How to Delegate Without Losing the Thread
Why this matters
Most work that comes back to an owner does not come back because it was done badly. It comes back because the owner went blind on it, and blindness on something you are still responsible for feels exactly like risk. So you ask how it is going. Then you ask again on Thursday. Then you are back inside it, and the person you handed it to has quietly stopped owning it, because it was obvious you never really did.
The fix is not more trust and it is not willpower. It is deciding, before you hand anything over, exactly what you will still see and how it will reach you. A handoff with a thread attached survives your own discomfort. A handoff without one lasts about three weeks.
Step 1: Name the one number that tells you the work is healthy
A thread is one indicator you can read in a few seconds without talking to anyone. Not a report, not a conversation, not a feeling. One number, checked on a stated interval, that moves when the work goes wrong.
Pick it by asking what failure of this task would actually cost you. Parts ordering fails as jobs delayed for want of a part. Scheduling fails as second visits that should have been one. Collections fails as invoices aged past your terms. Estimate follow-up fails as quotes that go quiet with nobody having called.
Two rules for the number:
- It measures the outcome, not the activity. Orders placed is activity, and it goes up when things go badly. Jobs delayed by parts is the outcome. If your number can look good while the customer has a bad week, you picked the wrong one.
- It has a baseline before the handoff. Measure it for two to four weeks while you still own the task. Without a before, every number the new owner produces reads as either fine or alarming depending on your mood that day.
Step 2: Write three tripwires that reach you without anyone deciding to escalate
The number tells you the trend. Tripwires catch the single event that cannot wait for the trend. Write exactly three, because a list of eight is a list nobody remembers and effectively becomes zero.
A tripwire is only usable if it is observable by the person holding the task, not by you. "Anything that seems like a big deal" is not a tripwire, it is a request for that person to read your mind, and it fails on the day it matters. "Any order that pushes a job past a date we already promised the customer" is a tripwire, because the person can see both facts on the screen in front of them.
Write each one as a condition and an action in the same line: when this is true, message me the same day. Say what happens next, too - do they stop, or do they proceed and tell you? A tripwire that stops work is a brake. One that proceeds and reports is a mirror. Most should be mirrors. Reserve the brake for the two or three cases where an undo is genuinely expensive.
Step 3: Set the reporting line, then take yourself off every other channel
Decide where the number lands and when. A single line, in a place you already look, on a fixed day. Friday afternoon is common because it lets you read the week without acting on it until Monday, which is exactly what you want for a task you are trying to stop touching.
Then do the harder half: turn off the incidental visibility. If you are still copied on every supplier email, still in the group thread where the questions get asked, still the one whose phone rings when the counter calls, you have not delegated the task, you have hired an assistant for it. Incidental visibility is the enemy here, because it gives you the raw material for a worry without giving you the whole picture, and a partial worry is what makes an owner take a task back.
Step 4: Set the decay schedule at handoff, not later
Write down, on the day you hand off, when the thread gets thinner. Something like: weekly for the first four weeks, every other week for the next four, monthly after that, and exception-only once the number has held inside its band for two consecutive months.
Setting the decay in advance does two things. It tells the person this is a ramp, not permanent supervision, which changes how they treat the early reports. And it takes the loosening decision off your desk in a moment when you will be inclined to keep it as it is because the current arrangement feels safe.
Step 5: Read the movement, do not react to a single point
One bad week is noise. Small shops produce small samples, and a single number from a single week is one job's worth of luck in either direction. Decide in advance what counts as a real move: two consecutive periods outside the band, or one period more than roughly half again the band's width.
When a real move happens, your first question is not "who dropped this," it is "which of the three tripwires would have caught this, and did it fire?" If a tripwire existed and did not fire, you have a tripwire problem, and that is yours to fix. If no tripwire covered it, you have found the fourth one you should have written, and you write it now instead of taking the task back.
A worked handoff: parts ordering
A shop runs about 55 jobs a month and places roughly 40 part orders a month against them. The owner has been placing every order personally for four years and it eats a piece of most evenings.
The baseline. Before handing off, the owner counts jobs delayed by a missing or wrong part for three weeks and finds it running at about 1 in 8 jobs, so roughly 12.5% of jobs, which on 55 jobs a month is about 7 delayed jobs a month. That is the thread number: parts-caused delays as a share of jobs, read monthly.
The band. The owner sets the acceptable band at 1 in 8 or better, which is where it already is. That is deliberate. The point of the first ninety days is not improvement, it is not getting worse while the task changes hands. Setting the target at 1 in 15 on day one would guarantee a red number in month one and guarantee the owner takes it back.
The tripwires. Three, written on the handoff sheet:
- Any order that pushes a job past a date already promised to a customer. Proceed, message me the same day, and call the customer before I hear about it from them.
- Any supplier substitution on a part going into warranty work. Stop, ask me first. This is the brake, because a substituted warranty part is expensive to undo after it is installed.
- Any top-ten truck-stock item hitting zero on hand. Proceed, order it, tell me at the Friday line.
The reporting line. One line, Fridays: orders placed, jobs delayed by parts this week, substitutions. Three numbers. The owner comes off the supplier email list the same day.
What happened. Weeks one and two run at 2 delayed jobs each, which annualizes worse than baseline and is exactly the noise the owner agreed in advance to ignore. Week three is 1. Week four is 0, and the tripwire fires once on a promised date, handled by the office before the owner heard about it, which is the first real evidence the arrangement works.
By month three the monthly count is running about 1 in 12 jobs, so roughly 8.3% of jobs, about 4 to 5 delayed jobs a month against 55. That is a drop of about 4 percentage points of monthly jobs, or about 2 to 3 fewer delayed jobs each month than the baseline of about 7. The owner did not do anything to cause the improvement. The person now holding it started calling the supplier at 7 in the morning instead of at 4 in the afternoon, which is a change the owner would never have made because the owner was never free at 7.
The decay. Two consecutive months inside band, so the Friday line goes monthly and the owner stops opening it on Fridays at all.
The tasks where a number does not exist
Some work has no clean indicator inside a quarter. Customer relationship repair, training a new hire, anything cultural. Delegating those blind is how a shop finds out in month five that nobody was doing it.
Use a sample instead of a number. Pick a fixed small count of instances and review those, on a schedule, in full: three of this month's completed follow-up calls, listened to or read end to end. Two of the new hire's job write-ups. You are not measuring, you are calibrating, and the discipline is that you review the sample and nothing else. A sample of three reviewed properly teaches you more than a vague awareness of thirty, and unlike a status conversation it cannot expand to fill an hour.
How to tell the thread is working
Three checks, at the ninety-day mark:
- You can state the number from memory. If you cannot, you are not reading the line, and an unread line is not visibility, it is paperwork you imposed on someone else.
- A tripwire has fired at least once and you did not take the task back. A tripwire that has never fired in ninety days is either set too loose to be useful or the work is genuinely quiet. Check which by asking the person whether anything came close.
- The number of unscheduled questions about this task has dropped. Count them for one week. If you are still fielding four or five a week on something you handed off two months ago, the thread is fine and the authority was never actually transferred. That is a different repair, and there is a sibling article on it.
The failure that looks most like success: the number is green, the reports arrive, and you have quietly kept doing the hard half yourself. The tell is that the person holding the task always brings you the same category of decision. Look at your own last two weeks and count how many times you touched the task. If it is more than the reporting line plus the tripwires, you are still holding it and the paperwork is a costume.
References
- See related: Delegating a Task: The Clean Handoff
- See related: Delegating the Decision, Not Just the Task
- See related: The Check-Back Cadence for Delegated Work
- See related: The Owner Who Is the Bottleneck and Doesn't Know It
- Trade-standard practice for owner-operator delegation in small field-service businesses