The Story Engine
Mode 3, finally specified · what makes a story a story
Universal tool
The Content Engine names three LinkedIn modes and gives two of them machinery. Mode 3 — the personal moment — has been a label with nothing underneath it. This is the machinery. Not everything called a story is one, and the difference is testable in about ten seconds.
The four-part test and the three-part shape adapt Kindra Hall, Stories That Stick. The refusal list and the expectation-gap engine adapt McKee & Gerace, Storynomics. Both are translated into your language with your examples — the source vocabulary is not used. Credited as inspiration, not reproduced.
01
The Five Impostors — what isn't a story
Start here, because most "our story" pages die on this block. Five things get called stories constantly and none of them are. They're recitations: this happened, then this, then this. The tell is simple — if you can reorder the parts and nothing breaks, you don't have a story. You have a list wearing a nicer word.
A process"First we scope, then we build, then we ship." That's a workflow. Nothing is at stake and nobody changes.
An org chartWho reports to whom. Order imposed on chaos, which is useful — and tells you nothing.
A chronologyFounded in 2019. Series A in 2021. Ten thousand users in 2023. Dates are not events.
A journeyThe softest word in business writing. A journey has a passenger. A story has someone who wants something and can't get it.
A narrativeEvery story is a narrative. Almost no narrative is a story. Reaching for the wider word is usually how you avoid doing the work.
The Five Impostors
Refuse these first
Five recitations that get called stories. None of them are.
The reorder test: shuffle the parts. If nothing breaks, it isn't a story.
Yours: this is a refusal list, and it works the way
the Humanizer's tells work — name the failure so you can see it in your own draft.
02
The Four Legs — the ten-second test
Four things have to be present or it won't hold weight. Hold any draft against them and find the missing leg. Most business writing that calls itself a story is standing on one. The one people skip most is the moment — a real point in time, not a period of time.
Leg 01
A person
Someone specific enough to stand next to. Not "teams" or "users." One person with a job and a problem.
Who is this happening to?Leg 02
A feeling
Felt, not labelled. "Frustrated" is a label. "Reading the same error for the fourth time" is a feeling.
What did it feel like?Leg 03
A moment
A point in time, not a period. Not "over several months." A Tuesday. A specific call. One sentence someone said.
When exactly? Where exactly?Leg 04
A detail
Too specific to have been invented. The odd, small, useless-sounding thing. That's what makes the rest believable.
What's the detail nobody would make up?
✓ Four legs
A founder I was on a call with muted himself, came back, and said his agent had just invented a customer that didn't exist. He'd already sent them an invoice.
✕ No legs
Many organisations struggle with AI reliability, which can lead to significant downstream business consequences.
The Four Legs
Person, feeling, moment, detail
All four present or it doesn't hold. Find the missing leg.
A moment is a point, not a period. "Over six months" is not a moment.
The detail nobody would invent is what makes the rest believable.
Yours: diagnostic, not decoration — same shape as
the Humanizer. Name the leg, fix the leg.
03
Floor, Break, New Floor
The shape. Three parts, and the weighting is the whole lesson: the floor gets the most room and the break gets the least. That's the opposite of what everyone does. The named failure here is the Missing Floor — opening on the event. It's the single most common reason a true story lands flat.
Floormost of the wordsHow things stood before. This is where the four legs get installed and the reader is handed a reason to care.He'd shipped the agent on Friday and spent the weekend telling people it was the best thing he'd built.
Breakfewest wordsThe thing that happened. One sentence is usually enough. If the floor is built, the break needs no help.Monday, it invoiced a customer who didn't exist.
New Floorthe meaningHow things stand now, and what it means. This is where the story earns its place — it points somewhere.He didn't have a model problem. He had a verification problem, and he'd never written one down.
Roughly the proportions to aim for. Everyone gets this backwards and leads with the break.
The Missing Floor. Opening on the event. "So my agent invoiced a fake customer —" lands on a reader who has no idea who you are, what you'd built, or why it mattered. No contrast, no stakes, no story. If a draft feels flat and you can't say why, check the floor first. It's almost always the floor.
Floor · Break · New Floor
The shape, weighted
Set the ground, break it, show where the ground is now.
The floor gets the most room. The break gets the least.
The New Floor is what makes it worth telling. Without it you told an anecdote.
Yours: the Missing Floor is the named failure. Flat draft → check the floor.
04
The Miss — the actual engine
Under the shape there's an engine, and it isn't the event. It's the gap between what someone expected and what they got. That's why the four legs matter — the legs exist to build the expectation, so there's something for the miss to violate. Not the event. The miss.
What he expected
Ship it, watch it work, tell everyone.
the miss
What actually happened
It confidently billed a customer who was never real.
✓ Has an engine
The expectation is stated, then broken. The reader feels the drop.
✕ No engine
A sequence of true events with nothing expected and nothing violated. Reads as "and then, and then, and then."
The Miss
Expectation minus result
The gap is the engine. The event is only where it shows.
No stated expectation → no miss → no story. Check this before you check anything else.
The floor exists to build the expectation. That's what it's for.
Yours: Not the event. The miss. — correction fragment, straight into the voiceprint.
05
The long engine — when three parts won't hold it
Floor, Break, New Floor is the default and covers most of what you'll write. When a piece has to carry more — a keynote open, a founder page, a long Substack essay — open the shape into five beats. The one that matters is Choose: the decision made with information the person didn't have before. That beat is where the meaning actually lives.
1Want — what they were after. Name it plainly, or nothing that follows has weight.
2Act — the first move they made to get it. Reasonable. The kind of move anyone would make.
3Miss — what came back instead. The world doesn't cooperate.
4Choose — the decision made with what they now know. The meaning lives here.
5Land — where it settles, and what it costs. Declarative. Never a question.
Want · Act · Miss · Choose · Land
The long engine
Five beats for pieces the three-part shape can't hold.
Choose is the load-bearing beat. If there's no decision, there's no meaning.
Land is declarative. Same rule as the six-beat verdict — never close on a question.
Yours: default to three parts. Don't spend five beats where three will do.
06
The four jobs — which story for which task
A story is chosen against a job, the same way a structure is. Four jobs cover almost everything you'll need. The one worth the most to you right now is the Witness, because it's the only one that produces evidence rather than mechanism — which is exactly the gap the Writer audit flagged.
| When you need to… | Tell | What it does |
|---|
| Make an offer feel real | The Sell | One person's before and after. Sits inside Problem → Promise → Proof → Price as the proof. |
| Explain why you exist | The Origin | The about page, the bio, the intro. Buys trust — never mistake it for a sales story. |
| Align people on an idea | The Banner | The shared reference everyone can repeat. Close cousin of the Mode-1 reframe. |
| Prove it actually works | The Witness | Their story, their words, their details. The top rung of the proof ladder. |
The Origin is not the Sell. The most common mix-up. An origin story that pivots into an offer stops being an origin story and starts being an ad the reader didn't agree to. Let it do its own job and stop.
The four jobs
Sell, Origin, Banner, Witness
The task picks the story. Same discipline as the structure picker.
The Witness is the only one that produces evidence. The rest produce mechanism.
Yours: go get Witnesses. That's the standing audit finding, and no page here can substitute for it.
07
The Vault — fill it, then cast from it
Nobody writes a good story from a blank page under deadline. You write it from an inventory. Two phases, and blending them is what kills the whole thing: fill without judging, then cast against a job. If you feel like you have no stories, that's a retrieval problem, not a supply problem.
Phase 1
Fill — no judgement
- One line per entry. A title, not a draft. "The one where the agent invoiced a ghost."
- Never ask "is this good enough." That question empties the vault.
- Prompts that actually work: the first time it went wrong · the thing a client said that stuck · what you believed two years ago and don't now · the moment you knew the common advice was wrong · the smallest fix that changed everything
- Add to it weekly. It compounds exactly like the hook bank.
Phase 2
Cast — against a job
- Name the job first: Sell, Origin, Banner, or Witness.
- Scan the vault for entries that fit that job. Pick one.
- Run the Four Legs. Missing a leg? Either fill it truthfully or cast a different entry.
- Never bend a story to fit a job. Change the story, not the truth.
Witness entries aren't yours to remember. You can't fill that shelf from your own memory — you have to ask for them. The Deep Ask on the Conversation page is the mechanic for getting a real story out of someone who's convinced they don't have one.
The Vault
Fill, then cast
An inventory, not an act of inspiration.
Judging while collecting is the failure. Separate the phases completely.
Cast against the job, then test the legs. In that order.
Yours: same philosophy as 70-20-10 — build a bank, let winners compound, never gamble on one.
08
The whole thing, running
A Mode-3 post with the parts tagged, so you can see the weighting in actual prose instead of in the abstract. Note the floor takes over half of it. Note the close is a verdict, not a question. The voice rules don't get suspended because it's a story.
FloorA founder I work with shipped his first real agent on a Friday. He'd been building it for five weeks. He spent the whole weekend telling people about it — his co-founder, his brother, me, twice. He kept saying the same line: it just handles the whole thing now.
FloorHe wasn't wrong to be excited. It did handle the whole thing. Intake, matching, drafting, sending.
BreakMonday morning it sent an invoice to a customer who didn't exist.
New FloorHere's what he told me after: he'd tested whether it could do the job. He'd never tested whether it could tell when it hadn't. Those are different tests, and only one of them was written down.
New FloorMost people building agents right now are running the first test and calling it done. Capability is easy to check. Knowing when the thing is wrong is the part nobody specs.
New FloorVerification isn't a feature you add later. It's the half of the build you skipped.
09
What this page can't do
Every page in this playbook gets an honest caveat. Here's this one's, and it's a real limit rather than a modest-sounding flourish.
The limit
This engine can tell you whether a draft is a story and exactly where it breaks. It cannot give you stories you haven't lived.
The Four Legs will find the missing leg. The Miss will tell you why a true account still reads flat. The Impostors will kill a chronology before you waste a morning on it. None of that produces raw material. The Vault is the actual work, it takes months rather than an afternoon, and it's yours — a playbook can't fill it for you. Same finding as the Writer audit, in a different domain: the system has limits, and pretending otherwise would be the exact thing this brand exists to argue against.