The Iterative Process
ReShelf's team now has a framed problem and a wall of ideas. The dangerous move is to pick the most exciting one and spend three months building it. The safer and faster move is to go round a loop many times, spending a little on each lap.
The rule
Discover, select, validate, execute — then loop back with what you learned.
- Discovery: find problems worth solving and gather ideas for them, using research, data and the idea techniques from this section.
- Selection: choose which ideas to pursue with explicit criteria rather than enthusiasm. A quick score is enough at this stage.
- Validation: test the riskiest assumption as cheaply as possible before building. A clickable prototype, a "coming soon" button that counts taps, or doing the job by hand for a few users all work.
- Execution: build and ship the smallest version that tests the idea for real, then measure. The results flow straight back into discovery.
Each lap reduces uncertainty. Early laps should be cheap and quick, because that is when you know least.
A quick selection score: rate impact, confidence and ease from 1 to 5, then multiply them.
| Idea | Impact | Confidence | Ease | Score |
|---|---|---|---|---|
| Wanted list (buyers post requests) | 4 | 3 | 4 | 48 |
| Barcode scan to list | 3 | 4 | 3 | 36 |
| Library drop-off shelf | 5 | 2 | 1 | 10 |
The score starts a conversation rather than ending one, but here it points clearly at the wanted list.
The shortcut
At any moment ask: "Which stage are we in, and what is the cheapest step to the next one?" The costliest mistakes come from skipping validation, jumping from "we chose it" straight to "we built it".
Discovery
Interviews show first-years often can't find a book that seniors own but never listed.
Drills
Answer fast — trust the rule.
What is the correct order of the four stages?
Meera matches wanted-book requests by hand for a week before any code is written. Which stage is this?
Idea A scores impact 4, confidence 2, ease 5. Idea B scores impact 3, confidence 4, ease 4. Which wins on impact × confidence × ease?
Which is the most common costly mistake in this loop?
Remember
- Discover, select, validate, execute, then loop back with what you learned.
- Ask which stage you are in and what the cheapest step to the next one is.
- Trap: skipping validation and paying for a full build to learn what a one-week test could have shown.