Bojhan Somers - product strategy, research and design


I helped Elsevier with creating a program that continuously evaluates new product experiments, to tackle challenging information quests that scientists face.


Scientists are time constrained, deal with more and more content each year and have to outperform their peers. Elsevier wanted to evaluate new concepts.

  1. Product ideas which require real data to be validated.
  2. Heavy upfront investment, before ideas are validated.
  3. Difficultly setting priorities amongst competing user insights.
  4. Lack of continuous user voice in the process.


Implement a process that enables product teams to quickly gather in-depth user feedback, qualitatively and quantitative to create products that align with user needs.

  1. Creation of a continuous evaluation process.
  2. Providing in-depth strategic feedback.
  3. Using insights to build business cases for new ideas.
  4. Integrating user research into agile practices.

Creating a continuous user feedback loop

We were able to evaluate dozens of products experiments in hundreds of user interviews with researchers from all over the world.

Using a Lean UX approach to validate and iterate concepts with increasing fidelity. Various concepts proofed valuable enough to be launched as MVP to scientists.

Creating the fundaments for the development of high value features on Mendeley, a social platform that helps scientists showcase their work, find information articles and collaborate in groups.

Providing in-depth strategic feedback

We created a measurement framework to quantify the value proposition of various experiments and their potential contribution to the overall product adoption and retention.

Using this value measurement framework we were able to help product leadership understand and prioritise features.

Recommending articles to scientists

Mendeley provides scientists with recommendations on articles, to help them discover new research based on their interests.

These recommendations are based on what these users have added to their personal library, trending in their domain, recent activity and reading behaviour. By bringing together millions of articles and behaviours across the Elsevier ecosystem.

The recommendation engine was build by the data science team at Mendeley and was able to provide high quality recommendations.

The initial phase we explored what it means for users for a recommendation to be relevant, uncovering many biases and attitudes that inform how you present and pick article recommendations.

We primarily focused on understanding: