Blog
Complex IT Projects: Manage Adaptively, Not Rigidly
A systematic review published on 24 October examines 136 sources on complex projects. Its central finding matters for IT delivery as well. Complexity cannot be removed through ever more detailed planning; it has to be governed as a system of dependencies, feedback loops and changing relationships.
The authors distinguish five interconnected dimensions: structural, organisational, technological, contextual and dynamic complexity. Feedback loops, time delays and non-linear effects are especially important. A decision that seems minor in one workstream can therefore shift the schedule, budget or quality elsewhere weeks later.
Complexity Is Not the Same as Chaos
PMI’s Pulse of the Profession 2026 shows how widespread this is. 97 percent of respondents managed at least one complex project in the previous year, and almost one third of complex projects failed. According to PMI, successful organisations respond with outcome focus, alignment and faster learning rather than more micromanagement.
For steering committees, this translates into five practical questions:
- Structure: Which dependencies could shift several milestones at once?
- Organisation: Where do divided ownership or supplier boundaries slow down decisions?
- Technology: Which interfaces, data models or legacy systems can trigger cascading effects?
- Context: Which regulatory, contractual or operational assumptions might change?
- Dynamics: Which delay could turn today’s green status red within four weeks?
In migrations I often see individual work packages planned carefully while the overall system stays unstable. One delayed business decision can change mappings, test cases, reconciliation and cutover at the same time. Adding detail to the Gantt chart does not solve that.
A Practical Pattern for the Next Steering Meeting
Add a small complexity map to the status report, covering the five most important dependencies, their owners, expected response times and affected decision windows. In a data migration, each unresolved mapping point should show its due date and owner as well as the affected test cases, acceptance criteria and downstream decisions.
Track decision latency as well. If the same dependency stays open across three meetings, it stops being a detail and becomes a systemic risk.
What Project Leaders Should Do Now
Do not try to plan complexity away. Make interactions visible, shorten decision paths and review assumptions at fixed intervals. What matters is whether the overall system can respond to the next change, not whether every individual plan is green.