PivotSearch
A LinkedIn Jobs feature helping mid-career professionals find companies with a real track record of hiring people who made their exact switch
In short
Pivot Search is a LinkedIn Jobs feature I designed to help mid-career professionals find companies with a real track record of hiring people who made their exact career switch, replacing guesswork with evidence. I took it from problem definition through a full PRD, a design sprint, a working prototype, real user testing, and a launch readiness review. The two biggest trust gaps found in testing were fixed and validated in a second round before recommending launch.
Overview
Mid-career professionals can't tell which companies actually hire people making their exact switch. Search tools treat a career change like an ordinary job search, ranking on title adjacency, so someone moving from IT support into product management is buried under experience-matched results and left guessing which doors are even open.
Pivot Search recognises a switch-style query and ranks companies by a switch-friendly score built from the volume, recency and background similarity of real switches into that company. Filters built for switchers — region, seniority of the switcher, and whether salary is disclosed — narrow the list, a company view shows the people who made the jump and when, and an intro request turns a shortlist into a conversation.
"As Ananya, an operations lead with six years of experience, I want to see which companies have actually hired people who moved from operations into product, so that I can spend my applications on places where my switch has already worked instead of guessing."
The decks
The Pitch
Background, business case, market sizing and the ROI argument for building Pivot Search.
Unrolling LinkedIn Pitch Deck…
How it came together
- Discovery — Framed the switcher problem against a ~$37.5B online recruitment market where Indeed and Naukri compete on title-matching. 40% of professionals are unsure which field to move into — and LinkedIn already holds the career-history data that answers it.
- Design Sprint — Five days of How-Might-We mapping, sketching and storyboarding narrowed everything to one testable concept: a switch-aware search bar plus a switch-friendly company score.
- Prototype v1 — A clickable flow inside LinkedIn Jobs: search a switch, scan ranked companies, open a company's switch history and people, then request an intro with a pre-filled editable message.
- User Testing — Moderated task-based sessions surfaced two trust gaps: participants wouldn't rely on a score they couldn't explain, and they wanted salary disclosure before deciding a company was worth exploring.
- Iteration (v2) — Added a plain-language 'what's this score?' explanation next to every score, and moved salary-disclosed into the results list as a scannable tag. Both fixes were validated in a second round.
- Launch Readiness — Test plan across P0–P2 cases, a dogfood survey, and a per-feature risk review. Feasibility checked out on a 6-month, 8-person squad (~$480K) against a plausible ~$20M/year Premium lift.
Risks & what we watch after launch
- Vague queries didn't always parse into a switch intent — mitigated with example query suggestions and autocomplete, monitored via the share of searches returning zero results.
- Score trust: one test case failed and the explanation modal didn't load on older app versions. The v2 explanation shipped and was re-validated; tap-rate on the info modal is tracked post-launch.
- Intro requests: a third of dogfooders wanted more guidance on what to say. The message stays editable with suggested prompts, tracked by send rate and reply rate.
Full detail — success metrics, feasibility and the complete risk table — lives in the PRD.
Retrospective
- I fell for the scoring algorithm before I validated that anyone trusted a score at all — the trust question should have been tested in week one, on paper.
- Salary visibility looked like a filter problem and was actually a scanning problem. Watching people scan the results list taught me more than the filter panel ever did.
- Holding the line on the last two weeks of validation was the right call: shipping early would have launched the score-trust issue unresolved.
Want the whole thing?
The PRD has every metric, risk and requirement. The prototype is clickable.