Put a name and a date on it

In short: Before you build anything, write one sentence that says what "this worked" looks like in numbers, with a name and a date in it. Nobody skips this on purpose. Everybody skips it. Without it, day 30 becomes a vibe check, and vibe checks always pass. This chapter ends with a short template you fill in, and that filled-in template is your deliverable for this lesson.

The sentence

Here's the shape:

"This worked if [named person] spends [X] fewer hours a week on [task] by [date]."

Every bracket does a job. The name means one person owns it, not a department and not "we." The hours mean the result is countable. The task means it's tied to the workflow you circled in the last chapter and nothing else. The date means there's a day when you sit down and make a call.

If you run a department instead of a business, the sentence works the same way. The named person might be you, or it might be the coordinator who handles intake. The task might be status updates, ticket triage, or turning meeting notes into action items. Same shape.

The number you'll count

For the next thirty days there's one metric: drafts approved per week. How many times Claude wrote something, you looked at it, and you sent it or used it.

Not logins. Logins measure curiosity. Not "time saved," because you'll estimate that number, you'll estimate it generously, and it'll tell you what you already wanted to hear. Not adoption percentage, which is a metric for people who need to make a slide.

Drafts approved is countable, it's honest, and it's tied to real output that reached a real person. If it's zero, nothing happened. No amount of enthusiasm patches a zero.

Track it on paper. Tally marks. I'm serious. Don't build a dashboard for this. Building the dashboard is procrastination.

What good looks like: week one you're at a handful and each one takes real editing. Week two the number goes up and the editing goes down. That second part is the actual signal.

The question at day 30

At the end of the thirty days you ask the named person one question: what do you do less of now?

That's the whole measurement. If the answer is specific, "I stopped spending Tuesday morning on inquiry replies," it worked. If the answer is vague or enthusiastic-but-fuzzy, it didn't, no matter how many drafts got generated. Count what stopped.

Then you make one of three calls. It worked, so you ship it as the normal way that job gets done and go pick number two. It half worked, so you fix one thing, usually the context or the review habit, and run thirty more days on the same metric. It didn't work, so you kill it cleanly, cancel anything you subscribed to, say out loud that you're stopping, and diagnose whether it was the workflow, the context, or the habit. A clean kill is a good outcome. A pilot that limps for eight months is the bad one.

The appointment

One more thing goes in the template, and it's the least glamorous item in the whole course: a recurring calendar appointment. Weekly, twenty minutes, same day and time. Put it in now, before there's anything to review.

Here's why. The single mechanism that kills more of these projects than anything else is that the approver quietly stops approving around week three. Week one is great. By week three the review has become "looks fine." By week eight nobody opens the queue. Nothing announces it. The drafts keep generating and the quality degrades in silence.

Three things prevent it. A named person, not a role. A recurring appointment defended like a client meeting. A visible queue where you can tell at a glance whether anything's been reviewed this week. That's the entire adoption strategy. There's no rollout memo that substitutes for a person, a time, and a queue.

Why write it down

Everything in this chapter could live in your head. Don't let it. A goal in your head bends to match whatever happened. A goal on paper, dated, with a name on it, is something you can be wrong against, and being able to be wrong is the only thing that makes day 30 a real decision.

Fill in the template below. Rough is fine. You'll revise the workflow line after Lesson 2's scoring, and that's expected.

MY GOAL

Business, or the department I run: ______________________

The one workflow I'm automating first (provisional):
______________________________________________________

Who owns it (a name, not a role): _______________________

This worked if ______________ spends ____ fewer hours a
week on ______________________ by ____________ (date).

The number I'll count: drafts approved per week
  Week 1: ____   Week 2: ____   Week 3: ____   Week 4: ____

Weekly review appointment (day, time, 20 minutes):
______________________________________________________

Day 30 question: "What do you do less of now?"
Day 30 decision: ship / fix one thing / kill it cleanly

Do this now

1. Copy the template above and fill in every line. Rough is fine.
2. Put the weekly 20-minute review appointment on your calendar today, recurring, before there's anything to review.
3. Save the filled-in template as your Work Order for this lesson.