A client asked us last year to review a reorganisation they had completed eight months earlier. Three layers had been removed, two functions merged, and the org chart was noticeably cleaner. Nothing had got faster. Pricing exceptions still took eleven days. Capital requests still went round the same loop twice.
The diagnosis took less time than anyone expected. The structure had changed; the decision rights had not. Every approval that previously required four signatures still required four signatures, because the delegation matrix had been copied across into the new structure unchanged.
An org chart tells you who reports to whom. It tells you almost nothing about who can decide what.
Start with the decisions, not the boxes
Before touching structure, list the twenty decisions that most affect performance. For each one, write down who currently recommends, who approves, who must be consulted and who merely needs to be informed. Do this with the people who live the process rather than from policy documents, because the two rarely match.
Two patterns show up almost every time. The first is approval inflation: a threshold set a decade ago that inflation has quietly made meaningless, so trivial spend now reaches the executive committee. The second is consultation creep, where a function that was once consulted informally has become a mandatory gate with no service standard attached to it.
Then design the structure to fit
Once the decision map is agreed, structure becomes a narrower question: which grouping of people puts the recommender, the approver and the information in the same place most often? That usually produces a less elegant chart than a consulting template would, and a faster organisation.
- Push each decision to the lowest level that holds both the information and the accountability for the outcome.
- Give every consultation gate a published turnaround time. A gate without one is a queue.
- Retire a forum for every forum you create. Governance accumulates otherwise.
- Re-baseline approval thresholds against current revenue, not the revenue of the year they were set.
What good looks like six months later
The test is not whether people like the new structure. It is whether the cycle time on those twenty decisions has fallen, and whether the number of items reaching the executive committee has dropped without a rise in the number of decisions being reversed. If both hold, the reorganisation was real. If only the chart changed, it was not.