Creation of a personalised article recommendation service

A free service that makes suggestions of the most relevant primary research papers based on the users history, chosen irrespective of the publisher.

Role

Senior UX Designer

Company

Springer Nature

Duration

8 months

a cellphone leaning against a wall
a cellphone leaning against a wall
a cellphone leaning against a wall

Company background

Springer Nature is a global academic and science publisher. Their main goal is to help Researchers and Academics learn, achieve and discover. It is home to some of the most widely used and trusted journals such as nature.com. 

The problem to solve

Springer Nature is exploring innovative ways to retain and engage their monthly active users. A survey was carried out to understand the challenges users face as researchers.

The volume of research being published has increased, statistics show that the number of articles published globally has grown from 1.3 million to 2.4 million from 2010 to 2015 with 4000 articles published each day.

Researchers are struggling to find the most relevant papers.

Company background

Springer Nature is a global academic and science publisher. Their main goal is to help Researchers and Academics learn, achieve and discover. It is home to some of the most widely used and trusted journals such as nature.com. 

The problem to solve

Springer Nature is exploring innovative ways to retain and engage their monthly active users. A survey was carried out to understand the challenges users face as researchers.

The volume of research being published has increased, statistics show that the number of articles published globally has grown from 1.3 million to 2.4 million from 2010 to 2015 with 4000 articles published each day.

Researchers are struggling to find the most relevant papers.

My role and responsibility

I worked as the Senior UX Designer within a multidisciplinary team, including Product Manager, Data Scientist, UX Researcher and Software Engineer. Our focus was less on roles and titles and more on our shared dedication to driving the product’s success. We followed a truly agile approach, continuously experimenting and making decisions based on hypothesis-driven insights.

Discovery of the solution

The proposition was to create a free service that makes suggestions of the most relevant primary research papers based on the users history, chosen irrespective of the publisher. 

Powered by an adaptive algorithm the service analysed the last 100 papers you’ve read across Springer Nature products. It then looks for similar papers from a large pool of journals, combines this with other data from other sources to create a score. The service can then use this score to deliver papers that meet a quality threshold from all publishers.

Discovery of the solution

The proposition was to create a free service that makes suggestions of the most relevant primary research papers based on the users history, chosen irrespective of the publisher. 

Powered by an adaptive algorithm the service analysed the last 100 papers you’ve read across Springer Nature products. It then looks for similar papers from a large pool of journals, combines this with other data from other sources to create a score. The service can then use this score to deliver papers that meet a quality threshold from all publishers.

Early concept

The original concept proposed a personalised dashboard requiring the user 
to sign-in to access recommendations. However, sign-in is confusing, as most users access content via their institution’s IP, and subscription models vary.

Proving the value

The key objective was to build a simple, standalone proof of concept with no dependencies to quickly test and learn the value of the service. Nature.com is an ideal experiment space, with over 8 million monthly active users, a trusted reputation, and high-traffic articles- maximising visibility and credibility for testing.

Mapping the user journey

I visualised the user journey to understand how users discover, access and interact with content. This revealed key pain points and decision moments. I learned that Researchers typically start on google scholar, find an article, read the abstract and then decide whether to continue reading.

Our hypothesis was presenting 3 recommended articles to 300 users after 
they read the article abstract will lead to clicks on at least one article or sign ups 
for more content.

Design explorations

With all this in mind, I began experimenting with how to present the recommendations. I explored different formats and placements with consideration for the other modules that already exist on nature.com. I also needed to be mindful of the design being mistaken for an advert. I ultimately decided on a card that slides in from the right after the user has finished reading the abstract.

Visual and animation design considerations

With such a small space, I needed to identify the key data points a user would need to successfully assess the relevancy of an article. In addition to this, I also needed to ensure the messaging encouraged users to sign up to the email. I chose to keep it on brand with nature.com styling. The use of animation plays an important part of guiding the user through the recommendations and once the key action of submitting email credentials had been completed.

Concept and usability testing

From the very start of the product, we knew it was essential to engage continually with our target users. This ensured that what we were building addressed real needs and integrated seamlessly into their workflows. We made contact with the Francis Crick Institute in London, which allowed us to quickly validate the concept of surfacing recommended content and test the usability of the feature.

Initial feedback revealed that users assumed the recommendations were exclusively from Nature. The use of Nature’s iconic branding and visual design made it difficult for them to associate the recommendations with other reputable publishers.

Iterating based on user feedback

Using this insight, I explored an identity that reflected Springer Nature’s values while signalling a publisher-agnostic service. This was not a nature-branded feature but a capability powered by Springer Nature. The identity needed to work at a small scale and be flexible enough to apply across multiple forms of communication.

Final visual identity

Following multiple iterations and conversations with stakeholders across the business, we collectively decided that a speech bubble best conveyed a sense of being personal, simple and helpful – a visual way of saying “We want to help you with your research.” The chosen typeface was both accessible and scalable, while the colour palette aligned with Springer Natures brand. The addition of yellow ensured the design stood out, drawing attention without feeling intrusive. I also considered how the pop up appears from the ‘R’, experimenting with different animations to fit with the new icon.  

Additional channels

The email plays an important part of the recommendations service. Current analytics show that although Researchers use desktop for research, they use mobile to browse content during their commute and when travelling. The email was optimised for mobile and provided an opportunity to aggregate more content.

Another vital part of the process was to create a landing page to be explicit about 
how we create the recommendations. The page also supported marketing campaigns to promote the service and acquire new customers.

Personalised data dashboard

At this point in the process we had started to understand that the users were engaging with the service. I collaborated with the Data Scientist to help improve our product team dashboard for ‘Recommended’ to view our key metrics. We already had a dashboard to monitor the click through rate, email subscribers and close rate but wanted one that was on brand and highlighted different experiments. 

Design, test and iterate

We continued to make incremental iterations in response to A/B testing. This helped us to track the areas that need to be improved. It also made it easier to run different types of experiments such as changes in the UI. The idea was to make small incremental changes. This was a continuous process of designing, testing and then iterating.

Defining what comes next

Springer Nature wanted to broaden the reach and start to integrate the service into other products, requiring me to design an iInline application. 

We continued to gather user feedback on the service, which led us to explore new concepts like the ‘The Digital Research Assistant’– a plug-in that tracks your scientific activity. Users can review their activity history, make adjustments such as removing papers and these changes feed directly back into the recommendation algorithm to deliver more relevant suggestions. 

Outcome and impact

We embodied a truly holistic approach to product development. Bringing together user experience, visual design, software development and algorithm design, with everyone on the team contributing at every stage.

Other projects

Copyright 2025 by Laura Gosney

Copyright 2025 by Laura Gosney

Copyright 2025 by Laura Gosney