Hulu: Advertising can also be very love

Last month, Hulu CEO Jason kilar A very good report, sharing Hulu's growth story, revenue data and some insights into the future of the video industry. Hulu's success is definitely a marvel, especially in business terms, and he has won 40% of the market share of the equivalent of Youtube 10%.

In the report video, you can hear Jason praising the Hulu Beijing team! I have had the opportunity to visit their office in the Tsinghua Science Park, a very low-key team, but it is really interesting. Working in Hulu is downright internet model, the business team in the United States, the technical team in Beijing, the goal of a consistent global collaboration, not XX China and so can match.

Hulu can be described as an example of the pursuit of integration of business and technology. When others are still hesitant about personalized technology, they have almost completed best practices. Hulu has applied personalized technology to many corners of the product, demonstrating its strength in detail, and is the perfect research object for the Shanzhai innovation.

The picture above is classic "If you like ... So you might also like ... "The recommended scenario, which, although common, is fastidious." Where should I put the recommendation, what is the presentation and when? Think about how you're going to solve these problems. The Hulu team's answer is a/b Test, which allows data to guide product decisions. Look at the details, "bookmarked" is a very fit scene of the small function, is a heart design. The effect Hulu does is that 10% of the users who see this recommendation click to watch or collect the recommended videos. But in fact the whole scene inside I want to say, is the upper right corner of the "does this recommendation acquires you?" this. I have always had a view: for the recommended products, the role of user feedback can not be overemphasized, although the user is really lazy, do recommend products do have to pay attention to implicit feedback, but this does not mean that you can not ask users to make a choice. Should be able to collect valuable feedback, but also make users feel the product of respect for him, how to balance this contradiction, it is necessary to have considerable wisdom. If you are hiring to recommend a product manager, it's worth asking.

The user does not like the advertisement, this matter believes everybody to understand. Jason gives a data that "the content maker earns one dollar each, and 41 cents comes from advertising," so all the advertising talk about the future of the video industry is a cloud, and the conclusion is that it will be better off without advertising. Hulu's burritos is divided into two parts: the first part is "to-help arranges find and enjoy" the world's premium content; Where and how tightly want it. The second part is "as we pursue our burritos, we aspire to create a service that users, advertisers and content owners unabashedly love." Basically, the first part, whether from the content coverage or the coverage of the audience, has not done; Obviously, we all can see that the second part is the real burritos, ^_^. The second part of the product is indeed Hulu's most work place: 1 The quality of advertising is very high, this should benefit from the flagship brand advertising strategy; 2 around the ads to give users a lot of choices, the red circle shown in the figure is only one example; 3 based on user selection and feedback, Using personalized technology to drive ads is increasingly relevant to user preferences. As far as the current public data are concerned, Hulu has achieved quite good results in the three aspects of user satisfaction, advertising effectiveness, and revenue. As a pursuit of the process of young men, I used to be very entangled: from a technical point of view need to idealize, from the business sector need to be straightforward, is not only a higher and faster and stronger search for fake drugs, technology can not be effectively realizable? Hulu has made me relieved.

There are many other interesting things. For example, 1 they use Ajax for the recommended module, and the recommended data is dynamically loaded only when the user drags the page to the appropriate location, which allows for more accurate consideration of the effect of the recommended algorithm. 2 in the same functional area, they will balance the proportion of ads and recommended films, the logic behind it is that users see more interesting films-> stay longer-> see the overall number of ads increased, this can be converted into a reasonably rational data model, according to maximize the benefits of continuous optimization. 3 They set up predictive models for unregistered users by analyzing the historical data of registered users watching video. Look at this example, from the global statistics, to watch the film on the left of the users, 63% is a woman, 37% is a male, the usual practice, will give the film with a female-oriented advertising, then, the 37% of the men are a cup. And Hulu is about matching targeted ads, for registered users, Hulu knows the sex data; For unregistered users who do not know the sex, Hulu predicts the sex by analyzing its browsing records, such as the one on the right, which is likely to be female. These mathematical models are not difficult, but they need some skill to get them in the right place.

Two days ago I sent a microblog, "the understanding of the data and the problem modeling is the first, with how advanced algorithm is not the focus." This is my personal feeling, I believe most of the scenes should also be suitable. But the recommended in-depth, is definitely a test of algorithmic strength. With a reliable team, with less effort!

Evaluating a team, I personally have a less mature standard of judgment, is to see if there are no dropouts, this at least can explain a question: whether the things are sexy enough, people willing to put aside the shackles of full love. Coincidentally, the Hulu Beijing team has a PhD student who dropped out of Peking University. They are recruiting, interested can come here to see, or can directly contact: Hua @twitter.

There are some funny photos here, too.

This article can be reproduced arbitrarily, reproduced, please be sure to hyperlink form to indicate the original source of the article and copyright information.
Http://www.guwendong.com/post/2010/hulu-team.html