At the time, Springer Nature had recently launched a beta service, Recommended, which used machine learning algorithms to suggest scientific research papers that users may have missed. The service was being trialled across four of the Nature portfolio of journals, and needed to go from beta to launch within six months.

What happened

Working with a small team of developers and designers, we successfully scaled the service to across 3000 journals, constantly refining our proposition through rigorous experiment-driven product development.

I instigated the hiring of Springer Nature’s first data scientist to assist us in bringing further discipline to our practice, and, inspired by our work, I wrote Experiment-Driven Product Development, to help share what we had learnt.

During my time on Recommended, myself and the team’s user experience designer set up a user panel at the nearby Crick Institute, which we would visit each month to get feedback and inspiration from working scientists.

It was on one of these visits that we uncovered an unmet need for a focused browser history experience, one that would help researchers keep a track of the literature they had read.

Through rapid development, we created the Digital Research Assistant, which was a viral hit. It provided more information for the main Recommended product to grow, as well as extending the user base, and showed the importance of listening to, and co-creating, products with real users.

Outcome

Over the lifetime of the service, Recommended maintained an average 5% click-through rate on its’ web recommendations, and 20% on email recommendations. Although the service was discontinued a couple of years after my time with the team, it was a pioneering service within Springer Nature for its’ data-informed, experiment-driven approach.

Employer:

Springer Nature

Project Duration:

September 2016 – December 2017

Themes:

Product development, experiment-driven, machine learning, publishing, user-centred design

Share this:

Like this: