Josh Hornby
  • Tags
  • RSS

Doing Leveraged Work

software-engineering (1)

For Staff and Principal Engineers, the goal is to identify and deliver leveraged work, efforts that multiply both throughput and quality across the organisation.

Two-axis graph plotting impact and longevity in relation to Leveraged work

Picture leverage as a simple chart: time on the x-axis, impact on the y. The real win is top right contributions that endure and change the trajectory. Not just a quick boost or flashy result, but work that compounds. It becomes part of the foundation others build on, shifts the baseline, and raises the bar for everything that follows.

If it… …then it probably belongs in that sweet-spot
keeps paying dividends long after the launch Longevity is doing the heavy lifting
turns one hour of effort into many hours saved for others Non-linear impact is showing up
lets teams build on it rather than around it clear Enablement
leaves a template, script, or doc that’s easy to copy strong Teachability

The art of leveraged work is finding where the opportunities lie, the majority of the time these can be found through conversations with team members and the wider business.

  • Pain-point mapping - sit with on-call, read incident reviews, notice recurring toil
  • Road-map stress-tests - ask, “What single failure would torpedo next year’s plan?” and front-load the fix
  • Glue-work synthesis - connect patterns observed across teams and turn them into shared libraries or services

Below are some concrete examples of leveraged work:

  • Feature-flag platform - internal service + SDKs for fast, reversible releases
  • Incident-retrospective programme - a lightweight 5-Whys process that looks at systemic fixes
  • Service-bootstrap template - a generator that spins up a production-ready app in minutes

Leveraged work is how Staff & Principal engineers multiply impact as it’s durable, scales beyond the individual, and spreads by design.


Published on May 31, 2025.
Previous Post Stuff I learned at cinch
Next Post Improve LLM prompts using LLMs