Two businesses left the same 4-hour session with the same enthusiasm. One of them, a 12-person insurance agency, had a written page: intake summaries, owned by the office manager, current time measured at 22 minutes each, decision on August 14. Thirty days later intake summaries took 6 minutes and they were arguing about what to do next.
The other had a good conversation and a set of notes. Thirty days later they had a good conversation and a set of notes.
The page is not paperwork. It is the difference between a month of trying things and a month that ends in a decision.
The four things a plan names
Before the weeks, the plan states 4 facts, and each one is a single line.
One workflow. Not a category, a workflow. “Turning inspection notes into the customer summary,” not “documentation.”
One owner. A person who does this work now, by name. Not a committee, and not you unless you’re the one doing the task.
One number, measured before you start. How long it takes today, timed across 5 real instances. Guessing at the baseline is how a project ends with everyone unsure whether it helped.
One decision date. Thirty days out, on the calendar, with keep, fix, or stop as the only 3 possible answers.
Week 1: measure and gather
Nobody builds anything. The owner times 5 real instances of the task and saves the inputs and outputs from each. If you already have 20 good past examples in a folder, that’s better than anything you’ll write from scratch this month.
Also pick the tool this week and settle the boring part: who pays for it, whether it’s a business account, and what is not allowed to be pasted into it. Ten minutes now, no scramble later.
Week 2: build the draft and run it beside the old way
The owner writes a first prompt, feeds it 3 of the saved examples, and runs the real task both ways. Slower this week than not doing it at all. That’s expected.
By Friday you want the prompt written down somewhere shared, with the review step attached: the 2 or 3 specific things somebody checks before the output leaves the building.
Week 3: hand it to somebody else
This is the week most plans skip and it is the one that tells you whether you built anything. Give the written prompt to a second person and have them use it without coaching.
If they can’t get a usable result, you don’t have a workflow yet, you have one person’s knack. Fix the instructions, not the person.
Week 4: measure again and decide
Time 5 more real instances. Compare to the Week 1 number honestly, including the review time, because review is part of the job now.
Then decide on the date you set. Keep it and write down the hours it returns each week. Fix it if the gap is one specific fixable thing, and give it 2 more weeks with a new date. Stop it if the honest number is close to where you started, and say so out loud. A stopped workflow after 30 days costs you a month. An undecided one costs you the year.
Write the page first
You can do this before you buy anything or call anyone. One page, 4 lines at the top, 4 weeks underneath. If you can’t fill in the number on line 3, that’s your first week’s work, and it’s the most valuable week of the four.