The Check-Back Cadence for Delegated Work
Why this matters
Two owners hand off the same task on the same day. One checks in whenever it crosses their mind, which turns out to be most days, and inside a month the person holding the task has stopped making decisions because there is always a check-in coming. The other says nothing for six weeks, discovers a problem that has been compounding for five of them, and concludes that delegation does not work here.
Both failed at the same thing: they never set a cadence. A cadence is a stated interval, a stated depth, and a stated end. Without all three, checking in is driven by your anxiety, and your anxiety is not a management system.
The three variables you are actually setting
Interval is how often. This is the one everybody thinks about and the least important of the three, because a wrong interval is easy to notice and easy to change.
Depth is what the check-back covers. A two-minute "all good?" and a thirty-minute walk through the work are both check-backs at the same interval, and they produce completely different behaviour in the person holding the task. Depth is where most cadences go wrong, in both directions.
Duration is how long the cadence itself lasts before it thins out or ends. If you never set this, the cadence is permanent by default, and a permanent check-back is a permanent statement that the task is still yours.
Start at weekly, and hold the check to 15 minutes
Weekly is the default starting interval for almost any task handed to someone already working in your shop. Commit to it and adjust from evidence, not from feel.
Weekly works because a week is the natural cycle of a service business. A full week contains a Monday, a Friday, a slow day and a bad day, so one week of a task is a fair sample of the task. Daily is the tempting alternative and it is usually wrong: a day is too small a sample to show a trend, so a daily check-back has nothing to read and fills the space with process questions instead, which is how you end up doing the work again through someone else's hands.
Cap the check at 15 minutes and hold that cap. A check-back that regularly runs 45 minutes has stopped being a check-back and become a working session, which is fine as a decision but should be a decision, not a drift.
Two cases justify starting tighter than weekly. A brand new hire in their first two weeks on anything customer-facing gets a daily touch, because at that stage you are teaching, not monitoring, and the cadence is training scaffolding you take down deliberately. And a task with a same-day irreversible consequence - anything that commits the shop to a customer promise you cannot walk back - gets a daily touch until you have watched the person handle it correctly three times in a row.
What a check-back contains: four questions, in this order
Order matters, because the first question you ask sets what the person prepares for next time.
- What does the number say? Read the indicator first, before any discussion. Starting with the number keeps the conversation on outcome rather than effort, and it means a good week reads as a good week even if it felt hard.
- What did you decide that you were not sure about? This is the highest-yield question in the whole cadence. It surfaces the judgment calls without implying they were wrong, and it is where you actually transfer decision-making. Ask it every time, and when the answer is "nothing," believe it once and probe it twice.
- What is going to be a problem next week? Forward-looking, so you are clearing an obstacle rather than reviewing a result you cannot change.
- What do you need from me? Last, deliberately. Asked first, it turns the check-back into a request queue and you leave with the work.
Notice what is not on the list: a walk through everything they did. If you want to inspect the work itself, inspect a sample of it on your own time, against a written standard, and bring the finding to the next check-back. Inspecting live turns the meeting into a performance review that neither of you scheduled.
The decay rule: halve the frequency after two clean cycles
State this at handoff so it is a rule, not a favour you grant later.
After two consecutive check-backs where the number sat inside its band and no tripwire was missed, halve the frequency. Weekly becomes every other week. Every other week becomes monthly. Monthly becomes exception-only, meaning the cadence ends and the tripwires carry the load.
Two cycles, not one, because one clean week is a coin flip in a shop that runs a few dozen jobs a month. Halving rather than a gentler step, because a cadence that eases at a leisurely pace never actually ends, and the point of the whole exercise is that it ends.
The tightening rule is not the mirror image. Go back up one step - not to the beginning - after either two consecutive cycles outside the band, or a single missed tripwire on something with a customer consequence. One missed tripwire matters more than two soft numbers, because a missed tripwire means the safety net has a hole in it and the numbers only mean the work is running rough.
Cadence by task risk
| Task shape | Start | Depth | Decay to |
|---|---|---|---|
| Reversible, high volume (ordering, scheduling adjustments) | Weekly, 15 min | Number plus decisions | Exception-only in about 3.5 months at the earliest |
| Reversible, low volume (supplier reviews, quarterly reordering) | Once per occurrence, first 3 | Full walk-through | Exception-only after 2 clean |
| Customer-facing, same-day irreversible (promises, pricing on the phone) | Daily until 3 correct in a row, then weekly | Sample the actual calls | Monthly, and stop there |
| Money-out (paying suppliers, approving spend) | Weekly, with a written authority ceiling | Every item above the ceiling | Monthly, ceiling raised in steps |
| Judgment-heavy, no clean number (a difficult customer, a struggling hire) | Every other week | Sample of 2 or 3 instances | Does not decay to zero, goes to monthly and stays |
The last row is the one owners get wrong. Some work should carry a permanent light cadence, and admitting that up front is better than pretending it will decay and then feeling guilty when it does not.
A worked cadence: three tasks on one calendar
An owner hands off three things over one month: collections follow-up, parts ordering, and building the next week's schedule. All three start weekly at 15 minutes.
Starting load. Three tasks, four check-backs a month each, 15 minutes each, so 180 minutes a month, or 3.0 hours a month of the owner's time spent checking on delegated work. Worth naming out loud, because owners tend to imagine delegation is free on day one and then feel ambushed by this.
Month two. Collections has run two clean cycles: aged invoices past terms held inside band both times, no tripwire missed. It halves to every other week. Parts ordering had one week outside band and one inside, so it stays weekly - it did not earn the decay and does not get it. The schedule task hit a missed tripwire, a customer promised a window that was already full, so it stays weekly and the owner adds a fourth tripwire covering that case.
Month four. Collections has run two clean cycles at every other week, so it goes monthly. Parts ordering has earned its decay and sits at every other week. The schedule task is still weekly by the owner's judgment, because the person holding it is newer and the consequence is customer-facing.
Monthly check-back time is now 15 minutes for collections, 30 for parts, 60 for the schedule, so 105 minutes a month against the 180 minutes a month it started at. That is a drop of 75 minutes a month, about 42% of the original monthly check-back time, arrived at without the owner making a single judgment call about whether it "felt right" to back off.
Month six. Collections goes exception-only. The owner's monthly check-back load on three delegated tasks is 90 minutes, and the two tasks still on a cadence are the two that genuinely warrant one.
Who runs the check-back once there is someone between you and the work
The moment a shop has a lead or an office manager supervising others, an owner-run cadence on every delegated task stops being supervision and starts being a second reporting line. Two people checking on the same work produces one of two outcomes and neither is good: the person holding the task reports twice, or they learn which of you actually acts on the report and quietly stop preparing for the other.
The rule is that a task has exactly one check-back holder, and it is whoever the person holding the task reports to on everything else. If you have put a lead in place and then kept the cadence on a task inside that lead's area, you have undone the lead.
Your cadence then moves up a level. You are not checking the task, you are checking the lead's cadence on it: did they run it, what did the number do, what decision did they make. Same four questions, one layer removed, and typically at half the frequency of the underlying cadence. This is where an owner recovers the largest block of time in the whole exercise, and it is also where the most common regression happens - a bad month arrives, the owner reaches past the lead to check the task directly, and the layer they spent a year building disappears in a week. A sibling article covers building that layer.
When a check-back has quietly become something else
Four tells, each with a different repair:
- It runs long every time. The cap is not the problem, the scope is. Something in the task needs a written rule so it stops being decided live. Find the recurring question and answer it once, in writing.
- They arrive with questions instead of numbers. The authority boundary was never set. They are not seeking a check-back, they are seeking approval, and the cadence is where they queue it. Write the ceiling and the decision rules.
- You do most of the talking. You are training, not checking. That is legitimate for a while, but call it what it is and give it a separate slot, or the training will never end because it is disguised as monitoring.
- You skipped it twice and nothing bad happened. Do not congratulate yourself and do not quietly abandon it. Move it formally to the next decay step. An abandoned cadence is worse than a decayed one, because the person cannot tell whether the silence means trust or means you have lost interest in their work.
References
- See related: How to Delegate Without Losing the Thread
- See related: Delegating Without Letting Quality Slip
- See related: The Owner Who Kept Taking the Task Back
- See related: Delegation Readiness: Keep vs Hand Off Decision Tree
- Trade-standard practice for supervision and handoff in small field-service teams