The Custodian Model: Keeping Objectives Accountable When Someone Is Away

Most OKR failures are not dramatic. They do not announce themselves in a post-mortem. They happen because the person whose name sat beside the key result went on parental leave in week three of the quarter, and nobody quite decided what came next.

In UK workplaces this is not an edge case. Shared Parental Leave can run to 52 weeks, with up to ten keep-in-touch days. Job shares are now routine in operations, finance and HR. Long-term sickness absence is a reality every manager will meet. Yet the OKR document usually assumes a stable cast of characters who are all present, all quarter.

Why leaving it to goodwill fails

The instinct is to say that a colleague will pick it up. Sometimes they do. More often, accountability in an OKR set is social rather than documentary: people act because a specific person is visibly watching the number. Remove that person and the key result does not fail loudly. It drifts, then it gets quietly reclassified as business as usual, and by the end of the quarter nobody can say when it stopped being real.

The second instinct is worse: appoint a deputy for everything, and a shadow hierarchy forms in which nobody is quite sure who decides.

Owner and custodian are different roles

The cleanest fix is to separate two things that UK teams usually bundle together.

An owner holds long-term accountability: they set the level of ambition, defend the trade-offs and answer for the result at the end of the period. A custodian holds the objective for a defined window, typically eight to twelve weeks, and is accountable for cadence, honesty and escalation.

The custodian's decision rights should be written down and narrow. They can renegotiate confidence. They cannot renegotiate the objective. They can move work between people. They cannot add a new key result.

Three rules keep the model from becoming a parallel management structure:

• Only one name appears on the owner line. The custodian is flagged separately, with an end date.
• A custodian does not inherit a deputy's seat at leadership review. They report; they do not represent.
• Custody is funded. If someone takes on an objective for a quarter, one of their own key results comes off the board. Unfunded custody is just additional unpaid work with a title.

Handover notes that do not become a second job

The failure mode here is a twelve-page document assembled the night before an absence begins, full of status history nobody reads. Cap it at three pages and six questions:

1. What does done look like by the end of the year, in one sentence?
2. Where are we against the number today?
3. Which decisions are pending, and who is allowed to make them?
4. Who are the five people who actually move this forward?
5. What have we already tried that did not work?
6. What would make you escalate in your second week?

The document is maintained inside the weekly review, not written on the eve of departure. If it is not already live, the handover has failed before it started.

Job shares: two names, one voice

A shared role creates a specific hazard, which is that both holders turn up to every review and the objective acquires two competing narratives. The fix is procedural rather than personal: one written voice per week, agreed between the pair before the review, and never both attending merely to be safe. The handover between sharers should be a standing twenty-minute item with a fixed agenda, not a rolling email thread.

It also helps to write key results that do not encode one person's working pattern. A key result phrased around daily standups, or availability on a Friday, will quietly punish the job-share arrangement rather than measure the outcome.

Sickness and sudden absence

Sudden absence admits no handover. The only workable answer is a default named at the point the objective is set: if I am unreachable for five working days, this person holds it. That line is written once, checked quarterly, and takes ninety seconds to agree.

Two boundaries are worth stating plainly. First, no colleague should be asked to open an absent person's inbox or notes to reconstruct a status. The objective is the unit of continuity, not the individual's working life. Second, the line manager owns the pastoral side; the custodian owns the number. Confusing the two is how teams end up with a spreadsheet that knows more about a person's absence than their manager does.

Re-entry: forty-five minutes, three questions

The return conversation is frequently treated as a status update, which wastes it. It should instead be a recalibration:

• What changed while I was away that I would not have predicted?
• What did the custodian decide that I would have decided differently?
• What should we stop now that I am back?

Follow it with a two-week ramp in which no new commitments are accepted. Handing someone four weeks of unread updates and a full objective slate on their first morning is not re-entry; it is a second absence waiting to happen.

The ten-minute quarterly orphan check

Run this alongside any existing review. It needs no tooling beyond the OKR document.

• Every objective has a named owner who is present and available for the next eight weeks.
• Every custodian has a stated end date, and no custody has silently rolled past it.
• No owner line carries more than two names.
• Any known absence over four weeks has a handover document that is under three pages and dated within the last fortnight.

Four checks, ten minutes, once a quarter. It will surface the objectives that were about to become nobody's business, which is the entire point.

Continuity of commitment is what this protects. Objectives survive people being away not because the paperwork is thorough, but because the team decided in advance who would be watching the number, and for how long.