Every region wanted its own career site. The brand needed one CMS to power them all.
Permanente Medicine's regional medical groups each recruit physicians through their own career site. Northwest Permanente is the largest multispecialty group in Oregon and southwest Washington: 1,500+ physicians caring for 615,000+ Kaiser Permanente members.
The kickoff brief asked for consistency and shared best practices across regions, accessibility and security standards, shared navigation with Permanente.org, and better search results. The regions also needed to publish news, events and jobs on their own, without waiting on developers.
From regional websites to one recruiting platform:
Building one site at a time does not insure brand consistency. So we built the system that builds the sites. With the Tempo UX design team, I helped design the UX and UI of Tempo, a multi-site CMS that now powers the Northwest, Northern California, Washington and Georgia career sites. I also used this CMS to build out these regional sites.
Fewer than 1 in 5 sessions started on the homepage. Most physicians landed straight on a job listing or news story from Google, so each one had to do the homepage's job: say who Northwest Permanente is, answer the physician's questions and point to a next step.
The job search turns each specialty filter into a page Google can find, and each listing into a path to a recruiter. From June to September 2026, interactions rose 38% on 16% more visitors, and recruiter contacts rose 15%.
One system, many sites
Each regional medical group is a site in Tempo, with role-based access (Designer, Publisher, Admin) and the same toolset: pages, media, taxonomy, blogs, contacts, events and forms. Editors build directly on the page, with publishing status, preview and undo in one bar. A new region starts from a working system, with components and patterns vetted by our UX, UI and development work.
Publishing with guardrails
Tempo split publishing by risk. Regional teams and designers could both publish jobs, blog posts and events, inside components that kept the words, images and people theirs and the layout fixed. Every other page was built and published by the design team, which on Northwest meant me. That kept UX best practices in place on every page, not just the ones someone remembered to review.
Forms without developers
A new form used to mean a developer ticket and a wait for the next sprint. In Tempo, the design team builds forms from standardized components and patterns, so regions get a working form when they need one, and every form behaves the same way on every site. Even multi-step flows, like Refer a Friend's eligibility check, ship without new code.
A consistent architecture for every region
I helped establish the information architecture every regional site shares. Put any two main navs side by side and the skeleton matches: About us, Excellence, Careers and National presence. Careers always opens with Career search, Contact a recruiter and Benefits. National presence always leads back to Permanente.org, which the kickoff brief asked for. Regions tune the details: Northern California, the largest group, moves Careers to the front and adds Locations to the top level. A physician who knows one Permanente site already knows the next.
Saying yes, and saying no
Good design means saying yes to the user and, sometimes, no to the client. This site needed both. Physicians land from Google deep in the site and want to know in seconds whether there’s a job for them. Saying yes to them meant turning every filter and listing into a page that stands on its own.
After launch, a regional client asked for filter pills above the job list. Saying no, with mockups and evidence, kept the job list clean and saved an estimated two weeks of development.
Two clicks from Google to a recruiter
How does a physician who lands from Google find the right opening, or a recruiter, in one or two clicks?
The job search had to serve candidates who arrive mid-site, and it had to work for every region on the platform. Five big design decisions shaped the experience for candidates using these job search pages.
Every listing is a landing page
Physicians search for a role and a place, like "anesthesiologist jobs Portland." Each listing's address and page title use those same words, so Google shows a match: "Anesthesiologist in Portland Metro Area, OR." Breadcrumbs then lead up to related openings.
Filters that generate searchable pages
Picking a specialty rewrites the address (/jobs/type/anesthesia), the headline ("Anesthesia jobs") and the browser title. Every specialty becomes its own page Google can find. And once physicians land, the job pages keep them moving: the main jobs page and the clinical and administrative job pages rank among the site's top internal referrers, sending visitors deeper into the site.
Answering the real questions on the page
Salary, total rewards and hiring incentives sit up front, with "Apply now" at the top and bottom. But physicians also want to know what the work is like, and nobody answers that like another physician. So listings carry videos of Northwest physicians talking about their own work. Visitors starting an application rose 10%.
One component, configured per region
Every region runs the same job search feature and switches on only the filters it needs, with no new build. Northwest keeps it lean. Northern California, with about 550 openings, adds subspecialty, job number and second language. We built one flexible system, tuned to how each region recruits.
Always a way to a recruiter
Not every physician finds the right opening on the first visit. "Connect with a recruiter" sits at the top, a "Don't see your desired position?" block follows the results, and nurse practitioners, physician assistants and current employees each get their own route. Visitors contacting a recruiter rose 15%, and their interactions rose 20%.
How the decisions paid off
Organic search did the heavy lifting. Weekly organic search sessions grew 4 to 5 times between late June and September 2026. Over the same period, unique visitors rose 16% to 40,947 and total interactions rose 38%.
Then regional clients asked for a new feature, and it raised a question worth asking every time: just because we can, should we?
The feature I proved we didn't need
Read the ticket, dig for the logic
After launch, a project manager opened a ticket: "Add filter pills above job list." A couple of regions had asked for it, and the ticket rolled it out to every regional job search and the national one, on desktop and mobile.
The ticket's own context line named the real problem: "an easier way for users to remove all selected filter options within a single facet." Pills can clear filters. But were they the best way to solve that problem in this product?
Mock it up, then push back
I mocked up the pills on desktop and mobile before arguing against them. The problems showed up fast:
Every filter would appear twice, as a checkbox and as a pill.
Two controls for one state add cognitive load and confusion about which one does what.
Pills multiply with every selection and crowd the job list.
On mobile, five filters stacked into five full-width pills and pushed the first job to the bottom of the screen.
I also knew the pattern. The ticket cited using the pills from Kaiser Permanente's Doctor Search redesign, which I helped design. Knowing the right context and use case for this pattern informed my judgement about their complications here.
Put the control where users already look
I recommended a single "Clear all" control at the top of the Filter panel. On mobile, the Filter button carries a count so candidates see what's applied without a single pill. One control meant less for candidates to process and far less for developers to build.
Before going back to the client, I walked the UX design team, research and the product manager through the recommendation and the principle behind it: "Our first design task is to validate the requirements by understanding the problem, and not simply executing on the requirements we are given." Design, product and development returned to the client with one position.
Let the simplest answer win
The pills were dropped, and "Clear all" shipped. Candidates got one straight-forward way to reset their search and a job list that stays readable as they filter, on every screen size. Fewer controls meant fewer decisions between a physician and the right opening. The team also got back an estimated two weeks of development across every regional job search.
Outcome and reflection
The actions that matter grew faster than traffic. All figures compare June to September 2026 with the 90 days before. Physicians don't impulse-apply. A few hundred serious candidates reaching a recruiter or an application in one quarter is exactly what a healthy physician funnel looks like.
