Enhancing user comprehension

Helping hirers better understand the candidates they are being recommended, when searching for new talent.

Company: SEEK

Role: Senior Product Designer

Duration: 3 months

Ideation workshops, Quantitative, Qualitative, Concept Testing, MVP, Ideal, Playroom Prototyping, Data Driven

Context

SEEK operates as an online job marketplace covering the entire Asia Pacific region.

Within its offerings, hirers can leverage the Talent Search feature to search, filter, compile, and reach out to potential new talent - a tool that facilitates headhunting and proactive sourcing.

The challenge

Within Talent Search, a sophisticated algorithm operates behind the scenes. It aims to optimise results based on factors like the posted job ad, and the skills and experience of the candidate.

The team received substantial feedback on these optimised results, particularly regarding hirers' occasional lack of understanding regarding recommended candidates.

After reviewing past research and business priorities aimed at enhancing connections between hirers and candidates, we identified a significant opportunity: improving "explainability."

Our goal was to find ways to better clarify our results and assist hirers in understanding why specific recommendations were made.

The highlight of the project:

Seek utilises a tool named Playroom to develop interactive prototypes using real code.

The highlight for me was acquiring proficiency in this tool and coding over 1300 lines to visualise both our MVP and Ideal state designs, across various themes and screen sizes.

Take a look at how Playroom lets me view designs in different visual languages and for for different devices!

The solution

I redesigned:

The way candidate information is shown on cards through card sorting and testing the most important information for hirers
The search results page, including a section about the job in order to make it clear how and why candidates are being recommended
Improved error states and messages for times when accurate recommendations cannot be provided.

When we tested these changes, all of our 45 participants (across moderated and unmoderated testing) easily understood the recommendations that were provided and knew what their next steps would be (either to connect with a candidate or improve their recommended results by editing their ad).

A look at some of the work

For this project, I followed the design process (with some adjustments due to time and resource constraints).

Take a look at some images of the research, user flows and the final designs.

To learn more about my process and learnings from this project, reach out!