skip to main content


Title: Optimizing Content with A/B Headline Testing: Changing Newsroom Practices
Audience analytics are an increasingly essential part of the modern newsroom as publishers seek to maximize the reach and commercial potential of their content. On top of a wealth of audience data collected, algorithmic approaches can then be applied with an eye towards predicting and optimizing the performance of content based on historical patterns. This work focuses specifically on content optimization practices surrounding the use of A/B headline testing in newsrooms. Using such approaches, digital newsrooms might audience-test as many as a dozen headlines per article, collecting data that allows an optimization algorithm to converge on the headline that is best with respect to some metric, such as the click-through rate. This article presents the results of an interview study which illuminate the ways in which A/B testing algorithms are changing workflow and headline writing practices, as well as the social dynamics shaping this process and its implementation within US newsrooms.  more » « less
Award ID(s):
1717330
NSF-PAR ID:
10096346
Author(s) / Creator(s):
;
Date Published:
Journal Name:
Media and Communication
Volume:
7
Issue:
1
ISSN:
2183-2439
Page Range / eLocation ID:
117; 127
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. Many journalists and newsrooms now incorporate audience contributions in their sourcing practices by leveraging user-generated content (UGC). However, their sourcing needs and practices as they seek information from UGCs are still not deeply understood by researchers or well-supported in tools. This paper first reports the results of a qualitative interview study with nine professional journalists about their UGC sourcing practices, detailing what journalists typically look for in UGCs and elaborating on two UGC sourcing approaches: deep reporting and wide reporting. These findings then inform a human-centered design approach to prototype a UGC sourcing tool for journalists, which enables journalists to interactively filter and rank UGCs based on users’ example content. We evaluate the prototype with nine professional journalists who source UGCs in their daily routines to understand how UGC sourcing practices are enabled and transformed, while also uncovering opportunities for future research and design to support journalistic sourcing practices and sensemaking processes. 
    more » « less
  2. null (Ed.)
    Headlines play an important role in both news audiences' attention decisions online and in news organizations’ efforts to attract that attention. A large body of research focuses on developing generally applicable heuristics for more effective headline writing. In this work, we measure the importance of a number of theoretically motivated textual features to headline performance. Using a corpus of hundreds of thousands of headline A/B tests run by hundreds of news publishers, we develop and evaluate a machine-learned model to predict headline testing outcomes. We find that the model exhibits modest performance above baseline and further estimate an empirical upper bound for such content-based prediction in this domain, indicating an important role for non-content-based factors in test outcomes. Together, these results suggest that any particular headline writing approach has only a marginal impact, and that understanding reader behavior and headline context are key to predicting news attention decisions. 
    more » « less
  3. This work proposes Dynamic Linear Epsilon-Greedy, a novel contextual multi-armed bandit algorithm that can adaptively assign personalized content to users while enabling unbiased statistical analysis. Traditional A/B testing and reinforcement learning approaches have trade-offs between empirical investigation and maximal impact on users. Our algorithm seeks to balance these objectives, allowing platforms to personalize content effectively while still gathering valuable data. Dynamic Linear Epsilon-Greedy was evaluated via simulation and an empirical study in the ASSISTments online learning platform. In simulation, Dynamic Linear Epsilon-Greedy performed comparably to existing algorithms and in ASSISTments, slightly increased students’ learning compared to A/B testing. Data collected from its recommendations allowed for the identification of qualitative interactions, which showed high and low knowledge students benefited from different content. Dynamic Linear Epsilon-Greedy holds promise as a method to balance personalization with unbiased statistical analysis. All the data collected during the simulation and empirical study are publicly available at https://osf.io/zuwf7/. 
    more » « less
  4. This work proposes Dynamic Linear Epsilon-Greedy, a novel contextual multi-armed bandit algorithm that can adaptively assign personalized content to users while enabling unbiased statistical analysis. Traditional A/B testing and reinforcement learning approaches have trade-offs between empirical investigation and maximal impact on users. Our algorithm seeks to balance these objectives, allowing platforms to personalize content effectively while still gathering valuable data. Dynamic Linear Epsilon-Greedy was evaluated via simulation and an empirical study in the ASSISTments online learning platform. In simulation, Dynamic Linear Epsilon-Greedy performed comparably to existing algorithms and in ASSISTments, slightly increased students’ learning compared to A/B testing. Data collected from its recommendations allowed for the identification of qualitative interactions, which showed high and low knowledge students benefited from different content. Dynamic Linear Epsilon-Greedy holds promise as a method to balance personalization with unbiased statistical analysis. All the data collected during the simulation and empirical study are publicly available at https://osf.io/zuwf7/. 
    more » « less
  5. This work proposes Dynamic Linear Epsilon-Greedy, a novel contextual multi-armed bandit algorithm that can adaptively assign personalized content to users while enabling unbiased statistical analysis. Traditional A/B testing and reinforcement learning approaches have trade-offs between empirical investigation and maximal impact on users. Our algorithm seeks to balance these objectives, allowing platforms to personalize content effectively while still gathering valuable data. Dynamic Linear Epsilon-Greedy was evaluated via simulation and an empirical study in the ASSISTments online learning platform. In simulation, Dynamic Linear Epsilon-Greedy performed comparably to existing algorithms and in ASSISTments, slightly increased students’ learning compared to A/B testing. Data collected from its recommendations allowed for the identification of qualitative interactions, which showed high and low knowledge students benefited from different content. Dynamic Linear Epsilon-Greedy holds promise as a method to balance personalization with unbiased statistical analysis. All the data collected during the simulation and empirical study are publicly available at https://osf.io/zuwf7/. 
    more » « less