{{ projectTitle }}

Problem

Approach

Balancing enterprise vs. small business users

Enterprise recruiters typically had more complex workflows and larger teams to collaborate with, while small business recruiters were the opposite in many ways. Our principle was to help enterprise users do more without getting in the way of small business users. This meant focusing on the essentials while ensuring more advanced features were readily available.

Managing and prioritizing messages

In tackling this problem, I initially started with concepts intended to make managing messages more "intelligent," but through our user testing, we narrowed down to a standard folder architecture for the MVP.

It turned out participants loved the familiarity and usefulness of folders, especially since that's already how they managed their email. This was a recurring theme for us: is this more of an email "thing" or a chat "thing"? We'd decidedly leaned into the former, and that suited most of our participants just fine.

One early direction explored surfacing folders as tabs across the top of the inbox, alongside a persistent left panel, and where profile details would go if they moved out of the message thread into a slide-in panel.

I ultimately placed these folders into a persistent left panel, making them more discoverable, while moving the right panel's contents to a secondary level in the slide-in profile. From there, I iterated on the panel itself across a few rounds before landing on a condensed folder list with a secondary status filter.

In the final design, I polished down the folder list to the most essential ones (Inbox, Awaiting Reply, Archived) and pushed others to a secondary-level filter (Accepted, Declined, and others). Conceptually, this filter also made sense because "Accepted" and "Declined" were message types that could exist across any folder, not just the inbox.

Finding past messages and files

According to our usage data, search was one of the more frequently used features. Users would want to find a past conversation but forget its exact name, most often searching by a role in the subject line, like "product designers." Other times, they were looking for a resume attachment.

This led me to two directions: a typeahead tuned to role-based subject lines, and a more advanced, filtered search for attachments.

Testing didn't give us a strong enough signal to call either a priority. Users were largely fine with search as it stood, and building either well would have taken a lot of time before the initial release.

We shipped simple performance improvements to the existing search and left the bigger ideas on the roadmap.

Refreshing your memory while replying

We needed to surface just enough information for a recruiter to take action, without pushing the message content too far down the page.

I explored modifying the profile card to elevate the essentials, pushing everything else into the slide-in panel or an overflow menu.

I'd originally favored keeping the full right panel in place, but internal feedback and user testing changed my view. It broke from how we represented profiles elsewhere in the app, an inconsistency that had already confused users in other feature areas.

Testing surfaced a real cost, too: several existing users felt that moving the profile card disrupted how they were used to scanning for information.

Following up and taking action

I was interested in "suggested actions," where we might suggest or even automatically trigger follow-up actions based on a message's contents.

It turned out it's not so easy to "AI magic" your way to relevant suggestions, and ensuring trust and privacy would be a major requirement of a feature like this.

Instead, I focused on cleaning up and reorganizing our existing Recruiting Tools to be discoverable for power users while getting out of the way of the main message content.

Before, adding a note meant a click-through flow with the field tucked away. After, quick actions sit right alongside the message thread, and the redesigned inbox carries that same clarity throughout.

Solution

Onboarding

Onboarding had never been a major part of the old inbox, so we had an opportunity to improve things this time around. As I refined the design, I explored onboarding by asking: what are the essential tasks and workflows for new users? What are the more advanced features we can introduce over time? What about different types of users? And how can we draw attention and guide them without disrupting their workflow?

Accessibility

Inclusive design had become a key focus area for LinkedIn's design team, and I made sure to include accessibility specs in my designs. One challenge was that, across all our apps, we over-relied on tabbing to select elements, so keyboard users would get "tab fatigue."

One change I proposed was to start introducing arrow keys into our specs, so users could quickly tab between UI areas while using arrows to navigate within those areas.

Transitioning design systems

If you're familiar with the LinkedIn app, you might've noticed a new design language being phased in. As an enterprise team, we'd started exploring how it'd translate to our UIs, but the transition would take many months, if not years.

The upshot was that I had to stick to a limited set of existing components to ensure we could make the transition, which was a bit painful at times, but also promised to pay off in the long term with a smoother transition down the line.

Outcomes

Reflections

An email paradigm reaching its limit

By the time I left, there was a lot of talk about revamping the Recruiter homepage for a future where AI powered much of the workflow and cut down on tedious, manual work. I immediately thought of the inbox: recruiters do want a personal touch in their outreach, but so much of the work is tedious and repetitive.

Much of the UI I built was about surfacing information, grabbing attention, and making sure nothing fell through the cracks, but it was fundamentally derived from an email paradigm: cranking through messages one by one. I think there was a real opportunity to rethink that model from the ground up, but I left before we started digging into it seriously.