Big Data myths in the video industry: gimmicks or tools?

The big fire in the card house this year has made Netflix, the original online rental DVD company, a model for the success of algorithmic programming. But in the opinion of Youku's director Chenhanze, "Solitaire" is simply a product of Netflix's precision marketing, not a drama that relies on big data analysis. "It was said that Netflix's boss was listening to his subordinates and decided to take the play's adaptation and distribution rights, and that the subordinate was a big fan of the British play" card House ". ”

Chenhanze, a media man, is wary of big data. She thinks big data is just a tool. Before large data, there are already statistical logic and data mining techniques. For example, the first application of large data in the Internet is traffic analysis. Around 2005, the Internet began to access through the portal, is the main entry of traffic. But by the year 2011, the biggest traffic comes from video sites.

Before the combination of Youku, Chenhanze was in charge of the homemade drama business at Tudou. The data model, she says, accurately identifies which minute the user leaves, looks back, and drags and watches in each play. These data provide great reference value for the choice of theme and design of the bridge section. In the refined production process of Hollywood screenplay, "script Doctor" plays a very important role, they usually adapt the script to strengthen a plot, highlighting a bridge. and large data analysis technology is Chenhanze mouth "script Doctor", with large data can analyze the audience like the plot elements.

The top ten films on Netflix represent the needs of different family members to visualize the movie. In addition, 10 films can be corresponding to different time, in different emotions under the needs of consumers. Behind this is the multi-dimensional analysis of the audience. Chenhanze says that video sequencing is a set of clear logical systems that understand sequencing and goal guidance to be more accurately recommended to consumers.

Five years ago, the explosion of a video on the network was usually based on a pattern of dots diffusion, which was spontaneous and shared; But as the network video industry gradually turned to making homemade dramas and purchasing TV programs, video companies had to have a quantitative analysis of the audience. The data report based on large data analysis can be used by film and television producers, advertisers, and third-party statistical agencies.

Advertisers are no longer satisfied with the traditional patch advertising marketing model. The original video site attached to PV, now becomes a more critical indicator of ROI. Ma mentioned in a recent speech that connecting everything, Internet + innovation emerges, open collaboration, consumer participation in decision-making, data becoming resources, the courage to conform to the trend and the risk of connecting everything. The trend of big data marketing is the rise of mobile end. "Chenhanze said.

In the PC era, using cookies and logical analysis to track and analyze user information is a common tactic. But as mobile browsers do not have cookies, coupled with the indulgence of cookies for user privacy, giants such as Google are starting to look for new models of data analysis.

Chenhanze that large data models could be used in the future. For example, a user opens a Web site first on a mobile phone, and after a long visit, it generates interactive behavior that will be shared to the next site he browses to. This must be based on the assistance between A and B sites, and they may be inextricably linked. All data is shared, which means "connect everything".

"Google's PPC model (for every click of Google online ads, advertisers need Google to pay a certain fee) may soon not apply." Chenhanze said, the future model is that users go through a website, b site, until n site, advertisers to achieve the final profit, but the previous browsing process can also participate in the profit sharing.

However, large data technology has its limitations, including the limited sample selection, data screening process of true and false judgments, analysis of the correctness of the logic, can affect the final results.

Victor Maire Schoenberg, author of the Big Data age, argues that the big data age only needs people to know what it is, without needing to know why. Just as businesses do not have to pay attention to why beer and diaper sales match so much, as long as they capture business opportunities, adopt a crossover strategy to boost performance without having to figure out the business logic behind it. If you have to analyze everything clearly before making a judgment, the loss is the horizon and the cost of time. Relevance is far more than causality, but it is an overreaction in Chenhanze's view. "Are all the pages sorted through the data model and no longer needed for editing?" ”

She thinks that selling stories and telling stories is immortal, and that editors still play their part. For example, traditional video sites with action movies, comedy, love movies and other categories, but on Netflix, there is a very small category, such as cult film is also regarded as a category, this is based on the editor of the content of the understanding and wisdom of the human brain products, which data can not be crawled.