--- name: pitch-deck-storytelling description: "Drafts pre-seed and seed investor arguments as a causal slide chain from positioning through the ask. Use for fundraising deck teardowns, one-line positioning, why-now unlocks, or bottoms-up TAM stories. Never a Series A metrics pack, Figma/Gamma layout chair, or SAFE/term-sheet legal desk." version: 1.0.1 --- # Pitch Deck Storytelling A pitch deck is not a document — it is an argument delivered as a story, engineered to survive a 3-minute skim by a pattern-matching investor who has seen 4,000 decks this year. This skill turns a founder's raw facts into a narrative that an investor *cannot unsee*: a logical chain where each slide makes the next slide inevitable. The output is structure and story, not graphic design. ## When to Use ### Trigger conditions Activate this skill when the user asks for any of: - "Write / build / draft my pitch deck" or "help me with my seed deck / pre-seed deck." - "Critique / tear down / roast / review my deck" or "why isn't my deck landing?" - "Fix my problem slide / solution slide / market slide / traction slide / the ask." - "Write my one-liner / positioning statement / elevator pitch / company tagline." - "How big is my market / build my TAM / SAM / SOM" in a fundraising context. - "What's my 'why now'?" or "how do I show momentum with little traction?" - "How much should I raise / what's my use of funds / what milestones do I pitch?" - "Turn these notes / this product into an investor narrative." - "Make my deck more compelling / tell a better story / sharpen the arc." - Any mention of: *Sequoia template, YC deck, a16z, demo day deck, investor update narrative, data room story, founder-market fit.* ### When NOT to use (boundary cases) - **Series A and beyond** — the burden of proof shifts from *narrative* to *metrics* (cohort retention, magic number, net revenue retention, sales efficiency). Use a growth-metrics/board-deck skill instead; this skill is calibrated for pre-seed/seed where story carries more weight than numbers. - **Pure visual/graphic design** — color, typography, slide layout, Figma/Pitch/Gamma execution. This skill defines *what each slide must say and why*; hand off the pixels to a design tool or skill. - **Financial model construction** — detailed 5-year P&L, cap table mechanics, 409A. This skill writes the *unit-economics narrative*, not the spreadsheet. Hand off to a financial-modeling skill. - **Legal fundraising mechanics** — SAFE vs. priced round, term sheets, valuation negotiation. Touch only the *amount and use of funds*; route the rest to a startup-legal skill. - **Non-fundraising storytelling** — sales decks, all-hands, product launch narratives. Different audience, different incentives. ## Prerequisites ### Narrative inputs to gather before drafting Before writing any slide, collect (or ask the founder for) the following **narrative inputs**: 1. **Stage and amount being raised** (pre-seed vs. seed; target dollar amount). 2. **What the product actually does** — the core mechanism, not the vision. 3. **Who the customer is and what they do today** — the specific persona and their current workaround. 4. **Any usage/revenue data** — real numbers only; never invent. 5. **The team's unfair advantage** — founder-market fit, proprietary insight, rare skills. 6. **The "why now" unlock** — the concrete recent change that makes this possible *now*. > **HARD RULE:** If three or more inputs are missing, ask for them. Do not invent traction or fabricate metrics. Investors triangulate during diligence; a fabricated metric is fatal. ### 2026 core principles (apply as non-negotiable defaults) The seed game has fundamentally repriced since the 2023–2024 generative-AI surge. Deviate from these only with an explicit reason. 1. **"AI-native" is table stakes, not a thesis.** Claiming you "use AI" differentiates nothing in 2026. Investors are saturated with thin GPT wrappers. The narrative must answer: *what is defensible once the underlying models commoditize?* Lead with the **moat** (proprietary data flywheel, distribution, workflow lock-in, a regulated wedge), and treat the model as plumbing. 2. **The traction bar has risen because building got cheap.** AI-assisted development means a solo founder can ship a working product in weeks. A demo alone no longer impresses — it's expected. Pre-seed now wants a live product plus *early signal* (engaged users, design partners, LOIs, a sliver of revenue). Seed increasingly wants *evidence of pull*: real retention, a repeatable acquisition motion, or six-figure ARR in many categories. Calibrate the traction slide to "momentum and pull," not "we built a thing." 3. **Capital efficiency is the dominant virtue again.** The ZIRP "grow at all costs" era is over. The 2026 hero narrative is **"a tiny team doing what used to take 30 people,"** powered by AI leverage. Show a low burn, a long runway, and a clear story for *how few people you need*. "We'll hire 40 engineers" is now a red flag, not ambition. 4. **"Why now" must cite a real unlock.** The strongest 2026 decks anchor timing to a *specific, recent capability or behavior shift* — a model crossing a usefulness threshold, an inference-cost collapse, a regulatory door opening, a platform shift, or a newly-formed buyer habit. "The market is growing" is not a why-now. "This was technically impossible 18 months ago and is now 10x cheaper" is. 5. **AI gross margins are under the microscope.** Inference is a real, variable COGS. Investors now probe: *what's your gross margin after inference costs, and does it improve or degrade as you scale?* Decks that hand-wave software's "90% margins" without accounting for token/compute costs read as naïve. Address margin structure proactively. 6. **Show, don't tell — with proof, not adjectives.** Replace claims with artifacts: a 60–90 second Loom or live demo of the magic moment, evals/benchmarks for AI quality, a retention curve, a real customer quote with a name and title. "Investors fund momentum they can see," and in 2026 a credible 90-second demo beats ten bullet points. 7. **Two decks, two jobs.** The **send deck** (read alone, in an inbox, in 3 minutes) must be self-explanatory and skimmable. The **pitch deck** (presented live) is sparser and leans on the founder's voice. Always clarify which you're building; default to a send-deck that can also be presented. 8. **Bottoms-up market math only.** "1% of a $4T market" is an instant credibility kill. Build TAM from units × price × frequency. Investors don't fund percentages of giant numbers — they fund a *credible path* from a beachhead to a large market. ## Procedure The deck is a **causal chain**. Each slide must earn the right to the next. The 2026 canonical order for the send deck: | # | Slide | The one job of this slide | Investor's silent question it answers | |---|-------|---------------------------|----------------------------------------| | 1 | Title / Vision | Position the company in one line | "What is this, in 5 words?" | | 2 | Problem | Make the pain visceral and expensive | "Does anyone actually care?" | | 3 | Solution | Show the magic moment | "Is the fix real and obvious?" | | 4 | Why Now | Name the unlock | "Why didn't this exist already?" | | 5 | Market / TAM | Prove the prize is big *and* reachable | "Can this be a fund-returner?" | | 6 | Product | Show depth and the moat | "What stops a competitor copying it?" | | 7 | Business Model | Show how money is made and kept | "Do the economics work?" | | 8 | Traction | Show pull and momentum | "Is it working already?" | | 9 | Competition | Show you'll win the wedge | "Why you and not them?" | | 10 | Team | Show founder-market fit | "Are these the people to bet on?" | | 11 | The Ask | Amount, use, milestones | "What does my money buy?" | The seven steps below map this chain. Steps fold the competition and product slides into adjacent steps. ### Step 1 — Hook & Vision **Goal:** In the first 10 seconds, make the investor lean in. The title slide is positioning, not branding. 1. Write a **one-line positioning statement** the investor could repeat to a partner. Use one of these proven structures: ``` A) Category-defining: "[Company] is the [new category] for [who]." → "Cursor is the AI code editor for professional engineers." B) Familiar-anchor: "[Recognizable thing] for [new domain]." → "Stripe for carbon credits." C) Outcome-first: "We help [who] [achieve outcome] without [old pain]." → "We help clinics cut no-shows 40% without hiring schedulers." D) Inevitability: "[Future state] is inevitable. We're building the [layer] for it." ``` 2. **Avoid** vague taglines ("Empowering the future of work"). They signal nothing and burn the most valuable slide. 3. Pair the line with a **vision sentence**: the world that exists if you win. Big enough to be venture-scale, concrete enough to be believable. *"Every small business has a 24/7 AI operations team"* — not *"transforming commerce."* 4. **Cold-open option:** open on the problem with a single shocking stat, then reveal the company. Use when the problem is non-obvious and the stat is genuinely arresting. **Output of this step:** one positioning line + one vision sentence + (optional) a one-stat cold open. ### Step 2 — Problem Deconstruction **Goal:** Make the investor *feel* the pain and believe it's a painkiller, not a vitamin. 1. Name the **specific protagonist** (a role, not "businesses"): *"A 3-person clinic front desk,"* *"a mid-market RevOps lead."* Specificity reads as real customer knowledge. 2. Describe the **status-quo workaround** — the spreadsheet, the offshore team, the duct-taped tools. The existence of an ugly workaround *proves* the pain is real and budget already flows to it. 3. **Quantify the cost** of the problem in money, hours, or risk: *"Each missed appointment costs $200; the average clinic loses $4,000/month."* A quantified problem makes the solution's value self-evident. 4. Apply the **"hair on fire" test:** is this an urgent, top-3 pain, or a nice-to-have? If it's a vitamin, the deck will fail later at traction — flag this honestly to the founder rather than dressing it up. ``` PROBLEM SLIDE TEMPLATE - WHO: [specific role / persona] - TODAY: [the broken status-quo workaround they use] - COST: [quantified pain — $, hours, churn, risk per period] - WHY BROKEN: [the structural reason the status quo can't fix it] ``` > **WARNING:** If the problem slide needs three paragraphs to explain why the pain matters, the pain probably isn't acute. Acute problems are stated in one sentence and felt immediately. Push the founder toward a sharper wedge rather than a longer explanation. ### Step 3 — Solution & Product Demo Strategy **Goal:** Show the **magic moment** — the single interaction where the user's pain evaporates — and make the solution feel like the obvious answer to Step 2. 1. Lead with the **demo**, not a feature list. In 2026 the default is a 60–90 second Loom or a live walkthrough that shows input → magic → outcome. Specify in the narrative *exactly which moment to film*: the one where a skeptic goes "oh." 2. State the solution as the **inverse of the problem.** If the problem was "clinics lose $4k/month to no-shows," the solution line is "we recover that revenue automatically." Mirror the structure so the logic snaps shut. 3. For AI products, **prove quality with evals/benchmarks**, not adjectives: *"94% intent-classification accuracy vs. 71% for the incumbent, measured on 10k real tickets."* Claims like "highly accurate" are discounted to zero. 4. **Product depth slide (folds in here):** show the one or two things that are *hard to copy* — the proprietary dataset, the workflow you own end-to-end, the integration depth. This plants the moat before the competition slide. 5. Keep features subordinate to outcomes. List at most three capabilities, each tied to a customer result. ``` SOLUTION ARC 1. "Here's the magic moment." → [demo / the single aha interaction] 2. "Here's why it works." → [the core insight or proprietary mechanism] 3. "Here's the proof it's good." → [eval, benchmark, or before/after metric] 4. "Here's why it's hard to copy." → [data flywheel / distribution / lock-in] ``` ### Step 4 — Market Dynamics & TAM Narrative **Goal:** Prove the prize is fund-returner-large *and* that you have a credible path to it. This is where most decks lose credibility. 1. **Build TAM bottoms-up.** Never "X% of a $T market." ``` BOTTOMS-UP TAM TAM = (# of target customers) × (annual contract value) × (realistic frequency) Worked example (clinic scheduling): 210,000 target clinics in market × $6,000 ACV/year = $1.26B TAM SAM = clinics in beachhead geography/segment you can actually sell to today SOM = what you can win in 3 years given your motion (be honest; investors discount this anyway) ``` 2. Present the **beachhead-to-expansion path**: the narrow wedge you win first, then the adjacent expansions. Investors fund a credible *sequence*, not a vague giant number. *"Start with dental no-shows → all outpatient scheduling → full front-office automation."* 3. **"Why now" lives here or as its own slide (Slide 4).** Name the concrete unlock: a model capability crossing a threshold, inference cost collapsing, a regulation opening, a behavior shift. Tie market timing to that unlock so the size feels *newly* capturable. | TAM red flag (reject) | 2026 credible reframe | |------------------------|------------------------| | "1% of the $5T healthcare market" | "210k clinics × $6k ACV = $1.26B, starting with dental" | | "Everyone is our customer" | "Our wedge is 3-to-10-seat clinics with no IT staff" | | "Market growing 30% CAGR (top-down)" | "Why now: voice models crossed human-parity on scheduling calls in 2025" | | Static TAM with no expansion path | "Beachhead → adjacent vertical → platform" sequence | ### Step 5 — Business Model & Unit Economics **Goal:** Show how the company makes money, *keeps* it, and gets more efficient with scale. 1. State **pricing and packaging** in one line: per-seat, usage, platform fee, % of value created. Tie price to the quantified pain from Step 2 (*"we charge $500/month to recover $4,000/month"* — an obvious ROI). 2. Address **gross margin honestly, including inference COGS.** The 2026 investor will ask. State margin after compute and whether it improves with scale (caching, smaller fine-tuned models, batching). Pre-empt the margin-compression worry. 3. Show **early unit economics directionally**, not precisely, at seed: rough CAC, expected payback, expansion potential. Label estimates as estimates — fabricated precision is worse than honest ranges. 4. Reinforce the **capital-efficiency narrative** (Core Principle 3): low burn, AI-leveraged team, a long runway on a modest raise. ``` UNIT-ECONOMICS NARRATIVE TEMPLATE - PRICE: [model] at [$amount], anchored to [customer ROI] - GROSS MARGIN: [%] after inference/compute; [improves/holds] at scale because [reason] - CAC / PAYBACK: ~[$] to acquire, ~[months] to recoup (early estimate) - EXPANSION: net revenue grows via [upsell / seats / usage] - BURN: [$]/month, [N] months runway on this raise ``` ### Step 6 — Traction & Milestones **Goal:** Convert whatever evidence exists into **momentum and pull.** Growth *rate* and *engagement* beat absolute size at seed. 1. **Lead with the strongest real signal**, in this rough priority: revenue growth > retention/engagement curve > paying design partners > signed LOIs > active-usage growth > waitlist. Pick the one true thing and make it the headline. 2. **Plot momentum, not a snapshot.** A "up and to the right" chart of weekly active users or revenue over the last few months beats a single number. Slope is the story. 3. **Use named proof.** A quote with a real name and title (*"— Dr. Patel, Smile Dental"*) outweighs ten anonymous logos. 4. If traction is thin, **reframe honestly around velocity and learning**: *"Built and shipped in 6 weeks; 3 design partners live; 40% week-over-week usage growth from a small base."* Never invent numbers — investors triangulate, and a fabricated metric is fatal in diligence. 5. **The milestone plan (bridges to The Ask):** state the *specific, measurable* goals this round buys, framed as the proof points needed to raise the *next* round. ``` TRACTION HEADLINE PATTERNS - Revenue: "$[X]k ARR, growing [Y]% MoM for [N] months" - Engagement: "[%] of users return weekly; [usage] per active user" - Partners: "[N] design partners live, [N] paying" - Velocity: "0 → [milestone] in [N] weeks with a team of [N]" ``` > **WARNING:** Vanity metrics (total signups, social followers, press mentions) without engagement *underneath* them actively hurt — sophisticated investors read them as a dodge. If retention is bad, lead with something else and be ready to explain the retention honestly when asked. ### Step 7 — Team & The Ask **Goal:** Prove *these specific people* are unusually suited to win this, then make a crisp, justified ask. 1. **Team = founder-market fit, not résumés.** Answer "why you?": the years lived in the problem, the proprietary insight, the rare combination of skills. *"I ran scheduling for 30 clinics for 8 years — I've felt this pain every day"* beats a logo soup of past employers. 2. For AI-native teams, signal **leverage and shipping velocity**: a tiny team that ships fast is itself the thesis (Core Principle 3). 3. **Competition slide (folds in here or just before Team):** never claim "no competition" (reads as "no market"). Use a positioning frame (2×2 on the two axes that matter, or a feature/wedge comparison) that shows *why you win the beachhead* — usually a sharper wedge, proprietary data, or distribution, not "more features." 4. **The Ask** must be specific and self-justifying: ``` THE ASK TEMPLATE - RAISING: $[amount] [instrument: SAFE / priced] (2026 norms: pre-seed ~$0.5–3M, seed ~$2–6M) - USE OF FUNDS: [%] product, [%] GTM, [%] team — as a small number of concrete hires/bets - MILESTONES: this round gets us to [specific metric] — the bar to raise our [next round] - RUNWAY: [N] months ``` 5. Tie the ask back to the milestone plan from Step 6: *"This $2M buys 18 months to reach $1M ARR and 70% logo retention — the proof points for a Series A."* An ask without milestones reads as "we need money"; an ask *with* them reads as a plan. ## Pitfalls | Pitfall | Symptom in the deck | Mitigation protocol | |---------|--------------------|----------------------| | No clear problem | Problem slide is abstract, needs paragraphs to justify | Force the one-sentence "hair on fire" statement; quantify the cost; name a specific persona (Step 2) | | Boring / flat narrative | Slides are a feature list; no arc; investor skims and drops | Rebuild as a causal chain — each slide must make the next inevitable; cold-open on the sharpest stat | | Unrealistic TAM | "1% of $X trillion"; top-down math | Replace with bottoms-up units × price × frequency; add beachhead→expansion sequence (Step 4) | | Weak / hand-waved moat | "Our AI is better"; no defensibility once models commoditize | Name the durable moat: data flywheel, distribution, workflow lock-in, regulated wedge (Core Principle 1) | | "AI" as the whole thesis | Deck leans on the model, not the business | Demote the model to plumbing; lead with customer outcome + moat | | No "why now" | Timing unexplained; "why didn't this exist?" goes unanswered | Anchor to a concrete recent unlock — capability, cost, regulation, behavior (Core Principle 4) | | Thin traction over-inflated | Vanity metrics; or fabricated numbers | Reframe around velocity + the single best real signal; never invent metrics — diligence will catch it | | Margin naïveté | "90% software margins" with no mention of inference cost | Address gross margin after compute and its trajectory at scale (Step 5, Core Principle 5) | | "No competitors" | Empty or dismissive competition slide | Reframe: positioning 2×2 showing why you win the *wedge*; "no competition" implies "no market" | | Vague ask | "Raising to grow"; no use of funds or milestones | Specify amount, allocation, and the next-round proof points it buys (Step 7) | | Too long / dense | 25+ slides; walls of text | Cut to ~11 core slides; push detail to an appendix; one idea per slide | | Wrong deck for the channel | A live-pitch deck emailed cold, or vice versa | Clarify send-deck vs. pitch-deck; default to a self-explanatory send-deck (Core Principle 7) | ### Edge cases - **Pre-revenue, pre-product (idea stage):** Shift weight to *team + why-now + market insight*. The narrative becomes "uniquely-qualified team + a timing unlock no one else sees." Do not fake traction; sell the *wedge and the velocity plan*. - **Deep tech / long R&D:** Traction is *technical milestones* and *expert validation/design partners*, not revenue. The arc emphasizes the moat (hard tech) and a credible commercialization sequence. - **Founder insists on a buzzword-heavy narrative:** Push back once, clearly, with the reason ("investors discount 'revolutionary AI platform' to zero — specificity is what reads as real"). If they still insist, comply but flag the risk in your summary. ## Verification Before declaring a deck or narrative complete, confirm every item. Any "no" is a blocker — fix it or flag it explicitly to the founder. ### Narrative integrity - [ ] The one-line positioning is repeatable from memory and names a category, anchor, or outcome — not a vague tagline. - [ ] Each slide makes the next slide feel inevitable (the causal chain holds end to end). - [ ] A stranger could read the send-deck alone in 3 minutes and explain the company back correctly. ### Problem & solution - [ ] The problem is stated in one sentence, tied to a specific persona, and quantified in $/hours/risk. - [ ] The solution mirrors the problem and identifies one explicit magic moment (with a demo plan). - [ ] AI quality is backed by an eval/benchmark or before-after metric, not adjectives. ### Market & timing - [ ] TAM is built bottoms-up (units × price × frequency), with a beachhead→expansion sequence. - [ ] "Why now" names a concrete recent unlock, not "the market is growing." ### Economics & defensibility - [ ] Pricing is tied to quantified customer ROI. - [ ] Gross margin is addressed *after* inference/compute cost, with a scale trajectory. - [ ] A durable moat is named that survives model commoditization (data, distribution, lock-in, or regulation). ### Traction & ask - [ ] The single strongest *real* signal headlines the traction slide; no fabricated or vanity-only metrics. - [ ] Momentum is shown as a slope (growth over time), not a static number. - [ ] The ask states amount, use of funds, runway, and the specific milestones it buys toward the next round. ### Team & format - [ ] The team slide answers "why you specifically" (founder-market fit), not just résumés. - [ ] The competition slide shows *why you win the wedge*; it never claims "no competition." - [ ] The deck is ~11 core slides (detail pushed to an appendix), one idea per slide. - [ ] Send-deck vs. live-pitch intent is confirmed, and the deck matches that channel. ### Final gut-check — the "so what" pass Read the deck and after every slide ask "so what?" If any slide doesn't visibly advance the argument toward "this is a fundable, fund-returner-scale bet," cut it or rewrite it. Investors fund a *clear, inevitable-feeling story* — momentum they can see and a logic they can't unsee.