An engine produces the report. Feed it the same inputs and you get the same output every time, and every judgment it makes is a rule written out below for you to check. I read each report before it ships. None of this is investment advice — it's structural and arithmetic checks on how your own story and numbers should be presented.
Most funded pitches run on one of a handful of story arcs. The arc, not a fixed slide list, decides which slide leads and where the burden of proof sits. The engine scores five arcs against your stage, model, metrics, and the why-now and wedge you wrote, then picks the winner. Here's the rubric:
| Arc | Wins when… | Leads with |
|---|---|---|
| Category creation | a sharp why-now is articulated, AI/new-approach language, revenue still small | why-now |
| Efficiency play | language centers on a cheaper/faster known job; healthy margins & real revenue | quantified before/after |
| Platform | NDR ≥110% and real ARR at a later stage — expansion is already earned | beachhead traction |
| Picks-and-shovels | positioned as infrastructure under a recognizable boom; design partners across apps | the boom + your position |
| Wedge → platform | the early-stage default: a stated wedge, pre-seed/seed, no stronger signal | a painfully specific problem |
Each signal carries an explicit point value; the report shows the exact reasons the winning arc scored what it did, and the runner-up as a fallback. Arc definitions are distilled from public startup-narrative craft (YC "How to Pitch", Sequoia's business-plan template, a16z/NFX market-narrative writing).
Here is the full point table — every signal, every arc, the exact points it adds or subtracts. This is the rubric behind the score; nothing else moves it.
| Arc | Signal | Points |
|---|---|---|
| Category creation | AI / "new way" / "reimagine" / "first" language in your text — a first-pass signal the human read confirms or overrides, never the final word | +2 |
| a why-now field is actually filled in (not left blank or a stub) — the presence of a stated enabling shift, which I then read on its merits | +2 | |
| a regulatory shift / mandate / compliance change is cited | +2 | |
| ARR > $3M (the category likely already exists) | −2 | |
| stage is Series B (must be won, not pitched) | −2 | |
| Efficiency play | "cheaper / faster / automate / replace / manual / spreadsheet" language | +2 |
| gross margin ≥ 60% | +1 | |
| CAC payback ≤ 18 months | +1 | |
| ARR > $500K | +1 | |
| model is B2B SaaS or fintech | +1 | |
| Platform | NDR ≥ 110% | +3 |
| NDR < 100% (platform story premature) | −2 | |
| logo retention ≥ 85% | +1 | |
| Series A/B and ARR ≥ $2M | +2 | |
| "platform / suite / multi-product / expand / adjacent" language | +1 | |
| stage is pre-seed (no beachhead yet) | −3 | |
| Picks-and-shovels | "infrastructure / API / tooling / middleware / SDK / rails" language | +2 |
| sits under a recognizable boom (AI, crypto, robotics, climate, EV) | +1 | |
| design partners ≥ 3 across applications | +1 | |
| model is deep tech or hardware | +1 | |
| Wedge → platform | baseline (the reliable early-stage default) | +1 |
| stage is pre-seed or seed | +2 | |
| a wedge field is actually filled in (not blank or a stub) — the presence of a stated wedge, read on its merits by the human reviewer | +1 | |
| ARR > $3M (may have earned a platform arc) | −1 |
Highest total wins; the second-highest is your runner-up. Ties break by table order. The report reproduces the exact lines that fired for your inputs — this table is the whole engine, not a summary of it. The language-based signals (e.g. AI / "reimagine" wording) are deliberately a coarse first pass to nominate a likely arc; they are not a judgment of whether your why-now is credible. That judgment is the human read on top, which can and does override a keyword-driven score before your report ships — so the mechanic isn't gameable by dropping in a buzzword.
The slide set is the canonical funded-deck sequence — title, problem, why-now, wedge, product, market, traction, model, GTM, team, ask. For each, we ship the single question it must answer, what to show, the diligence lens (what an investor actually checks here), and the failure mode. The arc supplies the ORDER; load-bearing slides are flagged. Per-slide attention estimates use DocSend deck-analytics findings on where investors spend a ~3m44s first read.
For your stage and model, the engine decides which metrics to lead with, support with, or leave off. Lead with the wrong one and you invite the wrong question. It then scores your actual numbers against a deliberately conservative "clears the bar / strong" band. The bands are the ranges US venture works from, 2024–2026:
| Metric | Seed bar (clears · strong) | Series A bar (clears · strong) | Lead / support |
|---|---|---|---|
| ARR | $100K · $1M | $1.5M · $3M | lead (seed+) |
| MoM MRR growth | 10% · 20% | 7% · 15% | lead at seed |
| YoY growth (×) | 2× · 3× (AI: 3× · 4×) | 2× · 3× (AI: 3× · 4×) | lead at A/B |
| Gross margin | 50% · 70% (AI: 40% · 52%) | 65% · 75% (AI: 45% · 55%) | lead at A/B |
| NDR | omitted (not table-stakes) | 100% · 110% | lead at A/B |
| Logo retention | 70% · 85% | 80% · 90% | support |
| Burn multiple (lower better) | ≤3× · ≤2× | ≤2× · ≤1× | lead at A/B |
| CAC payback (lower better) | context only | ≤18mo · ≤12mo | support at A |
| GMV (marketplace) | $1M · $10M | $10M · $50M | lead |
| Take rate (marketplace) | 5% · 15% | 8% · 20% | lead |
| Design partners / LOIs | 3 · 8 | omitted | lead at pre-seed |
| MAU (consumer) | 1K · 50K | 50K · 500K | lead |
A number at or above "strong" reads 🟢, at or above "clears" reads 🟡, below "clears" reads 🔴; "omitted" means the metric hurts more than it helps at that stage and the engine leaves it off. Pre-seed and Series B bars follow the same structure and ship in the report.
Model overrides apply where the standard differs — e.g. vertical-AI gross margin bars sit ~20 points below software-grade, reflecting inference cost. Verify against the source reports before quoting to an investor; markets move and these are deliberately conservative "bar to clear" bands, not top-decile vanity numbers.
Before a partner reads a deck, they run a few checks in their head. The engine runs the same ones on paper, and each flag shows the exact number that tripped it:
Round size vs. stage. Your ask compared to the US median round-size band for your stage. More than 2× the top of band trips a red flag.
Implied dilution. If priced at the stage median post-money P, an ask A implies A / (P + A) dilution. Outside a healthy 10–25% band, the ask is mispriced for the stage.
Valuation vs. traction. Stage median post-money ÷ your ARR gives an implied revenue multiple; well above the stage's sanity ceiling means a partner will re-anchor down without an AI/hyper-growth story.
Runway to the next bar. A round should buy ~18–24 months to reach the next round's ARR bar. Under 12 months of reported runway flags — you'll re-raise before you've generated the proof.
Burn-multiple coherence. At Series A/B, a burn multiple well above the stage guide flags in the current efficiency environment.
Round-size, post-money and dilution norms are US medians from Carta State of Private Markets (Q4 2025 / Q1 2026) and PitchBook-NVCA Venture Monitor (2025). These are wide arithmetic sanity bands — not a valuation opinion, and not a prediction of whether you'll raise.
No report ships until it passes a structural check: exactly 11 unique slides in the arc's order, at least two load-bearing slides, an ask analysis that isn't empty, and metric rules that apply to the stage and model. Then I read every one before it goes out.