Deeply decrypt Obama's big data-precision marketing Campaign

There is no difference in nature between political elections and product marketing, both of which are the process by which candidates are ultimately "sold" to voters through various marketing tools. Thus, with the further integration of social networks and large data analysis and marketing, the U.S. presidential election has evolved into a precision marketing showdown based on Big data analysis (the 2012 Obama campaign's number of scientists is five times times higher than the last one). In the article "How data scientists help Obama TKO Mitt Romney", we reveal the Obama campaign's winning password--social analysis + personalized precision marketing, but the tactical details, new tools used, new methods, etc. "micro-innovation" are still unknown.

More recently, an in-depth report from the MIT Tech Review detailed the data analysis war in the U.S. general election. The ideas and information in this article are of little use in our country's elections, but they are useful for corporate marketers-as the Harvard Business Review puts it: data-driven decisions can leave managers no longer reliant on intuition. Instead of replacing the president with "a brand of milk" and replacing voters with "consumers", you can interpret the U.S. presidential election as a precision marketing case that combines big data analysis. The following will compile the report for everyone as follows:

In 2010, two years after Obama's arrival in the White House, the Democratic Party was defeated in the mid-term elections. The Democratic Party's advantage in Congress has ensured the success of Mr Obama's financial reform and health-care reform. After the 2010 mid-term elections, the Democratic Party's position in the House of Representatives was replaced by the Republican Party. In the Senate, it is only by a narrow margin that the seat advantage is barely preserved. If the mid-term elections were the midterm exam of Obama and the Democratic National Committee, it was clear that the result was "failing".

"Failing" is not necessarily a bad thing at all. At least that's what Dan Wagner thinks. The data analyst was appointed by the Democratic National Committee as the director in 2009 to collect and analyze voter information to help the democratic election team conduct election campaigns by telephone and direct mail. Dan Wagner entered his statistical model with the collected voter information and produced a series of analytical reports on voter attitudes and preferences. He asked the technical engineers of the Democratic Party to develop a software program on that basis. He called the Software "Survey Manager".

The forecast error is less than 2.5%.

This fall, in a special election in the house of New York State, a few days before Election day, Dan Wagner's survey manager predicted the correct election results, with less than 150 votes to predict the actual results. One months later, with the death of Senator Edward Warkenedy of Massachusetts, Massachusetts held a special congressional election to fill Senator Kennedy's seat. In this campaign, it is widely believed that Democratic candidate Martha Coakley will undoubtedly prevail in the traditional Democratic stronghold of the state. Dan Wagner's survey manager correctly predicted that Republican candidate Scott Brown would eventually win. These forecasts have attracted the attention of the Democratic National Committee. Jeremy Bird, deputy director of the Obama national campaign, said: "When you correctly predict that we can win is one thing, and when you can correctly predict that we will lose, it is another matter." ”

The "Other thing" is a prediction that was made 5 months before the loss. Since June, Dan Wager has built a model to predict some of the Senate election results in 74 constituencies in the congressional election. The results of these predictions are "incredibly" accurate. Dan wager did not take the traditional method of sampling analysis, but rather a voter statistics. He initially predicted that Democrats would have trouble in the midterm elections from the thousands of-Democratic poll calls that were compared to the Democratic voter database. According to the survey call feedback. The Democrats ' hard-core voters say they are going to vote much lower than the percentage expected in the probability statistics. and wager can accurately measure the rate of voter support that the DPJ can raise for every campaign. From these analyses, Wagner argues, the effect that the party's propaganda campaign can do is not enough to compensate for the gap with the Republican Party's support rating.

The gap between his predictions and the end result is less than 2.5%. "This makes many people who do not understand the mathematical models behind these predictions understand the value that these mathematical models can produce." Mitch Steward, director of the Obama campaign, added: "Since the end of the special congressional elections, his predictions have become the golden rule in the Democratic Party." ”

(Responsible editor: The good of the Legacy)