Case study · Kelvin Group · 2024–25
Hiring hard-to-find trades talent — at national scale.
Role
Creative & Digital Strategy Lead
Timeline
Ongoing, 2024–25
Channels
Video · LinkedIn · Paid Social · Print · Events
Team
Solo creative + 1 producer, AI-assisted
Hero recruitment spot — cut once, then repurposed across LinkedIn, Meta, and event loops without a reshoot.
Results
64%
increase in job-application volume
18%
expansion of the qualified talent pool
85%
reduction in concept-to-deliverable time
01 · Challenge
Six brands, one hiring problem.
Kelvin Group had rolled up six mechanical, compliance and inspection businesses under one parent brand. Each had its own recruiter, its own creative habits, and its own way of talking about the exact same open roles.
The result: expensive job boards, inconsistent applicant experiences, and hiring managers waiting weeks for on-brand creative that never quite matched the other five BUs.
Key constraints: no in-house video team, no dedicated recruitment budget, and a rapidly expanding open-req list.
02 · Strategy
Templates over one-offs. System over sprints.
I built a shared employer-brand kit — one master calendar, six BU templates, and a plug-and-play asset library — so every recruiter could ship on-brand creative without a design ticket.
On top of that, an AI-assisted production workflow (copy variants, cutdowns, quick localization) let a two-person team keep pace with a national blitz across LinkedIn, Meta, print and career fairs.
- ↗One hero video, five platform cutdowns — no reshoots
- ↗Six BU-branded postcard variants driving to a single tracked apply page
- ↗Career-fair kits sized for Mass Maritime, Penn College, Thaddeus Stevens, and Gillette Stadium
- ↗Weekly performance readout feeding the next round of copy tests
03 · Artifacts
Career-fair & open-house kit
One template, four schools, zero design tickets per event.
Recruitment postcards — six BU variants
Same layout system, six BU personalities. Front-of-house is consistent; back-of-card content is BU-specific and swappable without re-typesetting.
04 · What I’d do differently
Not everything worked the first time.
The first version of the shared template was too rigid — recruiters worked around it instead of through it. Version two loosened the copy blocks and locked only the brand-critical layers (logo, color, hierarchy). Adoption jumped.
I’d also invest sooner in a proper UTM + apply-page tracking scheme. We backfilled it after the first blitz; doing it upfront would have made channel attribution cleaner.
Have a similar problem?