Overthe past decade, powerful business intelligence (BI) applicationshave been developed that allow business leaders to build new levelsof insight into performance. These capabilities promise to speedour identification of performance issues and generate improvedresults. Let's take a look at the practical challenges inherent inexpanding and realizing the value of these capabilities.

To begin, making the decision to acquire or expand the BI toolsat your disposal does not cause insight or capability toautomatically spring to life. There is a technical side to theseefforts, one that involves housing data and learning to work withthe technology. The breadth of reporting in these environmentsevolves over time as data and functionality are incrementallyadded. It is common to have traditional methods for acquiring andanalyzing information coexist with new functionality for a longtime. In fact, the need for specialized skill sets to bringtogether data and manipulate it in familiar tools like spreadsheetsdoes not go away, although the dependence on these approaches canbe reduced.

There is, more importantly, a strategic side to this effort.This side requires business leaders to decide what information isimportant, why it is important, and how it will be used. The rangeof possibilities in highlighting performance hot spots, slicingdata, and drilling to potential problem areas is enticing. It iseasy to be seduced by the possibilities. Remember, though, thathaving more insight doesn't make your organization more capable ofacting on that.

Making effective use of these capabilities requires planning. Toachieve enhanced reporting and insight, you must preparestrategically by doing three things: First, select and organizeyour measures. Second, develop a discipline for exploring data thatfocuses on root cause. Last, teach your managers and leaders how toreview and act upon data.

Select, Organize Your Measures

As the amount of data grows and your access to it increases, youcan quickly lose track of critical relationships among data.Building new reports becomes easier but can lead to a confusingarray that loses focus on what is important. Organizing data meansidentifying the operating outcomes that are most crucial to you andusing them as your mainstay.

The first step in gaining better control is to generate a listof measures that you care about and then put them into groups.Different companies will identify different measures; however, theyfall into four distinct categories:

• Quality measures, which include the averageand median amounts paid, quality review scores, and numbersindicating procedural compliance (such as alternative partsutilization in physical damage or validation of codes on medicalbills).
• Efficiency measures, such as cost per claimand claims per adjuster.
• Cycle time measures focus on speed ofcompletion and generally revolve around the time a claim isreported until closure.
• Customer measures often revolve around theresults of customer satisfaction surveys.

It is critical to select a small set of key performance measures(KPMs), as few as one measure for each of the categories listedabove. One should note that the average or median paid is a riskykey measure. Making it such can result in unintended actions thatare good neither for you nor your customers). These measures tiedirectly to your strategic objectives and are the ones that peopleat all levels of your organization know.

A small set of diagnostic measures should be identified. Thesenumbers have a direct relationship with your KPMs and provide acontext for more rapidly understanding your results. For instance,claim cycle time for a property damage claim might be supplementedwith measures such as the average time to make contact and completeinspections. If overall claim cycle time starts to slip, then youcan generally spot the trouble areas quickly in one or more ofthese diagnostic numbers.

Create a balanced set of measures that ensure your leadershipdoes not push results in one area only to negatively impact them inanother. Hold your leadership to a balanced assessment ofperformance. That is, set a standard that one-dimensionalperformance improvements do not make for excellence. Without astructured set of data using a format like the one described here,different managers will form different views of what is important.Consequently, they will react in completely divergent ways to getresults.

When choosing your KPMs and diagnostic measures, combinequalitative and quantitative information. Qualitative measuresrefer generally to those that are derived from observations bypeople about work. Claim quality scores and customer surveys aregood examples. Quantitative measures are derived by applyingalgorithms to the data in your systems. Calculating cycle time is agood example of a quantitative measure.

It is easy to focus on the quantitative and exclude thequalitative as you examine the data available. This is whereconsideration of measures is critical. Leaving out qualitative datadeprives you of the human observations most effective inidentifying root cause and understanding opportunity. In short, itproduces an incomplete picture that can cause reactions to symptomsand without curing the disease.

Root Cause Analysis

Use performance goals for each KPM to guide assessment ofresults. Having measures without goals makes your purpose ambiguousand can prevent being able to gauge progress. Any result can feelsignificant, while none seem to provide guidance. When specificperformance goals are identified, managers and leaders know whatresults to scrutinize to pinpoint causes of change. This startingpoint is critical for understanding what (if any) action should betaken.

Root cause analysis arises in two different scenarios, both ofwhich should use a “hypothesis-confirmation” approach. Identifywhich situation you face. The first scenario involves investigatingchanges in results where there is some certainty as to the cause.For example, you may notice a spike in the average age of closedclaims when reviewing cycle time. You may also know that one ormore teams within the organization started efforts to reduceinventory by focusing on closing out older cases. With this as yourhypothesis, honing in on these teams to confirm they caused thespike is quick and easy.

The second scenario is more complicated and infrequent. Thisscenario occurs when you are uncertain about the cause of changingperformance. Before jumping into the data, you must methodicallydevelop hypotheses to explain the change. These investigations areiterative, with each round offering insight as well as questions.Refer to the sidebar for more information about crafting hypothesesand acting accordingly.

To understand whether you have reached the source of yourproblems, answer this question: “If I changed X, then could I beconfident that results would improve?” In the example used here,you might find that the change in procedure has had some impact.However, when you consider changing it, you may conclude this wouldnot reverse the problem. Keep looking.

This may seem like a lot of work. Being anxious for results andbypassing these steps, though, often has the impact of prolongingthe problem and, even worse, creating new causes for variance. Tofix the problem, you have to remove the cause. Anything else istampering.

Teach Users How to Review Data

One of the most important (but often overlooked) steps isteaching your managers and leaders how to use data. Assumptions areoften made that the data will tell the story and the intuitivenature of reporting will make not only insight, but also action,self-explanatory. The simple fact is, most of us struggle to reachaccurate conclusions without guidance. While we are surrounded bydata in claims, and yet there is little done to formally educatemanagers on using it.

Develop rules of thumb for deciding when a result needsattention. When a number moves in a negative direction in a week ormonth, this does not mean that it needs to acted upon. Big movesand trends are important to dig into. Decide what constitutes atrend and a big move. Otherwise managers should note the change andmonitor it.

Stay focused on plans that address problem results. In manycases, results don't change overnight. When deciding whether torespond to changes, managers need to consider the capacity of theorganization to deal with the change and the recent history ofchanges. Sometimes, you will be faced with issues you want toaddress but must put off because your organization is currentlyresponding to other challenges. Conversely, your organization mayhave the capacity, but you recently made changes in response toother, unrelated results. Implementing another change can create a“flavor of the moment” impression. It is counterproductive to haveyour team cringe each month or week when new reports come out.Discretion is the better part of valor in these situations.

Implement periodic commentary by managers about results. Havingnumbers is great. Teaching leaders to put together a cohesive storyabout them is greater. Monthly reports are drudgery when they areexercises in paper shuffling. When used as a platform to discussresults and insight, they are developmental exercises that createbetter leaders. Be sure to use these opportunities to improvejudgment.