UXR Case Study – Yahoo Research / Verizon Media

Trust in Algorithmic News Feeds

Longitudinal Mixed-Methods Research on User Trust, Algorithmic Curation, and Responsible Product Innovation

UX researchers testing devices at Yahoo

Design Framing

How might we understand what shapes user trust in algorithmically curated news feeds — and use that understanding to build more responsible, empirically grounded product experiences?

As a User Experience Research Contractor at Yahoo Research (Verizon Media), I led a longitudinal mixed-methods study on trust in algorithmic news feeds — one of the most consequential and contested design problems in consumer media. The work required building cumulative user intelligence over time, triangulating behavioral data, survey analytics, and qualitative interviews to generate a layered picture of how users form, lose, and negotiate trust with algorithmic systems at scale.

Yahoo Research — AI and machine learning at scale

Methods

  • Longitudinal Study Design — tracked user attitudes and behaviors across multiple waves to capture how trust evolved over time rather than as a static snapshot
  • Survey Analytics — designed and analyzed quantitative instruments measuring trust dimensions, perceived algorithmic fairness, and feed satisfaction
  • Behavioral Data Analysis — integrated platform behavioral signals (clicks, dwell time, return rates) to triangulate self-reported attitudes against observed patterns
  • Qualitative Interviews — conducted in-depth interviews to surface the mental models, heuristics, and emotional responses users brought to algorithmic curation
  • Data Triangulation & Synthesis — wove quantitative and qualitative streams into coherent, actionable narratives for senior product leadership

The longitudinal design was essential: trust in algorithmic systems is not a fixed user attribute but a dynamic relationship shaped by feed behavior, news events, and platform design choices. Building intelligence over time allowed us to surface patterns that single-point studies routinely miss.

Key Challenges

  • Trust in algorithmic feeds is multidimensional — users conflate trust in the algorithm, the platform, and the news sources it surfaces, making clean measurement difficult
  • Behavioral data and self-reported attitudes frequently diverged, requiring careful triangulation rather than simple aggregation
  • Complex, nuanced findings needed to be translated into compelling executive narratives without flattening the underlying user complexity
  • Responsible innovation recommendations had to be grounded in empirical evidence, not design intuition — raising the bar for how findings were framed and defended

What We Built Together

Through cumulative waves of mixed-methods research, we developed a layered model of how users negotiate trust with algorithmic news systems — identifying the specific feed behaviors, transparency signals, and source patterns that either built or eroded user confidence. Findings were translated into product narratives for senior leads, directly informing responsible innovation strategy.

  • Wave 1 Baseline → Behavioral Integration → Qualitative Deep-Dive → Cross-Method Triangulation → Product Narrative → Roadmap Activation
  • Attributes: Longitudinal, Mixed-Methods, Empirically Grounded
Yahoo brands and products

Real World Impacts

  • Cumulative user intelligence delivered to senior product leads translated directly into responsible innovation strategies for algorithmic feed design
  • Longitudinal triangulation methodology established a replicable model for ongoing trust monitoring at platform scale
  • Complex multi-method findings distilled into compelling executive narratives that activated product decisions grounded in real user evidence