Five roles, not five steps
| Letter |
Role |
What goes wrong when it is left unassigned |
| Recommend |
Builds the proposal: options, analysis, a clear recommendation |
The Decide role receives raw noise instead of a testable proposal |
| Agree |
Must formally sign off: typically legal, regulatory or safety |
Compliance objections surface after commitment, at maximum cost |
| Perform |
Executes the decision once made |
The people delivering learn of the decision last and renegotiate it as they go |
| Input |
Consulted for expertise; heard, not necessarily heeded |
Functions escalate to reopen decisions because they were never asked |
| Decide |
One person makes the call and commits the organisation |
Consensus becomes the default decider, and pace dies |
The organisations that need it most
RAPID was designed by Bain & Company for precisely the environments where matrix structures, strong functions and global-versus-local tension make every significant call a negotiation. If your organisation regularly discovers that a decision it thought was made is still being debated two levels down, or that regional leadership treats group decisions as opening offers, the problem is unallocated decision rights. That is the specific problem RAPID addresses.
Putting it to work on a real decision
- Assign the D to a single named person and publish it. A decision right that is secret is not a decision right.
- Keep the A role scarce and justified. Every Agree holder is a genuine veto; if more than two or three exist, the decision will route around them or die.
- Distinguish Input from Agree explicitly when inviting people in. Most stakeholder resentment comes from people who believed they held an A and discover they held an I.
- Give the Recommend role real analytical resources and a deadline. A weak R starves the D, and the vacuum fills with lobbying.
A sketch from the field
Imagine a multinational deciding whether to consolidate three regional distribution centres into one. Group operations holds the R and builds the case. Legal and the works council process hold A roles, because consolidation triggers consultation obligations that can genuinely stop the move. Regional managing directors hold I: their knowledge of local customer commitments is essential input, but their preference to keep their own sites is not a veto. The COO holds the D; site operations teams hold P. The uncomfortable clarity of telling regional MDs they are Input, not Agree, is exactly the conversation RAPID exists to force early rather than late.
How RAPID gets defeated
The framework fails through role inflation and slow reversals rather than open rejection.
- The Agree column swells until it is a veto web: every function that asks becomes an A, and the decision now requires unanimity while pretending not to.
- The D is assigned by rank rather than by accountability for the consequences, so the formal decider defers to whoever has the strongest opinion in the room.
- Input holders who dislike the outcome escalate around the process, and leadership entertains the appeal, which teaches everyone that the RAPID chart is provisional.
- The P role is treated as an afterthought. Decisions get made that the performing organisation cannot staff, and the failure is later blamed on execution rather than on the decision process.
Strengthening the R before it reaches the D
RAPID concentrates enormous influence in the Recommend role: the decider mostly sees the options the R chose to present. That is where independent perspectives earn their place. Having selected senior operators pressure-test the recommendation (the discarded options, the assumptions doing the heaviest work, the execution risks the deck understates) gives the decider confidence that the choice in front of them survived challenge from people with no position to defend.
