursor of information, rather than devoting so much to the technology side of BI. While sound technology is important, hardware and software alone will not get a company where it wants to be.

To expedite progress, companies should consider creating a leadership role reporting to the CIO that has a job description likened to that of a "data czar" or a chief data officer. This role should be filled by a person who excels in the skills necessary to manage the numerous issues of enterprise data, has a passion for creating a data-driven organization, and can lead the rollout of BI. Surprisingly few companies have created such a permanent position to lead and shepherd the development of essential data assets.

From a project charter perspective, perhaps the most important foundational effort for a data czar to spearhead should be a holistic organizational information inventory, if one hasn't been done already. This is a time-consuming effort that should aim to scour the company's electronic and paper files, documents, and systems for the vast assortment of historical data typically available for policies, premium, claims, losses, billing, agency/distribution management, underwriting, quotes, call center detail, regulatory/compliance data, competitive intelligence, marketing/company-specific customer information, and other internal company data types.

It is essential to identify the external data scattered throughout the organization where specific vendor relationships and data purchases provide sources from the likes of commercial business and personal/consumer credit bureaus, other bureaus such as ISO NCCI AAIS, Choicepoint, Polk, MIB, MVRs, U.S. government sources, Census Bureau data, etc. It is common for external data purchases to have been made for production purposes and special projects, but the identity and location of this data is known to only its localized users. Since this data is likely owned, licensed, and paid for by the company, it can serve additional purposes and provide greater organizational value than the original intent. In any case, just knowing what it is and where it is located is Job One.

However, performance of an enterprise information inventory should not be overthought--the focus should be on practicality and simplicity vs. elegance. The primary objective should be to get it done. The focus should be to find, organize, and make the data available quickly for BI and analytics efforts so ROI opportunities can be realized sooner than later to help fund additional BI projects.

2. Don't obsess over readying enterprisewide information for use. You don't need sanitizing gel for your data--it's never going to be completely clean, complete, and documented, so keep this in mind when developing data marts for BI and analytics. Too often companies strive to repair and clean their data to perfection when that extra effort doesn't add proportional value. Also, too often companies spend vast amounts of time and resources designing and developing comprehensive data schemas with the objective of design perfection. However, industry experience has shown many BI efforts have taken too long and cost too much as data technicians tweak and tune a company's data while database architects work to create the perfect data warehouse with the most ideal normalized relational database--all resulting in a project scope and time line that drift far beyond the original project vision. These efforts can result in database structures too complex for mere mortal BI users.

A step back from the overall effort should be taken to ask whether the project is on the planned course and whether it continues to be focused on the essential objectives. Recognize that while good database and data design are important for technical query efficiency purposes, since disk storage is cheaper than ever before and processor speeds are exponentially increasing, it makes more sense to focus on getting BI ready for use by making reasonable sacrifices on storage and query efficiency.

Remember, too, your data already is being used to run your company, generate your financials, collect your premiums, and pay your claims, so it can't be too bad. Many companies literally obsess on efforts to clean and repair enterprise data or make it more complete than it ever was historically. Important projects can stall or get back-burnered waiting for the preliminary project work to be performed. Insurers should challenge the common thinking their data is very dirty and needs to be thoroughly cleaned before other projects proceed--industry experience shows it typically is clean enough, credible enough, and complete enough and reflects the company's past well enough.

Focus on the form and functionality that will be used most and that will provide the majority of business value. Think of all of this like a PC-based spreadsheet program commonly used in the insurance industry that has such a comprehensive feature set that a large number of the functional widgets never will be used. Similarly, when building enterprisewide BI information repositories, be sure to spend the needed time on the most necessary features and not on the arcane things to which many project teams naturally can gravitate, because it is human nature to try to find the "gotchas" in any effort.

3. Leveraging analytics is essential, but it takes the right people and tools. Industry experience has shown not every insurance professional has the aptitude, patience, or interest to become proficient in BI software or to learn how data is structured or how to do statistical analyses. By creating a team of BI gurus, the data czar can have, or can develop, individuals with the necessary passion and institutional knowledge to learn the organization's data structures, data history, and the software tools. This new group of internal specialists can be offered career-defining leadership roles to serve as the go-to people for BI within a company. Doing BI work is not for everyone and companies need to recognize this and staff accordingly. Companies that efficiently mobilize this new analytic SWAT team can realize faster, cheaper, and more efficient BI results than those companies that believe everyone throughout the company can learn BI methods or perform advanced analytics.

With a small group of BI missionaries embedded throughout a company, it becomes possible for management metrics and dashboard criteria to be developed as part of the foundation for an enhanced data-driven organizational culture. These new quantitative management mechanisms can provide corporate leaders with real-time views of the organization and can help initiate the development of proactive and reactive processes--processes evolving from advanced analytic data mining and predictive modeling tools such as:

o Underwriting and right-pricing predictive models.

o Risk portfolio management models.

o Claims/fraud management models.

o Price optimization and market demand analytics.

o Customer acquisition, retention, and sales/marketing analytics.

These and other techniques can be used to implement processes with high ROI potential when deployed effectively and on time. Companies that innovate and progress beyond the tried-and-true can develop new competitive advantage.

Gaining BI Insight

BI should provide management with the insights to help improve business performance through enhanced decision-making. To do so, companies must create a culture of BI innovation, and progress should be made with urgency by focusing on priority problems that need attention. It is important in the current market climate for companies to do more with less, seek continuous improvement, work methodically, and take advantage of organizational experience. Ask the people, at all levels and from all functions, what vexing business problems they see as requiring attention; circulate these ideas to seek feedback and prioritization; and work to finalize the organization's project docket. Focus on things you don't yet know or those things you need to know more about. Focus less on refining or rehashing learning that is already known or routine. Look for ways to challenge "tribal wisdom" and validate or invalidate facts that may inhibit new thinking or opportunities.

With the recent tumultuous events in the financial services and insurance marketplace, looking for answers to new business questions and resolving challenging problems never have been more important.

John Lucker, principal, Deloitte Consulting LLP, leads Deloitte's Advanced Analytics & Modeling practice. He can be reached at JLucker@deloitte.com.

The content of "Inside Track" is the responsibility of each column's author. The views and opinions are those of the author and do not necessarily represent those of Tech Decisions.