I agree with all of Margaret's points. I would like to add two things though that I think are absolutely essential to maximizing the selection, implementation, and eventual ROI of a BI project.

The first is in having a performance oriented culture. Too many efforts fail because the KPIs or leading indicators are outside of the span of control of a person, team, or department. Moreover, these people may not have been held to account in the past, and might not understand how to have an impact on those metrics, or what resources they may bring to bear to hit those target. Spend the time to map out what it is that you do, how you need to measure it, and make sure that accountability is in place. Also, make sure that budget control mirrors KPI accountability to avoid the situation where performance suffers because someone else was controlling the purse strings.

The second point is on the subject of metrics -- you need to have relevant layers -- similar to the leading and lagging indicators that Margaret spoke of. A KPI or metric that is too coarse does not provide actionable information as to how to affect it. This coarse grained metric should be served by lower level actionable metrics. This most often rears its head when you are looking at BI that has QA components, but the lessons apply across the board. I always approach this from a dual perspective -- what are the top level KPIs senior management and the board need to assess the performance of the organization, and what are the lower level KPIs that middle managers need to assess daily or period based operations. The secret is to make sure that there are adequate dependence linkages established between the levels. That way, variations in the top level metric should be traceable to a variation in the lower level metric.

One last point on BI and the use of metrics. Implementing too many metrics can cause confusion. If you read Jim Collins book "From Good to Great", there is considerable discussion in there about metrics and the danger of over-measurement. Collins, in his study, found that at the strategic level, all of the BI and metrics could roll up into one coherent metric that effectively galvanized the efforts of the organization. An example he cited in the book was 7-Eleven and the metric of Revenue per Customer Visit.

Good Luck!

Bruce Kearns

Margaret Johnson

Managing Director, Oakwood Systems Group, Inc

Posted on Dec. 10, 2010

Jeff - great question. Let me share my top three with you.

One of the most interesting basics to tackle is that of language. Do the different groups within your speak the same language? Does the term "revenue" mean the same to the finance team as it does to the sales team? What about "profit?" Does sales consider the sale complete when the order is placed, while finance doesn't book the sale until the product is shipped? Coming to agreement on the language that you will use to measure the success of the company is critical.

Secondly, do your systems all use the same terms? Does one system say "client" while another says "customer" and still another says "account?" An effective BI system will combine information from several systems, so ensuring that those systems use common terms - or can be made to use common terms when reporting to the BI system - is important.

Finally, work with the organization to define a key set of leading indicators as well as lagging indicators. Most reports provide information about what has already happened, and it's too late to take action. What sort of predictive indicators can you put in place that can be monitored and managed so that the organization can take action in time to affect the bottom line? For example, for the lagging indicator of "sales" or "revenue," what are the contributing factors to that number and how can you measure them? If you can track sales back to how many sales calls are being made by the sales team, then the leading indicator might be "sales calls." If you manage to that number in real time, you'll be able to affect revenue.

A great resource on this topic is the book "Performance Dashboards" by Wayne Eckerson. I recommend this book to everyone who is contemplating a move into BI. There is a lot of great fundamental information in this book, including how to define clear and relevant Key Performance Indicators (KPIs), which are as critical to a successful BI implementation as anything I've mentioned above.

I hope you find this helpful.

Margaret Johnson
Oakwood Systems Group, Inc.