# Will Larson Suggests How to Define a Good Goal
## Metadata
**Status**:: #x
**Zettel**:: #zettel/literature
**Created**:: [[2022-10-08]]
**Topic**:: [[♯ Management]]
**Book**:: [[Calibre 394 - Will Larson - An Elegant Puzzle Systems of Engineering Management|Will Larson - An Elegant Puzzle: Systems of Engineering Management]]
**Parent**:: [[Will Larson - An Elegant Puzzle (Highlights)]]
## Notes
### 4 Kinds of Numbers
A bad goal is indistinguishable from numbers. A good goal has 4 kinds of numbers:
- A target states what we want to reach.
- A baseline is the current state.
- The trend is how quickly the state changes now.
- The time sets up the bound on the goal.
A bad goal example: Limit the CI error rate under 5%.
A good goal: Limit the CI error rate under 5% in Q4. Now the error rate is 10%. In Q4, it has increased 3%.
### Investment and Baseline
Investments are goals that we want to improve. Baseline are states we want to preseve.
Without baseline goals, we can complete the goal *Reducing CI wait time by half* by doubling the CI machines. A baseline goal to avoid this looks like: CI expanses should not increase.
Baseline goals are restricts to help us to clip the solution space and prioritize different directions.