Skip to main content
All issues
The Sagentix Signal · Issue #4· Pitch Deck

Ninety-five decks in a hundred never reach due diligence.

Issue #4 · Pitch Deck

2026-09-03·By Stéphane Raby
Download PDF
Listen to this newsletter0:00 / 13:51

Your deck will not be audited. It will be eliminated. Those are two different jobs, and most founders build for the wrong one.

Gompers, Gornall, Kaplan and Strebulaev surveyed 885 institutional venture capitalists at 681 firms and reconstructed the deal funnel from the inside (Gompers et al., 2020). For every deal a firm closes, it considers roughly 101 opportunities. It meets management at 28 of them. Ten reach a partners' meeting. Due diligence starts on 4.8. Term sheets go out on 1.7. Which means about 96 of those 101 are gone before a single reference call is made — ninety-five in a hundred, before anyone opens a data room.

For every deal closed, 101 considered and 4.8 taken into diligence

  1. Considered

    101 · opportunities per closed deal

  2. Management met

    28

  3. Reviewed with partners

    10

  4. Due diligence begun

    4.8

  5. Term sheet offered

    1.7

  6. Closed

    1

Band widths rank the stages; they are not drawn to scale, because 1 in 101 is a hairline. The printed numbers are the finding (Gompers et al., 2020).

Averages from the 442 respondents who answered the funnel questions, of 885 surveyed (Gompers et al., 2020). Sector matters: an IT firm considers 151 opportunities per investment, a healthcare firm 78.

So the sharper question is not which of my slides survive an audit? Almost none of them will ever face one. It is: which slide is going to get me eliminated, and what would have to be on it to stop that?

A fatal flaw ends it in stage one

Maxwell, Jeffrey and Lévesque coded 150 recorded entrepreneur–investor interactions and tested the shape of the decision rather than assuming it. Angel investors do not weigh and score a long list of attributes. They use "a shortcut decision making heuristic known as elimination-by-aspects to reduce the available investment opportunities to a more manageable size" (Maxwell et al., 2011). One disqualifying finding ends the assessment: "If an opportunity is diagnosed with a fatal flaw, it is rejected in the first stage of the decision making process."

They name the eight grounds on which that flaw gets found: product adoption, product status, protectability, customer engagement, route to market, market potential, relevant experience, and financial model.

Two caveats belong on the record. Those 150 interactions were recorded from the Canadian edition of Dragons' Den, so they are angels on television rather than partners in a boardroom. And the paper is fifteen years old. It still matters, because Gompers' funnel — a different method, a different decade, institutional capital rather than angels — describes the same shape from the other end.

So what? A slide is not an opportunity to impress. It is the place where one of eight doors gets closed. A slide that closes none of them has spent a minute of a thirty-minute meeting and bought nothing.

How much evidence you need depends on how much you have

Here the advice most founders receive is not merely vague — it is backwards half the time.

Nyilasy, Yi, Herhausen, Ludwig and Dahl analysed every written proposal sent to the Creative Destruction Lab between 2012 and 2019 — 5,334 of them — against the panel's actual acceptance decisions (Nyilasy et al., 2025). They separated the steak (capital raised, patents, founder track record, marquee partnerships) from the sizzle (high-arousal words, and concrete, tangible specifics). The finding is an interaction, not a rule: passionate language pays only when hard evidence is already there, and concrete detail persuades only when it is not.

Get that pairing wrong and it costs you. Writing the HBR summary of their work, the authors report that amped-up passion deployed over thin resources "reduces acceptance odds by roughly thirteen percentage points," while aligning rhetoric to resources "can swing acceptance odds from only two percent to almost 35%" (Nyilasy et al., 2025b). Investors read the mismatch as lipstick, not conviction.

Passion pays only on top of evidence; without it, cool and concrete does

Under-sold

the steak is on the plate and the language does not say so

Passion pays

high-arousal language lifts acceptance only when the evidence is already there

Concrete detail persuades

the right register when you are resource-light: checkable specifics

Read as lipstick

passion over thin resources costs roughly thirteen points of acceptance odds

Every written proposal to the Creative Destruction Lab, 2012 to 2019, against the panel's own decisions (Nyilasy et al., 2025).

A second finding tells you which register to write in. Falchetti, Cattani and Ferriani ran four experiments and found that framing has to match the audience's expertise. Novices — their examples are lay people and crowdfunders — "appreciate more novel ideas framed in abstract why terms," while experts, meaning professional investors and innovation managers, want them "framed in concrete how terms" (Falchetti et al., 2022). Start with why is good advice for a conference audience and the wrong instruction for the people who write cheques.

What's in it for you: Audit your four capital buckets first — financial, social, human, intellectual. If you cannot point to one sizable cheque, one name-brand partner, one domain win or one defensible patent, you are resource-light, and your setting is cool, concrete and checkable. If you can, stop over-specifying next quarter's MRR and paint the horizon instead.

The market slide is not the slide they weigh

Ask the investors themselves what drives selection and the market is not the answer. The management team was named as an important factor by 95% of the firms and as the most important factor by 47% of the VCs. Business-related factors were mentioned often — business model 83%, product 74%, market 68%, industry 31% — but the whole bundle was ranked most important by only 37%. Valuation and the firm's own ability to add value were each mentioned by about half, and rated most important by under 3% (Gompers et al., 2020).

The market is the fourth factor VC firms name, not the first

  • Management team95%the most important factor for 47%
  • Business model83%
  • Product74%
  • Market68%
  • Industry31%
Share of VC firms naming each factor as important in selecting a deal (Gompers et al., 2020).

Our own primary research points the same way. A Canadian seed-stage investor managing two funds walked us through his five-minute filter and wrote ten items on a whiteboard in the order he works through them. The first five: team, technology, value proposition, market, competition — market the only word he put in capitals, and still the fourth thing he reaches for (Canadian seed-stage venture capitalist, personal communication, April 2026). His own screen rate is one investment for every hundred to two hundred companies seen. That range contains Gompers' 101, which corroborates the shape of the funnel and not the value of any number in it.

That does not make the market slide optional. It changes what it is for. A market slide is read as evidence about the team — as proof that these people know which buyers they can reach. OpenVC names the failure directly, listing "taking a top-down approach by quoting random studies found online" among the mistakes that sink the slide (Nasser, 2025). That it then reads as a team problem rather than a market one is my reading of the two findings together, not theirs.

The same four slides, before and after a claims audit

What the deck says

What survives reconstruction

  • $250B global market, 1% capture, $2.5B TAMMarket: NAICS-coded industry revenue, per-vertical fit, TAM to SAM to SOM
  • 100% YoY growthTraction: ARR $1.8M to $3.6M across 47 new logos, 38 of them outbound in three named verticals
  • A 2x2 with us in the upper rightCompetition: the three alternatives that appeared in our last 10 lost deals, and what the buyer chose
  • AI is hotWhy now: a dated inflection that did not exist 24 months ago and will not exist in 24 more
Illustrative deck language, not measured findings. Nothing on the left is false; it is simply unreconstructable, which at the elimination stage lands in the same place (Sagentix Cross-Engagement Benchmark, 2026).

Implication: For every factual claim on every slide, name three things — the source, how the number was built, and the date it was last refreshed. If you cannot name all three, cut the claim or hold it until the evidence exists (Sagentix 18-Check Quality Gate, 2026).

Across the decks we built this year the same architecture keeps arriving: a fifteen-to-sixteen-slide spine, six or seven audience versions built by swapping a minority of slides off a fixed core, and a claims grade on every slide before any of it reaches a designer. One recent build graded twenty-nine slide-level claims — twenty-three fully supported, five needing softened wording, one with no evidence at all — and the single unsupported claim carried a named owner, a deadline and the exact fallback sentence to use until it closed (Sagentix Cross-Engagement Benchmark, 2026). Writing the fallback before the deadline is what stops a founder inventing a number at 11pm the night before a partner meeting, which is where most unsourced claims are actually born.

Do this this week: take your market, traction and why-now slides, and against each one write which of the eight rejection grounds it closes. Any slide that closes none of them is decoration, and it is costing you a minute you do not have.

The Second Signal

Investors are moved by what they cannot audit

There is a finding in this literature that every evidence-first argument has to face honestly, including this one.

Tsay ran twelve studies with 1,855 participants using footage and real outcomes from live pitch competitions. Across all of them, "silent videos—but not sound recordings, video-with-sound recordings, or pitch transcriptions—best allowed both experts and novices" to identify which pitches the investors had actually picked (Tsay, 2021). Strip the words entirely and prediction improves. Worse, the paper reports that people "do not seem to fully recognize how much visual information factors into their decisions."

Hu and Ma reached the same place with machine learning over full pitch videos, and then added the part that matters: "positive (i.e., passionate, warm) pitches increase funding probability. However, conditional on funding, startups with higher levels of pitch positivity underperform" — and the mechanism, tested experimentally, is that delivery works "mainly through leading investors to form inaccurate beliefs" (Hu & Ma, 2025).

Read those two together and the conclusion is not presentation beats substance. It is sharper than that. Delivery predicts who wins the room. It does not predict who was right. An investor persuaded by warmth funds you and is later disappointed; the belief that got you the cheque was inaccurate, and the correction arrives at the board meeting instead of the pitch.

Bound it properly: Tsay's setting is pitch competitions, and Hu and Ma's is video pitches — neither is a Series B partner meeting. And the field is less settled than it looks. McSweeney and colleagues reviewed 173 articles on entrepreneurial pitching published between 2000 and 2024 and found that "prior research remains highly fragmented leaving us with a lack of conceptual clarity" about what pitching even is (McSweeney et al., 2025).

So rehearse the delivery — and never let it carry a claim. Coach preparedness, not passion.

From the Field

The same evidence gap runs through every go-to-market artifact we touched this cycle.

  • The Series A Board Deck Has an Evidence Gap — and It's Killing Founder Credibility. The blog's worked example is the 18-month post-Series-A board meeting where credibility leaks — a $50B TAM pulled from an analyst summary, a 2x2 with the company conveniently in the upper right, a forecast that needs a 38% conversion lift (Sagentix Phase 01 Market Intelligence, 2026). It is the elimination problem after the money has already arrived — the same eight doors, reopened by people who now sit on your board. Read it if your market slide has never been rebuilt from the industry codes up.

  • "Add More Visuals" Has No Support in the Research That Studied It. Directly relevant to anyone rebuilding a deck. Tversky's team found that "only carefully designed and appropriate graphics" help; figure count measures nothing. Cleveland and McGill ordered encodings by reading accuracy — position and length near the top, area and angle below. Given how much of an investor's judgment runs on what they see, a chart drawn against its own number is an expensive error. Read it before your designer starts.

  • Your TAM Slide Won't Survive Due Diligence. Here's How to Fix It.. Three patterns kill TAM credibility: top-down hand-waving, missing source attribution, and conflating TAM with SAM (Sagentix Phase 01 Market Intelligence, 2026). The fix is a bottom-up build with explicit SAM filters — the form OpenVC also recommends (Nasser, 2025). Read it before your next rewrite; it will save you a fact-check cycle with your lead investor.

  • B2B SaaS Took Tens of Billions in VC Funding. Most Founders Are Still Guessing Their GTM.. The paradox: software stayed the largest category of U.S. venture investment across every quarter of 2025 (NVCA & PitchBook, 2025), yet most funded decks rest on four unchecked guesses — pricing, sales process, value proposition, TAM. Each is one of the eight grounds an investor can eliminate you on. Read it if you are within six months of your next raise.

That discipline is being tested in a place founders are not watching yet.

What We're Watching

The fastest-moving signal is not the funding environment. It is what generative AI has done to the top of the funnel.

Davenport and Noyes interviewed seven venture capitalists and found AI now present as a core capability or a feature in virtually every technology investment they consider — and, more usefully, present in the pitches themselves (Davenport & Noyes, 2025). Vivjan Myrto of Hyperplane Ventures told them "noise is at an all-time high." Dan Von Kohorn of Broom Ventures described the quality distribution as barbelled: "a dramatic increase in low quality pitches, and also a sizable increase in the quality of the top pitches." Their practical warning to founders is the one worth pinning above your desk — partners have the same tools you do, so surface-level research on your own market will not impress anyone.

That is the same pattern as the deal funnel, in a different arena. When production quality stops separating decks, the only thing left that separates them is what survives a check. And the downside is asymmetric: profiling of legal hallucinations found the four models it tested — ChatGPT 4, ChatGPT 3.5, PaLM 2 and Llama 2 — fabricated between 58% and 88% of answers to direct, verifiable questions about randomly selected federal court cases (Dahl et al., 2024). One fabricated source in a deck is a fatal flaw under exactly the heuristic Maxwell's team described — and it gets found at stage one.

Work With Us

If your next raise or board meeting will test the market, traction or why-now slides, we can help. Phase 1 Market Intelligence is a 14-day, flat CA$4,500 rebuild of those three slides against NAICS-coded evidence, with a Phase 1 money-back guarantee (subject to terms). Phase 4 ships the storyboard, proof pack and proof tracker together. Reply with "Phase 1" or book at sagentix.ca/contact.

— Stephane Raby CMC · CISSP · P.Eng. · 25+ Years · Executive MBA

References

Get the next issue in your inbox

Bi-weekly. No spam. Unsubscribe anytime.