Siroker described how his team used the power of data analysis to raise an extra $44M or so for the Obama presidential campaign. Here are his five keys to success in getting actionable and profitable results using data analysis:

1. Define quantifiable success metrics. For Siroker, these were donations and votes.
2. Question assumptions.
3. Divide and conquer (segmentation of data).
4. Take advantage of circumstances – be prepared to leverage unplanned events.
5. Always be optimizing – continual improvement.

Siroker’s results speak for themselves!

SoCon is a conference on Social Media and Social Networking held in the Atlanta area every year since 2007. I attended in 2008, and I highly recommend it for anyone interested in this space.

Here is how SoCon describes this year’s conference:

“In the first three years, we introduced social media, user-generated content, blogs, podcasts, video logs, social networking, wikis, Twitter… but nothing stands still. Find out what you have to know in 2010 to stay ahead of the learning curve. Find out who is doing great stuff; who has great, innovative ideas. Network and learn — and maybe even partner with — independent content producers, new media pros, academics and people from across the spectrum of marketing, public relations, legal, human resources, and executive ranks.”

1. Maintain consistent leadership over projects implementing tools for data analysis. It is all too easy to continue to rely on the gut instincts and subjective assessments of the past. Strong leadership is required to break down old habits, evangelize the benefits of sustainable, data-driven approaches to running the business, and even challenge special interests who do not want mistakes revealed by evidence and facts.
2. Do not leave staff to wallow in the misery of collecting and managing data. Their time and effort is very valuable, and it is better spent analyzing the data and developing recommendations from it.
3. Build analytic systems that track changes over time and place. This is the only way to monitor performance and to develop benchmarks that serve as the basis for improvement.
4. Integrate systems across groups and divisions as much as possible. This leverages effort and creates larger potential for creating strategies that are optimal for the entire business and not just specific entities.
5. Carefully assess the make vs buy decision. If the company lacks the skills and/or the will to see an implementation through to its successful conclusion, then hire a firm that demonstrates those capabilities AND can train staff to maintain and use the system for the foreseeable future. Make honest assessments about whether or not the company truly has the resources to hire the correct staff and buy the correct hardware for making an analytic system work. Understanding the return-on-investment (ROI) of making data-driven decisions will help focus make vs buy decisions.

Follow these steps, and your business will be well on its way to executing data-driven strategies for improving business performance.

In my graduate department at Stanford University (Engineering-Economic Systems at the time, now named Management Science and Engineering), we had a saying: “Mathematical Modeling for Human Solutions.” We took pride in taking the tools and techniques of math and statistics to solve practical problems. The article correctly points out “though at the fore, statisticians are only a small part of an army of experts using modern statistical techniques for data analysis. Computing and numerical skills, experts say, matter far more than degrees. So the new data sleuths come from backgrounds like economics, computer science and mathematics.”

I was particularly encouraged to read that Peter Orszag, Director of the Federal Office of Management and Budget, has a keen interest in using statistics to drive sound policy.

In other words, if you are not taking advantage of data analysis to address some of your most vexing problems, you are probably going to get left behind!

Without data, we can have no impact. How can anyone really know whether or not a tax holiday not only works or is even worth its cost to the state budget? Certainly, everyone enjoys tax-free shopping, but the cost may outweigh the benefits if, for example, the state is not able to fund other important projects or increases taxes somewhere else to make up for the revenue gap.

These lessons apply in business as well as government. Without data to measure performance, business plans and strategies are subject to the whims of “gut instinct” or personal biases and subjectivity. One person’s success can easily be another person’s failure and power relationships may win the day instead of what actually improves profits.

How could the state of Georgia attack this problem? The first step is the collection of the retail data. This data needs to be daily so seasonal and cyclical patterns can be accounted for. The second step is to consider controlled experiments. For example, run the state holiday on different dates in different counties, skip a year, or change the dates around from year-to-year. In other words, establish a set of data than can be used as the control for comparison of performance. If the quantity of data or the number of stores present large obstacles, establish the data points for select stores that have an established record of sales in the local community. At this level, every effort should be made to track the specific items that consumers purchase as well.

I strongly suspect that if the state of Georgia applied some simple analysis to this program, the results will surprise them!