Find the work you should not be doing
Most owners cannot tell you where their week went. They can tell you it was full.
That is not a memory problem. It is a sorting problem. The work that fills a week arrives one message at a time, each one small enough to just handle, and none of it gets counted. By Friday you have done a hundred things and moved nothing.
This chapter is about counting. Before anyone builds anything, before you decide whether you even need a system, you need to know which parts of your week keep landing back on your desk and why.
Where the week actually goes
Dana runs a twelve-person plumbing company in Clovis. Four crews, one office manager, one dispatcher, and Dana, who does everything nobody else does.
On paper her job is estimates, hiring and the bank. In practice her Tuesday looks like this. Eleven texts from crews asking about parts. Three estimate requests that came in overnight and sat until she opened the inbox. Two customers calling because nobody confirmed the appointment. An invoice the office manager could not send because the job notes were in Dana’s head. Payroll questions. A supplier who needs a signature. And at four o’clock, finally, the estimate she meant to write at eight.
None of that is a crisis. All of it is hers, because somewhere along the way each of those things became “ask Dana.”
You will not find this pattern by thinking about it. You find it by writing down, for one ordinary week, every time something arrives that only you can answer. Not the big decisions. The small ones. The list is always longer than the owner expects, and it is always more repetitive.
The three kinds of work
Every item on that list falls into one of three buckets.
Work only you can do. Pricing a job nobody has seen before. Deciding whether to hire. Sitting across from an upset customer who is about to leave. Signing. This is the job. Nothing in this guide touches it.
Work someone should do. Answering the same parts question for the fourth time. Chasing a customer who has not paid. Turning a site visit into an estimate that looks like the last eighty estimates. This is judgment work, but the judgment is repeated. A good office manager does it. So does a good system, if it has been told exactly how you decide.
Work anyone could do. Confirming appointments. Filing the photo from the job into the right folder. Reminding a crew what is on tomorrow. Copying a number from one screen into another. This is the work that should already be gone, and in most businesses it is still on the owner’s desk because setting it up properly never made it to the top of the list.
The point of sorting is not to feel organised. It is to see that the first bucket is small and the other two are where the week went.
The "landed back on my desk" test
There is a faster way to find the second and third buckets than logging a week, and it works on the drive home.
Ask: what is the thing I have already handed off, that keeps coming back?
Dana handed dispatch to a dispatcher two years ago. Dispatch still comes back to her every time a crew is double-booked, because the calendar lives in one place and the job list lives in another, and the dispatcher cannot see both. Handing it off did not remove the work. It added a person who now asks her the question.
That is the signal. Work that keeps landing back on your desk after it was delegated is work where the process is broken, not the person. It comes back because the person doing it does not have what they need to finish it. Give them that, and it stays gone.
Make a short list. Three or four items is normal. If you have ten, pick the three that came back this week.
What is not worth touching yet
Not everything on the list should be fixed, and definitely not first.
Leave alone anything that happens less than once a week. The cost of setting up a system is roughly the same whether the work happens daily or monthly. Monthly work is not worth it yet.
Leave alone anything you are still changing your mind about. If you priced jobs one way in spring and another way in summer, a system will only lock in whichever version it learned. Decide first.
Leave alone anything that needs a real conversation. A customer who wants to renegotiate is not a workflow. Do not let anyone convince you an agent should handle it.
And be honest about the thing you secretly like doing. Some owners keep the estimates because estimating is the part of the trade they still enjoy. That is fine. Keep it on purpose, not by accident.
What is left is usually two or three pieces of work. Frequent, repeated, decided, boring. That is the shortlist, and it is short on purpose.
Common mistakes
Starting with the biggest problem. The biggest problem is usually the most tangled one. Start with the most frequent one. Wins that happen every day pay for themselves faster and teach you how to buy back the next piece.
Describing the job instead of the decision. “Follow up with customers” is a job. “If an estimate has not been accepted in three days, send a text asking if they have questions, and flag anyone who says no” is a decision. Systems run on decisions. If you cannot write the decision down, you are not ready to hand it off, to a person or to anything else.
Assuming it needs to be built. Half the time the thing that keeps landing back on your desk is already solved by a tool you own and have not set up. That is the point of a diagnosis, and it is what the next chapter is about.
Buying software to find out what you need. A subscription does not tell you where your week goes. A list does. Make the list first.
What comes next
You now have a shortlist: the two or three pieces of work that are frequent, repeated, already decided, and still on your desk.
The next question is not “what should I build.” It is “what is actually going wrong here.” Chapter II covers the diagnosis: what a good one tells you, why the answer is often a tool you already have, and how to know the difference before anyone writes a line of code.
Rather write than book? Tell me about your week.