Leadership & judgement
Video · 5 min read
The short answer
Watch: How does a leader know where they add the most value and when they should be involved?
Watch the explanation, then explore the practical implications below.
Where does leadership involvement improve the outcome?
A leader adds value where experience, accountability or knowledge of the business changes a decision. Being copied into every task is a poor substitute for identifying those points. It can create queues while leaving the real reason for approval unclear.
Review a small sample of recent decisions. Include an intervention that prevented a problem, a routine approval and a decision that reached you too late. Identify what you noticed and whether that insight was available to the team before the work began.
Separate repeatable guidance from judgement in context
Some corrections can become explicit instructions: use the approved rate card, include a required exclusion or obtain approval before offering a non-standard term. Others depend on context, such as the value of a customer relationship or a conflict between two important commitments.
Capture repeatable guidance in a maintained brief or workflow rule. Give it an owner, a scope and a review date. A correction for one customer should not silently become a rule for every customer. The business needs a deliberate route for approving and updating shared knowledge.
Design a clear route for exceptions
Define which situations people can resolve within their role and which need escalation. Useful triggers may include conflicting evidence, an unusual commitment, a missing approval or a decision outside an agreed commercial limit. Set the actual limits with the responsible leaders.
Also define who receives the exception and what happens if that person is unavailable. AI can prepare a concise decision pack, but the workflow should not infer approval because nobody replied. Where a decision requires authorisation, the system should wait or follow an agreed fallback.
A practical example: reviewing proposals
Imagine a commercial director repeatedly correcting proposals because the delivery team has not confirmed capacity. Requiring the director to inspect every draft treats the symptom. Add a capacity check before the proposal reaches approval, with the supporting evidence visible.
Routine proposals can then follow the agreed approval route. A proposal requiring a delivery date outside confirmed capacity becomes an explicit exception. The director sees the proposed commitment, the capacity gap, available options and the customer’s deadline rather than having to reconstruct the problem.
Make review useful, not ceremonial
An approver needs the source evidence, the proposed action, important uncertainties and a clear account of what has changed. They also need time and authority to reject, correct or return the work. A button labelled “Approve” does not establish those conditions.
Measure the pattern of involvement. Track waiting time, repeat corrections, avoidable escalations and problems missed by the process. Sample work that was not escalated as well as work that was. A quiet exception queue can mean clear rules, but it can also mean weak detection.
How CSM would approach the work
We would map a real workflow with its owner and the people doing the work. At each decision point, we would ask what judgement is required, who holds it and which information should be available. We would then separate routine preparation from decisions requiring authority.
The first implementation should make responsibilities visible and test difficult cases, including missing information and unavailable reviewers. Over time, approved learning can reduce repeated corrections while preserving meaningful escalation. The goal is to use leadership attention where it changes the result.
Questions for your leadership team
Where has my involvement changed the outcome?
Use recent examples to identify the judgement that mattered, rather than assuming every approval adds value.
Which corrections keep recurring?
Decide which should become approved guidance and which remain specific to a customer or situation.
What must trigger escalation?
Define conditions that people and the system can recognise, with a named decision owner.
What evidence should reach the reviewer?
Specify the recommendation, sources, uncertainty, alternatives and deadline.
How will we detect missed exceptions?
Review a sample of non-escalated work and examine mistakes as well as turnaround time.
Full video transcript
The full video transcript is included below for reference.
I think you have to make deliberate decisions about that. You can’t assume the answer is “everything important comes to me”, because that very quickly becomes the thing that stops the whole process working. But you also don’t want to hand over choices where your experience, your understanding of the customer or your responsibility for the business genuinely changes the result.
So I’d start by looking at where your involvement has made a difference in the past. What did you notice that someone else missed? Was it a commercial risk?
A commitment the company couldn’t meet? Something that looked sensible on paper but didn’t fit the relationship? Those are useful clues about where you belong in the process.
Then there’s another question. Does that decision need you every time, or is there something about your judgement that you can explain so other people, and the AI, can handle more of it? If you keep rejecting proposals for the same reason, that reason probably needs to be part of the brief.
Otherwise, the organisation is repeatedly creating work that only you know how to correct. Some decisions will still need you. A difficult exception, a trade-off between customers, a commitment that changes the direction of the business.
You should make those points clear, along with what information you need when the decision reaches you. And I’d review the pattern over time. If everything is being escalated, the boundaries may be too vague.
If nothing is being escalated, that doesn’t necessarily mean the system is working beautifully. It may mean nobody has told it what should cause concern. You’re designing how judgement moves through the organisation.
That takes some thought, but it gives you a much better chance of being involved where you’re useful rather than wherever the process happens to run out of instructions.
Put these questions to work in your business
Bring a real decision or workflow to a conversation with CSM. We can explore the evidence, responsibilities and practical steps needed to make AI useful for your team.