Cheap to Make, Costly to Check · Volume One
Tools
Eleven tools, twenty-six handholds, and where each one is built. The worksheets are in Part Five of the book.
The eleven tools
| Tool | Built in |
|---|---|
| The inventory — what is actually running | Chapter 2 |
| The three failure modes | Chapter 2 |
| The five triggers — does this need a rule at all | Chapter 8 |
| The four zones — how strictly handled | Chapter 8 |
| The constitutional-rank test | Chapter 9 |
| The pattern review | Chapter 9 |
| The duty to challenge | Chapter 9 |
| The pace rule | Chapter 11 |
| The coupling points | Chapter 11 |
| Configuration power | Chapter 13 |
| Unassisted practice | Chapter 14 |
Twenty-six things to do
A handhold produces a finding. A commitment sets something that stands afterward. None of them is a project.
| Ch. | Form | What to do | Time |
|---|---|---|---|
| 1 | Handhold | One pair of numbers: items per week at a checkpoint against minutes for a serious check | two hours |
| 2 | Handhold | The list of your systems, sorted into three layers — count before you explain | an afternoon |
| 3 | Handhold | The same pair of numbers on your longest-running AI use | fifteen minutes |
| 4 | Handhold | Three things that were true about your last significant decision and appear in no file | twenty minutes |
| 5 | Handhold | Mark where a standard tone would be noticed | half an hour |
| 5 | Handhold | What work did you learn to judge on — and who gets that practice now? | ten minutes |
| 6 | Handhold | Two phone calls: who stands behind the last thing you relied on, and for what exactly | fifteen minutes |
| 7 | Handhold | Write down who signed the maintenance agreement and who installed the last update | fifteen minutes |
| 7 | Handhold | Run the last AI decision through all five perspectives | — |
| 7 | Handhold | Approvals issued last week against those where you saw the material yourself | ten minutes |
| 8 | Commitment | Apply the five triggers before every new use and every new rule | — |
| 9 | Handhold | The pattern review: who was invited, who was not, what separates the groups | half an hour, quarterly |
| 11 | Handhold | Assign each process one of the four paces, with a date and a name | a morning |
| 11 | Handhold | Measure one coupling point instead of estimating — three numbers | two hours |
| 12 | Handhold | The load check: four questions, with the people affected | half a day |
| 12 | Handhold | Not only what a tool saves — what happens to the time it saved | — |
| 13 | Handhold | Write a name behind every pre-sorting system | one hour |
| 13 | Commitment | Fix your own position before looking at the system; log dissenting views | ten minutes |
| 14 | Commitment | Set a recurring date on the calendar at which work happens without the system | — |
| 14 | Handhold | Mark ten decisions where your information advantage actually decided | — |
| 15 | Handhold | Write the actual pace next to each zone | — |
| 15 | Handhold | Count the empty fields where no name stands for the configuration | twenty minutes |
| 16 | Handhold | Hold a reviewed decision against the five conditions and cross off whatever fails | — |
| 16 | Handhold | Write the question you would ask a colleague — then look for who can answer it | twenty minutes |
| 16 | Commitment | Write the one sentence on update review, and ask for two numbers | — |
| 16 | Handhold | Look for a review hours figure in your last business case | — |
Where no time is given, the chapter gives none — usually because it depends on the size of the organization.