The first 90 days.
Hiring a delivery leader is a bet on how fast they can read a function, steady it and make it perform. This is the pattern I run when I land — the same one behind the case studies: diagnose honestly, design with the team, then deliver and embed so it runs without heroics.
Understand the machine before touching it
- Meet every squad, delivery lead and the key stakeholders. Shadow the ceremonies, sit in on live escalations, read the last quarter's reporting cold.
- Map demand honestly — roadmap, paid client work and SLA-bound support — and find where the same people are being pulled three ways at once.
- Baseline the numbers that matter: delivery predictability, escalation volume, budget burn against plan, and where work actually queues.
- Land two visible quick wins early — unblock a stalled release, kill a report nobody reads. Credibility is earned in the first month, not the first year.
By day 30: a one-page honest diagnosis the exec hasn't seen before, baseline metrics agreed, quick wins delivered, and no surprises in the pipeline.
Build the operating model with the team, not at them
- Stand up a prioritisation model that balances roadmap, revenue-generating projects and support — with a trade-off forum where the exec sees exactly what a yes costs.
- Reset governance to fit the organisation: RAID discipline, clean escalation paths, and reporting that reads the same from squad level to the boardroom.
- Put commercial guardrails in place — budget tracking, supplier and third-party cadence, and forecasting that tells the truth early rather than late.
- Review team shape and capability: progression clarity for engineers and delivery managers, coach-or-hire decisions made, third parties brought into one way of working.
By day 60: the model is live, the first governance cycles have run, the commercial picture is visible weekly, and the team believes the plan because they helped build it.
Prove it with delivery, then make it stick
- Milestones start landing under the new model — and the predictability trend is visible on a chart, not claimed in a slide.
- Modernise where it pays: engineering practice, automation and AI adoption grounded in what I ship hands-on, not vendor theatre.
- Embed the cadences with named owners and clean documentation, so the function runs on rhythm rather than firefighting.
- Close the loop with a 90-day review: outcomes against the day-30 diagnosis, and a costed plan for the next two quarters.
By day 90: measurable lift in predictability, fewer escalations reaching the exec, roadmap moving while clients stay served — and a function that scales beyond me.