Today, a culture of experimentation pervades the industry as publishers work to balance the tradeoff between maximizing revenue per user and maximizing the number of users in the long term. In a new paper published today, researchers at HP Labs call this “The Provider’s Dilemma.”
To add some predictability to the experimentation process, coauthors Christina Aperjis and HP Senior Fellow Bernardo Huberman propose a new algorithm that may help predict how online audiences will react over time to the introduction of different business models, like advertising or subscription fees.
“The problem is that while free content will result in maximum audience attention, it creates minimal revenue,” says Huberman. “Added to this is the reality that people have come to expect ‘free’ content on the Web, which gives publishers even fewer options to recover costs.”
Although A/B testing and other techniques can deliver some insight into the issue, constantly imposing new business models might wear the audience out and cause them to leave for good.
Aperjis and Huberman’s solution is based on adaptation theory. Applied here, the theory posits that even though a change affects a person’s use of the product initially, as time goes on people tend to adapt and become less aware of these changes.
Here’s how the algorithm might work for an online publisher.
First, the publisher publishes a change to her site (e.g. a new type of advertisement) for a limited amount of time. Then, the algorithm analyzes the traffic data before and after the change and presents the publisher with one of two predictions:
1. Users don’t care about the advertisement and the publisher should go ahead make the change to the site all at once.
2. Users are sensitive to the advertisement, and changes should be made over time (at a rate that is also specified by the algorithm) to minimize attrition and increase revenue.
At this stage, the results of this work are still mathematical in nature and a real world implementation is needed to see how it can help content providers. We’ll continue to update readers on the status of this research as it evolves.
To read the full research report, click here.
Why is HP conducting this research?
Increasingly, online media is becoming many people's principal interface to technology, and these media interactions produce an enormous amount of data. Creating software, hardware, and services that can automatically analyze large data sets and help people make informed decisions is an extremely challenging technical task and an area of focus at HP Labs.
HP believes that information is the greatest resource we have for addressing problems in business and society. HP is the world's largest technology company and HP Labs is its advanced research group.
