Ask for an org chart and you will get a tidy tree: boxes, lines, names, a comforting geometry of who reports to whom. Now trace how one real piece of work actually gets done, a customer problem, a product change, a pricing decision, and watch the geometry dissolve. The work flows sideways, loops through two people who appear nowhere near it on the chart, waits eight days at an approval that adds nothing, and gets rescued, as always, by the same three individuals everyone quietly depends on.
The chart is not lying maliciously. It is answering a different question: who is accountable to whom for people and budget. The question that determines whether your organisation is fast or slow, resilient or brittle, is different: how does work move, where does it wait, and who can say yes. Designing structure without mapping that flow is decorating a building you have never walked through.
If the structure only works when everyone is heroic, it is not a structure.
Spans, tiers and the cost of management
Two numbers reveal more than most reviews: spans and tiers. Where spans of control run at two or three, you are paying managers to relay messages, and each relay adds delay and distortion. Where tiers multiply, decisions climb five levels to someone who lacks the context to make them, then descend again stripped of nuance. Wide, flat structures are not automatically right, but every narrow span and every extra tier should have to justify itself, and most cannot.
Orphaned work and double ownership
The two design failures that cost the most are mirror images. Orphaned work belongs to nobody: it is everyone's job in the values statement and no one's job on a Tuesday, so it gets done badly or late or never. Double ownership belongs to two people who each believe the other defers, and it surfaces as the meeting where a decision is relitigated for the fourth time. A sound accountability map has neither: every outcome has exactly one owner, and decision rights are written down where anyone can check them.
None of this requires cleverness. It requires the discipline to map the work as it actually happens, design roles around outcomes rather than incumbents, and then hold the shape while it settles, because every reorganisation feels wrong for a quarter, and the difference between a good one and a bad one is whether it feels right in two.
What to do about it
- Trace three real pieces of work end to end before touching the chart. The waits are the findings.
- Count spans and tiers, and make every span under five and every tier past four justify itself.
- Write one owner against every outcome. Where you cannot, you have found the design problem.
- Publish decision rights. Ambiguity about who decides is the most expensive ambiguity there is.
Our management consulting practice does this mapping and design work with the people who live in the structure, which is the only way it holds. The chart should describe the organisation, not flatter it.