Delivery field notes
Risk owner vs action owner: who does what?
The difference between a risk owner, an action owner and an approver, how to write a response with a trigger that arrives in time, and a worked register row.
A register full of names can still be empty
A fictional delivery team keeps a tidy risk register. Every row has an owner, a probability, an impact and a red, amber or green rating. Yet the top risk, a supplier who might deliver late, has said "Monitor" in the response column for six weeks. Nobody knows what would trigger a fallback, and nobody has asked the rental company whether a fallback exists.
The register looks maintained. It records concern rather than directing work. The usual cause is that one column, "owner", is doing three different jobs.
Three roles, often confused
| Role | Responsible for | Typical person |
|---|---|---|
| Risk owner | Keeping the risk understood, watching the trigger, and making sure the response happens | The person closest to the cause, with enough standing to raise it |
| Action owner | Completing one specific response task by a date | Whoever can do that task: often not the risk owner |
| Approver | Authorising a response that is outside the team's delegation: money, scope or date | The sponsor or budget holder |
One person can hold more than one role. A delivery lead might own the risk and also do the first action. The mistake is assuming that writing one name covers all three. If the sponsor's name is in the owner column, the preparatory work (getting a quote, checking the alternative fits) has no owner at all, and the sponsor finds out about the risk when asked to approve money at the last minute.
The response is the part that changes the exposure
The UK government's Orange Book, HM Treasury's guidance on managing risk, is explicit that choosing a response is not the end of the work. It says that, when selecting and developing risk treatments, an organisation should specify how the chosen option will be implemented, so that the arrangements are understood by those involved and their effectiveness can be monitored. It adds that monitoring should play a role before, during and after the treatment is implemented.
In register terms, that means a response needs:
- A specific action, not "monitor" or "manage closely".
- An action owner and a due date.
- A trigger: the evidence that tells you the risk is turning into an issue, early enough that the fallback still works.
- A decision point: who approves the fallback, and by when.
The Orange Book also lists the broad choices for responding to a risk, including avoiding it by not starting or continuing the activity that gives rise to it. Most project risks are reduced or prepared for, not avoided, but writing down which choice you made stops "monitor" from becoming a default.
Set the trigger by working backwards
A trigger is useful only if it arrives while you can still act. "Escalate when installation is delayed" is too late: by then, the fallback has no time to work. Work backwards from the last date the fallback can still save the plan.
Fictional example: card readers for an office move Go-live date ........................................ 2 Dec Installation takes .................................. 5 working days Latest installation start .......................... 25 Nov Rental readers need ................................ 5 working days to source Latest date to order rental readers ................ 18 Nov Supplier must confirm a delivery date by ........... 14 Nov ← trigger
The trigger, "no confirmed delivery date by 14 November", gives the team two working days to get approval and order rental readers before the 18 November deadline. Any later trigger means the fallback cannot arrive in time.
NASA's Systems Engineering Handbook describes two complementary risk processes: risk-informed decision making, for direction-setting choices such as design decisions, and continuous risk management, used throughout development and implementation to keep safety, technical, cost and schedule requirements on track. Triggers belong to the second. Working back from a deadline is how continuous management becomes concrete: each trigger has a date, and the date comes from the plan.
A worked register row
Here is the same risk before and after, using the fictional office move.
| Field | Before | After |
|---|---|---|
| Risk | Supplier delay | If the card-reader supplier does not deliver by 22 Nov, installation cannot finish before 2 Dec go-live, and 140 staff cannot enter the building |
| Owner | Sponsor | Delivery lead (Sam) |
| Response | Monitor | Prepare a fallback: rental readers, quoted and checked for compatibility |
| Actions | none | Get rental quote (Sam, 11 Nov). Confirm rental readers work with the access system (Facilities, 12 Nov) |
| Trigger | none | No confirmed delivery date from supplier by 14 Nov |
| Decision | none | Sponsor approves rental cost (est. €7,500) by 18 Nov if triggered |
| Status | Amber | Amber: quote received 10 Nov; compatibility check due 12 Nov |
The "after" row tells a new reader what could happen, what is being done, what would set off the fallback and who decides. The sponsor appears once, in the decision field, for the one thing only they can do.
Where probability and impact scores mislead
Most registers score each risk for probability and impact, multiply them and colour the result. The scores help sort a long list, but they can hide the things that matter for action.
Two risks with the same score can need opposite responses. A risk with high probability and low impact (a small, likely delay) and one with low probability and high impact (the building fails its fire inspection) can land on the same number. The first needs a buffer in the plan. The second needs a contingency plan and a named approver.
Scores do not say when. A risk scored "high" that cannot happen for three months is less urgent this week than a "medium" risk whose trigger date is Friday. Sort your weekly review by trigger date, not only by score.
Scores drift without evidence. If a probability moves from 4 to 2, the row should say what changed: a confirmed delivery date, a test passed. A score that changes without a stated reason is an opinion that has changed.
Use scores to decide which risks get a full response. Use trigger dates to decide which ones get attention this week.
A better weekly review
Most risk reviews walk through every row and ask "any change?". A shorter set of questions works better:
- What changed since last week, and does any trigger date fall before the next review?
- Which actions are overdue, and what is blocking them?
- Which risks have said the same thing for three reviews? Either the risk is no longer live, the response is not working, or the trigger is wrong.
- Does any response now need a decision outside the team's authority?
The third question catches most stale registers. A risk that has read "Monitor" for six weeks is either not a risk or not being managed.
When you close a risk, write one line on why: the trigger passed without firing, the response worked, or the risk turned into an issue and moved to the issue log. Closed rows with reasons are the raw material for the project's lessons learned.
Try it on one row
Take the riskiest row in a register you know, or invent one for a project you have run, and spend fifteen minutes on it:
- Split the owner into risk owner, action owner and approver.
- Replace any "monitor" with a specific action and a due date.
- Work backwards from the deadline to set a trigger date.
- Write who decides on the fallback and by when.
Keep the original next to the rewrite so you can see the difference in reasoning. The risk register template has these fields ready, and the RAID log example shows how risks sit alongside assumptions, issues and dependencies.
To practise writing risk responses with owners and triggers, try the open exercise Municipal permit portal: risk register.
Sources
Method background
These sources provide general background. The worked examples and worksheets are original Response Shift teaching material; the organisations do not endorse this product.
