Thumbnails for Data Stories: A Survey of Current Practices

When people browse online news, small thumbnail images accompanying links to articles attract their attention and help them to decide which articles to read. As an increasing proportion of online news can be construed as data journalism, we have witnessed a corresponding increase in the incorporation of visualization in article thumbnails. However, there is little research to support alternative design choices for visualization thumbnails, which include resizing, cropping, simplifying, and embellishing charts appearing within the body of the associated article. We therefore sought to better understand these design choices and determine what makes a visualization thumbnail inviting and interpretable. This paper presents our findings from a survey of visualization thumbnails collected online and from conversations with data journalists and news graphics designers. Our study reveals that there exists an uncharted design space, one that is in need of further empirical study. Our work can thus be seen as a first step toward providing structured guidance on how to design thumbnails for data stories.

💡 Research Summary

The paper “Thumbnails for Data Stories: A Survey of Current Practices” investigates how visualizations are adapted for use as thumbnail images in data‑driven journalism. Recognizing that thumbnails are often the first—and sometimes only—visual cue that a reader encounters, the authors set out to map existing design practices, identify the dimensions of the design space, and uncover the goals and constraints faced by practitioners.

Data collection and coding
The authors harvested articles published between November 1 and December 31 2018 from seven prominent data‑journalism outlets (Pew Research Center, The Economist, The New York Times (Upshot and DealBook), FiveThirtyEight, The Wall Street Journal (Graphics), Bloomberg News, and First Tuesday Journal). They focused on political and economic stories, which tend to rely heavily on charts. From an initial pool of 139 articles, 91 contained visualizations in the article body. After filtering out articles whose thumbnails used photographs (48) and those employing rarely used chart types (24), they arrived at a set of 67 visualization thumbnails for detailed analysis.

Each thumbnail was examined for two dimensions: (1) the editing strategy applied to the original chart—categorized as modified, cropped, or resized; and (2) the status of individual chart components—whether a component was removed, added, or remained. To capture component status, the authors built on three prior taxonomies (Borkin et al.’s visualization components, Byrne et al.’s graphical vs. figurative distinction, and Ren et al.’s annotation categories) and synthesized a new classification comprising 14 basic components (x‑axis, y‑axis, axis labels, tick marks, data labels, chart title, explicit legend, implicit legend, etc.) and 4 added components (explanatory text, highlights, human‑recognizable objects (HROs), and graphics not related to data (GNRDs)). Inter‑coder agreement was high (96 % agreement, Fleiss’ κ = 0.75).

Findings from the thumbnail corpus

• Chart‑type trends: Line charts (30 of 67 thumbnails) most often omitted the x‑axis title, the y‑axis, and legends, while frequently adding highlights and explanatory text. Bar charts and scatter plots displayed a more heterogeneous mix of retained, removed, and added elements.
• Outlet‑specific strategies: FiveThirtyEight tended to strip almost all basic components from line‑chart thumbnails and to insert many HROs and GNRDs. Traditional print‑derived outlets (NYT, WSJ, The Economist) mostly performed simple cropping or resizing of the original chart without major component removal. 538 and Pew showed a preference for modifying charts (e.g., removing axes, recoloring), whereas NYT leaned toward cropping the chart region.
• Editing strategy distribution: Across the 67 basic‑axis charts, modification (28) and cropping (29) were the dominant strategies; resizing was less common. For an additional set of 24 less‑common visualizations (maps, pies, parallel coordinates, sankey diagrams, etc.), cropping was the most frequent approach.

Practitioner interviews
Six professionals (two journalist‑engineers, an interactive graphics developer, a senior news editor, a data scientist, and a computational journalist) were consulted via asynchronous email and Slack. Their responses revealed four overarching design goals:

1. Brand consistency – color palettes, typography, and the inclusion of HROs must align with the outlet’s visual identity.
2. Aesthetic appeal / click‑through – thumbnails must be eye‑catching to compete for limited attention on homepages, social feeds, and mobile apps.
3. Interpretability – the thumbnail should convey the core story or trend quickly, often through highlights or brief explanatory text.
4. Ethical clarity – avoid misleading representations; the thumbnail must not distort the underlying data or narrative.

Practitioners noted tensions among these goals. For instance, brand‑mandated colors can clash with data‑driven color encodings, and adding decorative HROs may obscure the chart’s message.

Implications and future work
The study highlights a substantial “uncharted design space” for visualization thumbnails, with little consensus on best practices. Crucially, the paper points out the lack of empirical evidence linking specific editing strategies to user outcomes such as click‑through rates, comprehension, or recall. The authors propose a research agenda that includes: (a) controlled user experiments (eye‑tracking, A/B testing) to quantify the impact of component removal/addition; (b) development of automated thumbnail generation pipelines that take the 18‑element taxonomy as input; and (c) longitudinal studies to assess how thumbnail design influences long‑term audience trust and misinformation.

In sum, this work provides the first systematic taxonomy and empirical snapshot of how data‑driven visualizations are repurposed as thumbnails, surfaces the divergent practices across media organizations, and sets a roadmap for rigorous, user‑centered investigations that can ultimately guide designers, journalists, and platform engineers toward more effective and responsible thumbnail designs.