User Research
More than just making assumptions, we looked at data and interviewed users.
We selected a variety of users to interview based on messaging behaviours, like frequency and origin of message - Jobs page v Messages page.
There were two major workflows, reviewing candidates and messaging candidates already interviewing.
Users were more likely to message from the jobs page because they preferred to review candidates there
Reviewing candidates from the Messaging page was unclear or completely missed.
Finding candidates they previously messaged was difficult.
Hierarchy wireframes
Based on the research, I created a few of lofi wireframes to communicate hierarchy of filters to the team.
This is defined by grouping messages by jobs, application stage or other relevant data.
Users indicated they…
Reviewed candidates based on a job that needs to be filled.
Are not power users of our tool, they want to be in and out.
Want to review candidates and message them right away.
With that information we chose layout 1. It prioritizes jobs as the top level navigational tool, which aligns with recruiters' mental model, and does not rely on complex filters.
Iterations
We tried many different iterations of how we could structure the solution, exploring shallower elements like colour and spacing and deeper elements like navigation and scalability
User Testing Prototype
A good user test is all about giving the user prompts in the scenarios they might have while using the product, never direction them to "Click here" or "Look there."
It's best to save direct questions for the end of the test to avoid bias.
Some of the prompts...
You're logged into FactoryFix in the morning, it's time to review all your CNC Machinist candidates.
Let's check on the new messages and review the new applicants.
Michelle looks like a good fit so let's move her straight to the review phase.
You notice she also has some experience in managing processes, so you want to consider her for the Floor Supervisor role.
Let’s go to a new job. You want to review the job that Josh is monitoring.
You only want to see the candidates in the interview stage.
User testing prototype
Made with Figma
Key Takeaways
The tabs were a big win - Users loved having more space to see the content. They didn't express the need to view both messages and content at the same time.
Users don't need to view multiple jobs at once - Filtering by multiple jobs or a single job was a point of contention on the team. Interactions with the multiple jobs selections were more complicated and most importantly didn't fit the user's workflow of reviewing candidates job-by-job.
Filters needed to be surfaced better - We tested filters hidden behind a button that would open a small dropdown. Users would find it only after some directing as they weren't used to our product having filters.
Final product prototype
Made with Figma Make
Taking it all in
After user testing, it's time to make decisions around what to prioritize in our first cycle. We're not quite ready to deliver every desired feature.
Decisions made
Surfacing jobs filters - We used the space provided by removing the Candidate Information Panel to better surface the jobs filter. This allows users to easily filter jobs by who owns them, where they are (for multi-location users), and whether they're live or archived.
Surfacing sort and filtering for candidates - Users can now sort candidates and filter candidates by job.
Tabs to view candidate information - Instead of jumping back and forth between different pages (jobs page and messages page), rely on a constrained drawer or use pop-ups to view resumes, users can now select the right tab for candidate to display the desired information.
Scalable - This solution provides the space to scale, as we tested it internally with mock-ups of soon to release features - including Agentic AI.
Results
Users moved their workflow from Jobs page to Conversations page - Fairly quickly after release, users migrated their workflow to the messages page.
Candidates were easier to find - We received less support requests for lost candidates and conversations.
Foundations for Agentic AI - With users no working mainly in the conversations page, we could start building our Agentic AI on top of it.
One more thing...
The tabbed view created a challenge.
How do we let users both message and review a candidate at the same time?
After moving the candidate information from the right side drawer to a tab, users now must jump back and forth losing context if they had a question about the candidates resume or application.
Enter the always on messaging field!
No matter what tab the user switches to, the message box will follow them around. Now the user can review the candidate's application, resume, other applied jobs and more without losing the ability to message them.
More thinking
Candidate don't respond right away.
The context of the conversation is not needed to ask questions about a candidates application.
I explored typical solutions a message pop out and messaging drawer, both defeated our goal of scalable space for content.
The final question that solidified the decision was - do users need to see the entire conversation to send a message about the candidate's application? Absolutely not!