Tight Five Matrix
What is most important? How do you know?
80 cents of your dollar is spent on the headline. The PromptDeck IS the headline.
Pitch to Prompt
Each slide has two layers. The pitch (headline + table) sells. The prompt (open-loop question) hooks. The Zeigarnik effect is the engine — the question stays active in working memory until the reader resolves it by clicking the depth link.
| Layer | What it does | Bias engine |
|---|---|---|
| Pitch (headline + table) | Confirms what they already feel | Confirmation, loss aversion |
| Prompt (question) | Opens a loop they can't close | Zeigarnik, hyperbolic discounting |
| Depth (SlideDepth link) | Offers resolution through co-creation | IKEA effect, reciprocity |
The prompt prop on Slide renders the open-loop question beneath the headline. Use it:
<Slide
number={1}
headline="Strong claim that compels belief."
prompt="The open-loop question they can't ignore?"
>
We all need a prompt in the right direction — human and phygital agent alike. A mantra prompts you. A system prompt prompts the machine. The PromptDeck prompts the room. Same mechanism. Different scale.
How to Use
Data-Driven
All slide data lives in one file: src/data/prompt-decks.json. Every deck on the site reads from it. Add your deck, import, render.
import { PromptDeck } from "@site/src/components/prompt-deck";
import promptDecks from "@site/src/data/prompt-decks.json";
<PromptDeck slides={promptDecks["your-slug"].journeys.customer} />;
For dual perspective (customer + feature dev):
<PromptDeck slides={promptDecks["your-slug"].journeys.customer} />;
{
/* perspective flip */
}
<PromptDeck slides={promptDecks["your-slug"].journeys.featureDev} />;
Data Contract
Each slide in prompt-decks.json follows this type:
| Field | Type | Purpose |
|---|---|---|
number | 1-5 | Position in the Tight Five |
headline | string | ≤7 words. Rhetoric, not labels |
label | string | Optional nav pill override |
prompt | string | Question that opens a loop |
picture | string | Vision description or image URL |
imagePrompt | string | Hidden AI generation prompt |
points | string[] | Max 5. Last often a question |
depth | {href, label} | Intelligent hyperlink to full depth |
biases | {name, href}[] | Behavioural biases loaded |
Each entry in the file has a journeys object:
| Journey | Slides | Arc |
|---|---|---|
customer | 5 | Pain → trust → method → proof → renewal |
featureDev | 5 | Job → stories → build → sprints → validation |
Single-deck entries omit featureDev. The flip between journeys is the pikorua — same person, different moment.
The depth.href is an intelligent hyperlink. Here it points to a static page. In the app it points to live data. Same contract, different pipe.
Slide Shape
Every slide follows this exact shape:
<Slide
number={1} // 1-5, maps to 5P position
headline="Strong rhetoric that compels belief." // The argument in one sentence
picture="Positive vision of the greater good" // Vision description until image exists
>
| Column A | Column B | Column C | | -------- | -------- | -------- | | Row 1 | | | | Row 2 | | |
<SlideDepth href="#depth">Depth: Position</SlideDepth>
</Slide>
For new decks, prefer the data-driven approach — it means one source of truth, and when the component improves, every deck upgrades.
Headlines
The headline is the most important element. 80 cents in the dollar. It must be strong rhetoric that people (and agents) feel compelled to believe in.
| Good | Bad | Why |
|---|---|---|
| "Fonterra owns the data, farmers don't." | "Player Analysis" | Strong claim earns attention |
| "Zero friction tracing scales judgment." | "Performance Metrics" | Vision of the greater good |
| "Discipline earns the compounding edge." | "Platform Overview" | Positive rhetoric that inspires action |
Pictures
The picture is not a placeholder illustration. You must paint positive pictures backed by strong rhetoric that people (and agents) feel compelled to believe in. Sell reasons to believe that good things can be earned through disciplined dedication to a greater good. Go positive, go first, be constant in doing it.
- Visual Art Prompts Starter
Tables
The table IS the argument. If it doesn't fit a table, rethink the argument.
| Rule | Why |
|---|---|
| Max 5 rows | Forces compression — cut until only the essential remains |
| 2-4 columns | More than 4 = too much for a slide |
| No prose in slides | Prose hides weak thinking — tables expose it |
| No bullet lists | Bullets are unstructured — tables force dimensions |
| One sentence per cell | If it takes more, you don't understand it yet |
Mistakes
| Mistake | Fix |
|---|---|
| Headline is a label ("Principles") | Make it a claim ("No shared nomenclature exists") |
| Table has 8+ rows | Cut to 5. Can't cut? Split into two slides |
| Prose between table and SlideDepth | Delete it. The table said everything |
| Bullet list instead of table | Find the dimensions, make columns |
title/mantra/questions props on PromptDeck | Remove — the page H1, opening line, and Questions section handle these |
| Missing SlideDepth | Every slide links to its depth page |
Context
- The Tight Five — The schema this instrument implements
- Headlines — 80 cents in the dollar
- Aligned Priorities — The card sort exercise
- Mantra Cascade — How phrases graduate to systems
- Create Your Deck — The interactive instrument
- Prompt Deck PRD — The product behind this page
Questions
If the table IS the argument, what arguments are you still making with prose?
- When you write a headline as a label instead of a claim, what are you afraid to commit to?
- Which implementation in the list above best demonstrates the instrument — and what makes it work where others don't?
- If the headline is 80 cents in the dollar, how much time do you spend on headlines vs table content?
- What's the difference between a slide that informs and one that creates urgency to click the depth link?