Abstract
Search engines have introduced health information boxes to provide expert-vetted summaries for common medical symptoms. This study analyzes 28,552 unique Bing searches to understand how users interact with these boxes compared to standard results and advertisements. Findings indicate that information boxes capture the highest share of user attention (25%), and their design—specifically reading ease—can influence search behavior, including reducing clicks on commercial advertisements.
Introduction
When individuals experience new health symptoms, search engines are often their first point of contact. However, the vast amount of online information can lead to "cyberchondria" or the consumption of misinformation. To address this, search engines like Bing and Google have implemented health information boxes—curated summaries that appear alongside search results. While these features are common, empirical data on real-world user interaction has been limited. This research investigates how these "information anchors" influence user attention and click behavior.
Methodology
The researchers analyzed archival, deidentified data from Microsoft Bing searches in the United States between September and November 2019. The dataset comprised 28,552 unique searches for the 17 most common medical symptoms, such as anxiety, cough, and headache.
To measure user attention at scale, the study utilized mouse pointer hover time as a proxy for eye-tracking—a validated metric in information retrieval research. The researchers manually coded the characteristics of the information boxes, assessing "reading ease" via Flesch scores and identifying features like "related conditions." Linear and logistic regressions were used to analyze the time spent on various page elements and the likelihood of clicks.
Key Findings
The study revealed that the information box is the primary driver of attention on the search results page. When all elements were present, users spent 25% of their time (7.1 seconds) on the information box, significantly more than on standard web results (23%), advertisements (20%), or itemized results (10%).
- Design Influences Engagement: Higher reading ease and the inclusion of "related conditions" were associated with longer hover times.
- Impact on Ad Clicks: Notably, high-quality information boxes were negatively correlated with clicks on advertisements. This suggests that providing immediate, accessible answers can steer users away from commercialized health content, potentially serving as a public health benefit.
- Itemized Results Gap: Users spent the least amount of time (10%) on itemized web results, preferring either the high-level curated summary or the full context of a standard link.
Conclusion
Information boxes successfully compete for user attention and serve as a critical tool for providing vetted health information. By optimizing the readability and content of these boxes, search engines can better support users in their health-seeking journeys. Future research should investigate whether these interactions lead to improved real-world medical outcomes and how behaviors might differ across various search platforms.
Summary updated