Decision rights are part of the architecture
Why faster information needs a workable route to action

- By
- Cam Myers
- Content type
- Point of View
- Reading time
- 6 min read
- Status
- Author’s perspective
- Topics
- Operating Models
In brief
- Responsibility for an outcome needs sufficient influence over the decisions that determine it.
- Approvals have an operating cost, and their capacity can become the constraint.
- Delegation needs explicit boundaries, exception handling, and review of outcomes.
The argument
A company can identify a problem on Monday and still spend three weeks deciding who is allowed to fix it.
The delay may be blamed on communication or resistance to change. Sometimes those explanations are fair. Often the business has designed an arrangement in which a decision crosses several departments, carries consequences for all of them, and belongs clearly to none of them.
Software will encounter that arrangement. It can expose the delay and move information faster, but the authority to act still has to exist.
When I look at a business system, I include the people who can change its behavior. A pricing rule, an inventory allocation, and a campaign budget all reflect decisions about resources. The technical design and the management design meet at those decisions.
The decision underneathWho can act, within which limits, and by when?
Approval has an operating cost
Each approval can serve a useful purpose. It can also add waiting, preparation, repeated explanation, and opportunities for a request to be lost.
The right question is whether the protection gained is worth that cost for this class of decision.
A one-time decision to replace the accounting platform deserves broad review. A reversible adjustment to a small advertising test probably should not travel through the same route. Treating both alike consumes senior attention and delays ordinary work without necessarily improving control.
Decision volume matters. A process that works for five exceptions a month can become a bottleneck at fifty a day. As software reduces the effort needed to identify and propose changes, approval capacity can become the new limit on improvement.
Queueing makes this worse. A manager who is nearly fully occupied may have little room to handle a rush of requests or an unusually difficult exception. Individual decisions might take only minutes, while the wait to receive those minutes stretches across days.
Adding another dashboard does little for that problem. The business may need clearer delegation, fewer cases requiring approval, or dedicated time to resolve the queue.
Owners should pay attention to total elapsed time, including how long decisions wait between people. That is often where the largest delay sits.
Distinguish a dependency from a preference
Cross-functional work also stalls when everything becomes a prerequisite.
A new capability must wait for perfect data, a complete process manual, another department's redesign, and agreement from every affected person. Some of those requirements protect the business. Others reflect a preference for completing related work together.
The distinction becomes clearer when management describes the consequence of proceeding. Which failure could occur? How serious would it be? How would the business detect it? Could the exposure be limited or reversed?
An unresolved inventory identity problem may be a genuine blocker for automated customer promises. It may be tolerable in an internal analysis that flags uncertain records and makes no external commitment. The same data issue can require different decisions depending on the intended action.
This is where bounded trials are useful. Management can authorize a limited scope, define unacceptable outcomes, and establish who can stop the test. A trial produces evidence while keeping the potential consequences proportionate.
The trial needs a decision date and an accountable reviewer. Otherwise temporary permission can become permanent uncertainty, and the team remains unable to commit to either expansion or closure.
Own the outcome across departmental boundaries
Departmental goals can pull a business in different directions even when every manager behaves reasonably.
Purchasing may earn praise for lower unit costs while buying quantities that tie up cash. Sales may maximize bookings that strain delivery capacity. Marketing may reduce lead costs by attracting customers who are expensive to serve.
Improvement requires someone to judge the combined effect. That person needs access to the relevant information and enough authority to resolve the tradeoffs. A title alone is insufficient if each department can independently block the resulting decision.
The owner should make that authority concrete. Which decisions can the responsible leader make? Which limits require escalation? Who resolves disagreement, and by when? What evidence will show whether the chosen intervention helped the business as a whole?
Consultants and technical teams need the same clarity. They can design a capable system and recommend an operating change, but they cannot manufacture authority the client has not assigned. The engagement should identify who can authorize the changes required for the promised result.
Better delegation also needs follow-through. Review decisions against their outcomes, adjust limits when experience warrants it, and distinguish an honest result within agreed rules from a failure to follow them. People will keep escalating routine decisions if using delegated authority leads to unpredictable punishment.
Make the cost of waiting visible
Approval discussions tend to examine the risk of making a change more closely than the risk of continuing as usual. Yet slow stock keeps consuming space, a weak route keeps losing money, and an unanswered customer keeps looking elsewhere while the company deliberates.
Estimate those continuing costs alongside the possible damage from intervention. The numbers may be uncertain. Even a reasonable range can improve a discussion that otherwise treats delay as free.
Different decision routes are appropriate for different exposures. A limited, reversible test can justify a shorter route than an irreversible commitment affecting the entire company. The owner can retain control over important capital and policy decisions while delegating ordinary adjustments within an agreed operating budget.
The arrangement should also allow information from the work to change the rule. If every week brings the same exception, management may need to revise the policy rather than congratulate itself on reviewing each case. Repeated exceptions are evidence about the design of authority, just as repeated software failures are evidence about technical design.
Before the next technology rollout, take one important recurring decision and trace its path through the company. Count the handoffs, identify who can say yes, and find the point where responsibility becomes unclear.
Fixing that route may produce more value than making the information arrive another hour sooner.
For the leadership team
In practice
- Trace a recurring decision through the handoffs, approvals, and waiting time.
- Distinguish necessary safeguards from preferences that have become prerequisites.
- Make the continuing cost of delay visible alongside the risk of intervention.
Questions worth asking
- Does the person accountable for this outcome have authority to change its conditions?
- Which recurring exception suggests that the approval rule itself needs to change?
Weighing this decision now? Tell us what you are considering and we will say plainly whether we can help.
Discuss your business

