Thursday, November 8, 2012

Data Analytics wins 2012 US Presidential Election

Data analytics was the big winner in the 2012 US Presidential race. In fact, 11:17 PM (US ET) November 6th was the moment data analytics went mainstream. This was when Ohio was officially projected to go to Obama. It was the ultimate validation for Nate Silver and his data analytics approach to election forecasting. To much fanfare he accurately predicted the results of the election in all 50 states without doing any of his own polling. He used sophisticated analytic models based on data from as many third party polls he could find. To this he added the secret sauce of data analytics - a keen understanding of how different types of data from different sources relate to one another in context.

His FiveThirtyEight blog drove as much as 20% of the web traffic to the New York Times website - the 6th most visited US news site on the net - leading up to the election. As a result, data analytics is officially mainstream. Any business leader at any level that does not immediately embrace its power is putting his or her career and company in jeopardy.

Data analytics works. It does not produce miracles, but it does produce results that far outperform human judgment on its own. The Obama campaign employed an army of retail data analytics wonks to beat the Romney campaign in every battleground state. They did it by applying analytic techniques proven in the supermarket industry:
  • Standardizing records: Unifying the customer (voter) database
  • Widening perspective: Combining diverse data types: demographics; buying/voting history; response by media; donation/activity by trigger (celebrity dinner), model (contest) and method (mobile); group/church  membership, social networking activity (Reddit), etc.
  • Judicious targeting: Carefully identifying the potential for influencing voters that could influence the election. Not worth targeting easily influenced voters if they don't live in a county that can help swing a state. Not worth targeting difficult to influence voters even if they live in a critical county. This is essential for achieving impact and ROI.
  • Media mix modeling: which media channels have the greatest impact on which kinds of voters?
  • Action oriented outreach: Understanding the specifics of why and how certain people act and designing multiple outreach experiments (progressive offers, channel mix, social references, etc.) based on that.
  • Openness to innovation: data driven models may point to approaches that are counter intuitive for some decision makers. They can seem risky and mysterious. They will not be right all the time. Controlled risk is part of the evolutionary process to effectiveness. Without a tolerance for experimentation however, you will not develop a data driven culture, you will in fact kill it.

Marketers in the world's largest high tech companies are finally acquiring the enterprise data services needed to apply data analytics to long cycle B2B customer creation processes. We are already seeing signs of how significant the impact of these new approaches to marketing and sales can be:

  • $200M EU lift based on a sophisticated solutions recommendation engine
  • 45% more subscription revenue with no increase in a multi-million dollar marketing budget
  • Tens of millions of dollars in revenue uplift from simple web behavioral changes

Embracing data driven decision making is now a matter of survival. You simply cannot win against competitors that have faster, deeper market insight. They will beat you in every stage of the customer creation process. Your marketing will be months behind, your inside sales reps will be calling customers already committed to alternatives, your field sales reps will miss opportunity after opportunity to get more revenue from existing customers. Your funnel will collapse, your pipeline will dry up, your renewable revenue will shrink, and at that point it will be hard to recover. Hyperbole, you say? In the great A/B test of who uses data analytics and who does not, stay in the B group at your peril.

IDC EAG group has done extensive research on the key ingredients needed to create the enterprise data services that are a prerequisite for data driven customer creation and has ongoing research into how to create a data driven culture. To find out more please contact Gerry Murray - gmurray(at)idc(dot)com. 

No comments:

Post a Comment