You walk into a marketing meeting. Someone is presenting a 60-slide deck. The first 57 slides are titled with various functional labels — Strategic Overview, Market Analysis, Competitive Landscape, Product Architecture. The last three slides? Nobody has built them yet. We haven't gotten to those. Matt Dicks: That's where we start, though. We want to land somewhere — otherwise we're just collecting slides and changing the order three days before the event, and then two days before the event, and then two hours before the event we're going to say something or we're not going to say something.1
The procedural discipline is the inversion of the natural order. Most people build deck → ending. Dicks's principle: build ending → deck. Build story-end → story. Build talk-conclusion → talk. In storytelling, regardless of what the story is, it's always: what am I trying to say at the end?2 The ending is not the place you arrive at after working through the material. The ending is the place you start, and the rest of the material is constructed to deliver you there.
Start at the end is a procedural discipline with three properties:
1. Identify the destination before drafting anything else. Before writing the opening, before identifying the middle scenes, before considering structure — name where the piece is supposed to land. In oral storytelling, this is the 5-second moment of change. In a keynote, it's the single transformation the audience should experience. In a marketing deck, it's the specific commitment or insight the deck should produce. We start at the end because we want to know what we're going to say.3
2. Refuse to start drafting until the end is concrete. Dicks describes the typical client meeting: Often times they say, "Well, we're trying to say a bunch of things." And I say, "Well, if you say a bunch of things, you're not saying anything. You have to have something that you're aiming at at the end."4 Vagueness about the end produces vague pieces. The destination has to be one specific thing the storyteller can name in a sentence, not a list of three or four candidate goals.
3. Once the end is set, work backwards. What does the audience need to experience just before the end for the end to land? What scene leads into that? What scene leads into that? Each previous scene is engineered to produce the audience-state that makes the next scene work. The forward-flow of the eventual piece is the audience's experience; the construction-flow is the backwards engineering that produced it.
This is not the same as outlining. Many beginner writers think they outline. What they actually do is forward sketching — write down their starting point, then list scenes that might happen next, then keep going until they reach an ending. The piece's destination emerges from the sketching process. Dicks's procedure inverts this: name the destination first, then engineer scenes that lead to it.
Forward drafting produces sequences. The writer puts down events in the order they happened, and the resulting piece is a chronological recounting. Dicks's binary classifies these as anecdotes (story vs anecdote) — sequences without inversion-shape.
Reverse engineering produces inversions. By starting at the destination, the writer is forced to identify what kind of moment the destination is — a realization, a transformation, an interior shift — and the rest of the piece becomes the scaffold that supports the shift. The shape is built into the construction process. The writer cannot drift away from inversion-shape because the destination is anchoring everything.
This is why Dicks treats the reversal discipline as procedural rather than stylistic. Style can be added or removed; procedure can't. A writer who skips the start-at-the-end step is operating from a different procedural footing than a writer who applies it, and the resulting pieces will have different structural shapes regardless of how good the sentences are.
The discipline applies beyond storytelling. Dicks notes that he applies it equally to marketing decks: I always say what are the last three slides? They said, "Well, we don't know — we haven't gotten to those." And I said, "That's where we start though."5 The same procedural discipline that produces good stories produces good business presentations, good keynotes, good policy documents. The class of pieces is wider than storytelling; the unifying property is change-shape — pieces whose function is to move the audience from one state to another. Anywhere you're trying to produce audience-shift, start at the end.
Start at the end is the procedural form of the Big Three Step 1. The Big Three is the meta-framework (three steps, fixed order); start-at-the-end is Step 1 considered in its operational discipline.
Start-at-the-end makes the 5-second moment of change locatable. Without committing to find the destination first, the 5-second moment remains an abstract category. With the procedural discipline, it becomes the actual entry point to drafting.
The procedure feeds beginning as opposite of ending: once the end is set, the beginning is derivable (the opposite of the end). Without the end, beginning construction is guesswork. With it, beginning construction is reverse-symmetry.
The procedural discipline also makes the no-one-wants-to-hear mindset operationally usable. Knowing the destination lets you engineer every sentence backward from there to maximize each one's contribution to audience-arrival. Without the destination, you're not earning attention sentence-by-sentence — you're hoping the audience stays through wandering.
Dicks demonstrates the procedure in real-time during the Knowledge Project interview when Shane Parrish asks him to construct a story.6 The construction-flow he describes runs backward from the destination:
Step 1 — Locate the destination. I knew this was a moment, I felt it. My son is getting ready to go to Scout camp, and part of scout camp is the swim test. He passed on the second try. He told me, "I heard you in my head, Dad, and I swam the other two laps." The 5-second moment is the realization that the story Dicks told his son traveled with him into the moment that mattered. Destination identified.
Step 2 — Identify what the beginning needs to be. Given the destination (the lesson I gave my son carried him), the beginning must establish the opposite (a moment when I refused the same lesson and it cost me). Dicks reaches into his own past — the 1983 swim test where he gave up after 50 yards and ended up classified as a beginner swimmer, stuck in a rowboat for two days. Beginning identified, and crucially, identified because the destination required it.
Step 3 — Compress the timeline backwards. The destination is 2024. The beginning is 1983. Raw chronological distance: 41 years. The piece needs to span both moments without being a 41-year story. Dicks's solution (BABC structure) is engineered to serve the destination — start in the present (son's trunk, packing for camp), flash back briefly to 1983, return to the present for the pickup-and-reveal. The structure is selected because it serves the destination, not because it's a generic best practice.
This is the procedural discipline in operation. Dicks could not have constructed the piece by starting in 1983 and working forward — he would not have known which 1983 details to include, how to compress the intervening years, where to land in 2024. The destination is what enabled the structural choices.
Saturday morning. You're going to write a new piece. You're going to apply the start-at-the-end discipline from scratch.
You sit at the kitchen table with a notebook (paper, not screen). At the top of the page you write: END. You give yourself ten minutes to identify the destination. You don't try to write the piece. You don't draft any other content. You sit with the question: what is the 5-second moment of this story? What is the audience supposed to experience right before I stop talking?
You write candidate destinations. Most of them are too vague — I realized something about family doesn't pass the test. You revise. I realized my mother had given me a single sentence I'd been hearing in my head for twenty years without knowing it. That passes — specific, locatable, contains an actual interior shift.
Now you flip the page. You write BEGINNING. You give yourself five minutes. The beginning is derivable from the end — it must be the opposite. What's the opposite of "I realized she'd given me the sentence twenty years ago"? The opposite is a moment when I assumed she had given me nothing. You scan your memory for scenes that have this shape. You pick one: the day after her funeral, looking at her empty closet and feeling the lack of inheritance she had to give.
Now MIDDLE. You give yourself ten minutes. The middle is the bridge between the two states. You sketch three or four scenes that move from she gave me nothing to she gave me the sentence twenty years ago. The scenes are not random; each one is engineered to produce a shift the audience needs to feel to reach the destination.
Total time: 25 minutes. Output: a structural skeleton built backwards from a known destination. The actual writing of sentences can now begin — but with the structure already in place, the writing is much faster and the resulting piece holds together because every scene is doing destination-arrival work.
You start drafting without naming the destination. You sit down and start writing the opening. You're operating in forward-sketch mode, not reverse-engineering. Stop. Go back to the destination-naming step before continuing.
Your destination is too vague to anchor structure. Something about my mom is not a destination. I realized my mother had given me a single sentence I'd been hearing in my head for twenty years is. Specific. Locatable. Anchorable.
You name a destination but don't engineer scenes backward from it. You found the destination and then resumed forward-drafting, hoping the scenes would naturally arrive. The procedural discipline requires that scenes be selected based on destination-arrival service. Without that selection, the destination is decorative.
Your destination changes during drafting. As you draft, you discover a different destination than the one you started with. This is not necessarily a failure — discovery-drafting can produce better destinations than initial-naming. But if it happens, stop and restart the procedure with the new destination as the anchor. Don't drift between candidate destinations.
You apply the discipline only to short pieces. The principle scales to keynotes (start with the takeaway), marketing decks (start with the last three slides), book chapters (start with the chapter's structural function), and even full books (start with the central argument or transformation the book is supposed to produce). Limiting it to short forms wastes the discipline.
The start-at-the-end discipline is named explicitly in both Dicks transcripts as Step 1 of the Big Three.7 It is consistent with how working storytellers across the Moth-StorySLAM community describe their crafting process. The discipline also appears in Dicks's coaching work with marketing teams (last three slides first), confirming it transfers across narrative and non-narrative formats.
A real tension exists with Karlsson's animating-question method: Karlsson's essay-writing process explicitly does not start at the end. Karlsson begins with an animating-question whose answer he doesn't know, and the writing is the discovery of the answer. The discovery process is the essay. The split reflects medium-deep differences: oral storytelling is a delivery form where the destination must be known and managed; essay writing is a discovery form where the destination emerges through writing. Both methodologies are right for their forms.
A tension also exists with Macfarlane's relational nature-writing, which builds long-form nature pieces through years of with/by presence rather than through destination-first construction. The Big Three / start-at-the-end procedure would distort Macfarlane's work by forcing it into change-over-time inversion structure the work doesn't naturally have.
Open question: do practitioners need to be told which of their pieces require start-at-the-end vs which allow discovery-drafting? The two procedures produce different artifact-types and confusing them produces bad work in either direction. A forward-drafted speech fails because it has no destination; a backward-engineered essay often feels mechanical because the discovery is missing. The decision rule (when to apply which procedure) deserves more articulation than current Dicks teaching gives it.
Dicks and Bilton converge cleanly on destination-first construction. Bilton's Tetris database research method for narrative non-fiction requires the reporter to know the arc and shape of the story before collecting material — the database is organized retrospectively, not chronologically. Same discipline at radically different scales (60,000-word book vs 5-minute Moth piece). Both writers reject forward-sketching as a working method for change-shaped pieces.
Dicks and Karlsson are in productive opposition. Karlsson explicitly inverts the Dicks procedure — animating-question first, destination discovered through writing. The opposition reveals two legitimate epistemologies of authored work. Some content is discoverable only through the act of writing (Karlsson's essays); some content is destroyed by attempting discovery-drafting (Dicks's stage pieces). The form constrains the procedure. Which procedure you apply depends on whether your form allows in-process discovery.
Start at the end is fundamentally a reverse-engineering procedure — the discipline of working backwards from a known target through the choices required to reach it. That move has direct analogues across disciplines.
History/strategy: Strategic planning as backwards construction — military and business strategy literature has a long tradition of backwards planning (start with the desired end-state and work backwards to identify required moves). Dicks's procedure is the craft-level instantiation of the same epistemic move at the scale of single artifacts. What neither domain produces alone: strategy supplies the philosophical framing (forward construction is path-dependent and rarely reaches optimal end-states); Dicks supplies the teachable, granular, single-artifact procedure that converts the strategic principle into daily-practice discipline. Strategy operates at scales of years and millions of dollars; Dicks operates at the scale of one 5-minute piece. The structural insight is identical — the same epistemic discipline that works at military-campaign scale works at single-story scale. Practitioners trained in either can recognize the other.
Behavioral mechanics: Six-Minute X-Ray (Hughes) — Hughes's elicitation work begins with the target disclosure the elicitor is trying to obtain, then works backwards through the rapport-building and conversational moves required to produce that disclosure. The mechanism is structurally identical to Dicks's start-at-the-end: identify the destination (target disclosure), work backwards through scenes (conversational moves) that produce the audience-state (target receptivity) needed at each step. What neither domain produces alone: Dicks teaches the storyteller to reverse-engineer for audience transformation (cooperative); Hughes teaches the elicitor to reverse-engineer for target disclosure (adversarial). The cross-orientation parallel shows that reverse-engineering from the change-point is a meta-skill across live-craft disciplines. Once you have the skill in one domain, it transfers to others with minor retraining.
The Sharpest Implication Start at the end implies that most public communication is procedurally upside-down. People sit down to write speeches, decks, articles, and stories starting with their opening lines and proceeding forward, with the destination emerging or not emerging through the process. The procedural discipline inverts this: name the destination, work backward. Most pieces produced under the forward-procedure are anecdote-shaped, slide-collection-shaped, or wandering. Most pieces produced under the reverse-procedure are change-shaped, focused, and effective. The reversal is small (10-25 minutes of pre-drafting work). The cumulative effect across a career is enormous. The implication: the difference between competent and excellent communicators is often just procedural discipline, not native talent.
Generative Questions