Case Study · HR Tech · 0→1 · 2020–2021
JobPixel was building video-first hiring. I joined as the founding designer while the company was actively selling to its first customers. We were building and pitching simultaneously — showing prototypes to hiring managers in the morning, iterating on their feedback in the afternoon. My Jobs is where that pressure produced something real. This is the story of how we figured out what it needed to be.
The Problem
Traditional hiring tools are built around rows in a table — name, status, date applied. That model works when candidates are documents. It falls apart the moment candidates become faces and personalities on video.
When we joined JobPixel, there was no employer-side dashboard at all. Hiring managers were reviewing videos through ad-hoc links and tracking applicants in spreadsheets. The product needed to figure out how to manage people, not just route videos — and it needed to do it while we were simultaneously selling the platform to its first real customers.
The pressure that shaped everything
We were showing prototypes to hiring managers in the morning and iterating on their feedback in the afternoon. Every design decision had to work in a live demo before it could be validated in production. That constraint — build it and pitch it simultaneously — turned out to be the most useful forcing function we had.
Act 01 · The Wrong Bet
When the founders brought me on, the vision was clear: build a video-first job board. Think Indeed or LinkedIn, but instead of text listings and uploaded résumés, every job post would have a video from the hiring manager and every candidate would introduce themselves on camera.
It was a genuinely compelling idea. Video humanizes the hiring process on both sides. Job seekers could see who they'd be working for before applying. Employers could see personality and presence before investing time in interviews. The grid of faces felt alive in a way that a table of names never could.
So I designed it. A full discovery feed — 6-column grid, video cards, company branding, location search, filters. We took it to early customers. The prototype looked great in a demo. The feedback was polite and non-committal in a way that told us everything.
The customer signal that changed everything
The customers who stayed on the call longest weren't asking about job discovery. They were asking about what happened after someone applied — how do I manage them, how do I share them with my team, how do I know which ones to call first. We weren't listening to that soon enough. When we did, the job board direction collapsed immediately. That's what gave birth to My Jobs.
Try it: Hover any card, click to simulate a video intro, use search to filter. This is the product vision that didn't survive contact with customers. It looked great in a demo. The feedback told us to build something else.
Act 02 · The Pivot · My Jobs Is Born
Killing the job board direction wasn't a clean decision. We had customers who'd seen the demo and liked it well enough. We had a sales story built around it. Pivoting meant going back to those customers with a different product. Some of them came with us. That was the real brief — build something the ones who stayed would actually use.
The question wasn't where to post jobs — it was what happens after someone applies. For customer-facing roles especially, a résumé tells you almost nothing about what actually matters: how does this person present themselves? Are they clear and articulate? Do they seem genuinely excited about the opportunity? Those things don't fit in a bullet point.
Video changes that equation entirely. A 60-second introduction from a candidate reveals more about fit than three rounds of résumé screening. JobPixel's core value was inserting video into the hiring loop at exactly the right moment — after the application, before the phone screen — giving employers a human signal they'd never had before.
My Jobs was the product built around that insight. A dedicated dashboard where recruiters and hiring managers could review video applications across all their open roles, in one place. The first version used a collapsible accordion to organize candidates by job post. Clean on paper. Brutal in practice.
What customer feedback revealed
We were on calls with customers who'd posted four, six, eight jobs. They'd open My Jobs, click through each accordion row to check for new applicants, lose track of where they were, and close the tab. We saw this pattern across multiple accounts. The accordion wasn't just inconvenient — it was causing churn in a product we were still trying to prove.
Try it: Click any job row to collapse or expand. Open two at once and feel the scroll problem. The ··· menu on each job shows all available actions. Green Hills starts expanded — notice how it dominates the view and pushes the other three jobs into thin collapsed rows. This is exactly the feedback we received from customers.
Act 03 · The Persistent Grid
The third iteration was the most significant rethink. We eliminated the accordion entirely. Jobs became persistent sections — always visible, always showing their candidates. No collapsing. No clicking to reveal. Land on the page and see everything.
Card states forced a product definition
Designing the states forced a conversation we'd been avoiding: what does it mean to "review" a candidate in this product? Is it watching the video? Rating them? Adding a note? We were getting this question from customers too — they wanted to know which teammates had seen which candidates. The card state system wasn't a visual design decision. It was the product's first answer to a workflow question customers were actively asking. Pink = unreviewed, needs attention. Green = reviewed, rated, in consideration. The eye goes to pink first — that's the action queue.
The card states became the communication system. Pink border meant new and unseen. Green border meant reviewed and rated. Star ratings on the card meant a team member had weighed in. A hiring manager could scan all candidates across all roles in seconds.
We also added the filter bar — View, Posted by, Sort by — and search with typeahead. "Future Applicant" placeholder cards showed exactly how many seats remained open, adding a sense of progress and urgency.
One of our earliest customers — a multi-location retail brand managing hiring across several stores — was the clearest signal that V3 was working. They'd started with one job posted. Within a few weeks they had four active, were using the team ratings panel to align on candidates across locations, and asked us about volume pricing. They weren't evaluating the product anymore. They were running their hiring on it. One hiring manager told us: "Seeing a video beats reading a resume every time — and now I can actually manage all of it in one place." That kind of expansion, that early, told us the persistent grid was the right call — not because it looked better than the accordion, but because it was the first version customers actually came back to.
Customer feedback · Post V3 launch
"Now I can actually see everything the moment I open the page. I know instantly which jobs have new applicants and who I still need to review. I don't have to manage the UI anymore — I just manage the candidates."
— Recruiter at a multi-location retail brand, after switching from the accordion version
Act 04 · The Mature Product
By the time we got to V4 a new customer type was showing up that V3 wasn't built for. Talent acquisition specialists — people managing hiring across many roles simultaneously, often at large organizations — weren't using My Jobs the way a single hiring manager would. They weren't watching every video. They were scanning for job health: which roles had traction, which were stalled, which needed more candidates. They needed bulk analytics, not a per-role candidate grid.
The customer who changed Act 04
"Fantastic overall experience. We love what this has done and will continue to do for our recruitment strategy. I enjoy the stats that show how many views jobs are getting and the ease of sending and receiving videos."
— Jen DeSwarte · Talent Acquisition Specialist · Caterpillar
Jen wasn't reviewing one candidate pile. She was managing a recruitment portfolio. V3 gave her candidates front and center — the right model for a hiring manager working a single role, completely wrong for someone scanning eight roles for pipeline health. That pressure is what inverted the information hierarchy in V4.
The final iteration made jobs the primary object. Each job card surfaced the metadata a TA specialist actually needed: candidate volume, view count, who's managing it, active or paused. The hiring manager could still drill in and see every candidate — the V3 grid was still there underneath — but the top level was finally built for someone scanning a portfolio, not a single pile of applications.
We also shipped the grid/list toggle — grid for visual scanning, list for density and power-user management. The product now served two distinct mental models: the hands-on hiring manager who wants to see faces, and the operations-focused recruiter who wants data.
Try it: Click any job card to drill into its full candidate view — that's the V3 persistent grid we built in Act 03, now nested inside the V4 job card model. The View filter, Sort, and Grid/List toggle all work. The key shift from V3: jobs are now the primary object, not section headers above candidates. You navigate into a job to see its candidates rather than seeing everything at once.
On visual design
Both screens were designed end-to-end in Figma — component library, auto layout, interactive prototypes, and dev handoff specs included. The candidate avatar stack with "+N" overflow, the view count badge, and the three-action row (Share / Preview / Manage) were all custom components built to scale across any number of job cards without breaking the layout.
Customer Outcomes
This is where the product ended up. Three customers — a multi-location retailer, a regional recruiter, and a TA specialist at Caterpillar — each representing a different stage of the product's evolution. The story of how we got here is four acts of building in public, selling while shipping, and letting customers tell us what the product needed to be.
The core tension across all four iterations: video hiring is a fundamentally different management problem. When candidates are faces and personalities instead of rows in a table, the entire mental model for managing applicants breaks. The product had to solve that before any of these outcomes were possible.
"Seeing a video beats reading a resume every time — and now I can actually manage all of it in one place."
Ashley HomeStore · Hiring manager
After expanding from 1 job to 4 active roles · Asked about volume pricing
"Now I can actually see everything the moment I open the page. I know instantly which jobs have new applicants and who I still need to review. I don't have to manage the UI anymore — I just manage the candidates."
Recruiter · Multi-location retail brand
After switching from the accordion version to V3
"Fantastic overall experience. We love what this has done and will continue to do for our recruitment strategy. I enjoy the stats that show how many views jobs are getting and the ease of sending and receiving videos."
Jen DeSwarte · Talent Acquisition Specialist · Caterpillar
Managing a portfolio of roles across a large organization · V4
What I Learned
Every layout decision flows from one question: what is the user trying to manage? When candidates are the primary object, you design a candidate grid. When jobs are the primary object, you design a job dashboard. The clearest decision we made was committing to jobs as the top-level navigation layer in V3 — and giving users the ability to drill into each one when they needed the full picture.
The job board vision was compelling, well-designed, and wrong. The customers who told us this didn't do it directly — they did it by asking the wrong questions for the product we'd shown them. We killed the direction before writing a single line of production code. That was only possible because we were talking to customers while we were still in prototype. If we'd shipped first, the feedback would have cost us months instead of a week.
The pink/green/grey border system on candidate cards was one of the most effective decisions in the whole product. Three colors, three states, zero labels on the grid view — but hiring managers understood it immediately. Designing the card states forced us to define exactly what "reviewed" meant in the product, which turned out to be an important product question we hadn't answered yet.
The core insight that drove the entire product was right: for customer-facing roles, a 60-second video introduction tells you more about fit than three rounds of résumé screening. How does this person present themselves? Are they articulate? Do they seem genuinely excited? JobPixel put that signal into the hiring loop at exactly the right moment — after the application, before the phone screen — and that's the insight the entire UX was built to serve.
Want this kind of thinking on your product?