Something Wicked This Way Comes… It’s BI season in Florida

Usually it is around this time of year in Tallahassee when Business Intelligence (BI) vendors descend on government CIOs and executives to show the latest tools for slicing-and-dicing government data into something that may bring transparency to government or save taxpayer money.

The latest news of shortfall in the state budget will prompt many state agencies to aggressively seek out opportunities to do more with less. Add to this scenario the ever-present stimulus money from the  2009 American Recovery and Reinvestment Act (ARRA) and you have an orchard of seemingly low hanging fruit with an army of BI vendors and lobbyists waiting to help each agency pick their share.

While a well thought out Business Intelligence solution could probably have a dramatic impact on optimizing an agency to meet its goals at the lowest tax payer expense, there are a few points (truths) for the government executive or CIO to consider as he or she listens to each vendor’s presentation.

First – There are no quick or cheap Business Intelligence solutions. No matter what the vendor tells you.
It is true that some of the major vendors may have products which significantly reduce the effort required to create a BI solution. When considering these rapid off-the-shelf solutions be honest with yourself about your expectations; you want a solution which will tell you something about your organization that you don’t already know. Once you get beyond the presentation (sales pitch) and get to an estimate on the amount of time it could take the vendor to show you real, actionable data specific to your organization you may find it to be much more effort (money) than you were originally led to believe.

Second - A vendor is not likely to know your state government business better than you do.
When a BI vendor shows you a neat graphical dashboard displaying dials, and knobs, and charts full of indicators on the health of your organization you should be asking yourself and the vendor “who decides what symptoms make my organization healthy or sick?”. The common executive dashboards in private sector organizations display indicators geared toward sales trends and profit. None of these indicators normally apply to the “health” of a public service organization. So who will develop your specific measures? Your organization will. So back to item number one; the “quickness and cheapness” of the overall solution is going to depend on how fast your own management team can develop strategic measures for your organization.

Third - Data integration is never easy; especially in government.
Government information systems are basically a patchwork of ad-hoc adjustments designed to meet the latest legislation, tack on improvements, or apply data integrity fixes. You can see evidence of this by reviewing the long list of Legislative Budget Requests each year asking for funding of IT projects to “re-engineer” or “consolidate” systems. I will certainly not lay blame on the state agencies for not having the fore-sight to build monolithic, integrated systems that do all things for all people – that notion is just plain silly. The current ad-hoc nature of state government systems is a direct result of limited budgets. The fact that there is generally no income generated by a state information system means not much measurable return exists to warrant the significant investment a monolithic system would require. Also, with no income there is virtually no measurable return on investment to fund regualr upgrades to these systems. The integration of systems and data is probably the greatest challenge facing any state government organization and will likely be the most expensive IT problem an agency will ever face.

So, with that said. What sort of magic beans could a BI vendor possibly drop which would immediately integrate all of your data into a neat little package and generate factual measures on that pretty display? What if the vendor integrates some of the data incorrectly? Will you find the error before you make an operational decision based on that faulty data or worse report that bad information to the public? How much time and cost will be required from your organization to test and verify the “facts” which are being presented by that BI solution are correct? (back to point #1 – nothing is cheap or quick)

Fourth - Who will feed, walk, and play with this new puppy.
Who in your organization will be using this BI solution? If it is just your immediate executive team then how much money is it worth to you to see your agency’s statistics on the big screen? Being logical business men and women you will likely say the data and indicators would be most useful to operational managers further down the chain of command. With that reality in mind, what you are seeing in the vendor presentation is only the tip of the iceberg. The underlying data presentations required for operational reporting are much more detailed and may require months of group discussions (your staff resources) just to pin down what is meaningful operational data and what is trivial.

Now here comes million dollar question… Who in your organization will maintain this BI solution? Assuming that today you don’t know each and every measure you may ever need to run your organization, who will make the necessary changes to the technology solution to address the new measures and indicators you discover down the road. Better yet who will correct the faulty measures that were not discovered when testing the first release? Will it be this vendor? You can be certain that most technology vendors will not have a graceful exit strategy or transition plan and are not even looking for one. You can also be certain that your IT organization is probably not prepared to support a data warehousing solution that can grow as big and as fast as a BI solution can.

So How Can You Approach Business Intelligence in State Government?

This article is not intended to scare you away from Business Intelligence, just to warn you of the current pitfalls of vendor presentations for BI and just about any other technology solution. The moral is don’t buy into “silver bullet” solutions or you will waste what little budget and time you have. Take your BI strategy one day at a time and in small bites. Focus on fixing the underlying system integration issues and a few intelligence measures may actually appear on their own through improved operational reporting.

One solid approach for state government is to build multiple, smaller BI solutions in specific business areas. At its core, a BI solution consists of a data warehouse pumping data up to a decision support tool (dashboard, operational reports, etc.). The data warehouse is the consolidated data from all of your systems with additional data mixed in to provide historical perspective.

In business intelligence terminology a smaller data model exists referred to as a data mart and you can create a data mart for a particular subsection of your organization – maybe a single business unit or agency function. Building a data mart is a much smaller investment and might possibly be supported (or even created) using your existing in-house IT resources. In fact it is much easier to integrate multiple, well designed data marts into a larger data warehouse down the road than it is to create one big data warehouse from the beginning.

Pick one area of your organization and try out a small data-mart with a basic dashboard. If nothing else you will at least gain experience on what is truly involved in the process.

UPDATE – September 28th, 2009:

The Recovery Accountability and Transparency Board unveiled their revamped Recovery.gov web site today. “The site was revamped at a cost of $9.5 million in stimulus funds and provides easier-to-use tools, such as a ZIP Code search that shows stimulus projects in local communities” (LA Times: “Website upgrade makes it easier to monitor federal stimulus”)

At first glance the site just appears to be re-arranged, maybe a bit less photogenic, but with a clear emphasis on the information. A further review reveals that the web site is actually an ASP.NET web application with multiple zones containing data indicators. Yes, this appears to be your basic BI dashboard.

So why am I posting an update? Well a comment in the LA Times article “Website upgrade makes it easier to monitor federal stimulus” from Eric Gillespie, the chief information officer of Onvia (the company behind Recovery.gov) caught my attention.

“This is not about technology. You could build the sexiest user experience on the front end, but unless you have the data on the back end it doesn’t matter,” Gillespie said. “It’s not about the map.” – Well that says it all.

At 9.5 million this was clearly no quick and cheap solution, this appears to more like a planned attack at a true business intelligence solution. Still for the volume of data the ARRA stimulus is likely producing and the fairly rapid turn-around for the dashboard and underlying data back-end this seems to be a rather rapid deployment of a BI solution. A bit of further research into Onvia and I uncovered from their corporate web site just how this was possible:

“For more than twelve years, Onvia has delivered unparalleled coverage of government purchasing activity from federal, state, local and education agencies…”

“Onvia covers over 89,000 federal, state, local, and education purchasing offices and is tracking the government economic recovery projects. New opportunities are added to the Onvia database daily.”

Well there you have it, I will say once again “you just can’t cheat at business intelligence”. You either know the “ins-and-outs” of the industry and how to analyze the data or you don’t. Onvia evidently has invested a significant level of time and resources in the government purchasing and procurement industry and therefore they were in a unique position to leverage their knowledge investment to provide this specific BI solution when needed.

Advertisement