# Studies of how investors decide on a pitch

Thirty studies with data on what moves an investor's decision: 11 experiments, 14
observational studies with an outcome, and 5 surveys, meta-analyses or platform reports. Use it
before you accept a pitch rule as fact: check whether anything measured it.

Each link goes to the full text where one is open, otherwise to an abstract record, and the DOI
is given as text. Quotes are from the abstract or text, under 25 words.

**Strength** is our grading, not the authors': **strong** = randomized or quasi-experimental with
real investors or large n and the full text read; **moderate** = a real design with one big limit
(abstract only, one investor, self-report, stated interest); **weak** = small n, non-investor
subjects, edited TV, vendor data or numbers from a secondary summary.

## Experiments (causal)

| Study | Design and n | Finding | Strength |
|---|---|---|---|
| Bernstein, Korteweg & Laws 2017, *Journal of Finance* ([full text](https://www.haas.berkeley.edu/wp-content/uploads/BernsteinKorteweg_032314.pdf)), doi:10.1111/jofi.12470 | field experiment; ~17,000 emails to 4,494 active AngelList investors about 21 startups (2013) | Showing the team raised clicks; the published effect size could not be re-read (the linked draft refuses automated requests). Traction and lead investor had no effect on the average investor; inexperienced investors reacted to all three. Abstract: "responds strongly to information about the founding team, but not to firm traction". | **Moderate**: randomized, real investors, but the full text could not be re-read and the outcome is a click, not a check. |
| Bapna 2019, *Management Science* ([abstract](https://api.openalex.org/works/doi:10.1287/mnsc.2017.2833)), doi:10.1287/mnsc.2017.2833 | field experiment; not given in the abstract; investors on an equity platform | Third-party signals work together: product certification plus a prominent customer raised interest 72% over no signal; certification plus social proof, 65%. Abstract: "a signal about product characteristics is the key to unlocking" the rest. | **Moderate**: randomized, but stated interest and no n. |
| Gornall & Strebulaev 2024, *Management Science* ([abstract](https://api.openalex.org/works/doi:10.1287/mnsc.2024.4990)), doi:10.1287/mnsc.2024.4990 | field experiment; 80,000 cold pitch emails to 28,000 VCs and angels (2018) | Identical pitches: female founders got 9% more interested replies, Asian founders 6% more than white founders. Reply rate about 6.5% (as cited by Zhang 2022). Abstract: investors "do not discriminate" on unsolicited pitch emails. | **Moderate**: large and randomized, but fictitious startups and replies, not investment. |
| Zhang 2022, working paper (arXiv:2010.16084) ([full text](https://arxiv.org/pdf/2010.16084)) | field experiment; >17,000 VCs emailed (Mar-Apr 2020), plus an incentivized profile-rating experiment | Cold emails: 12% opened, about 24 s of reading, about 1% clicked or replied. An elite school in the subject line raised opens 0.7 pp; a female name, 1 pp. | **Moderate**: randomized, but run during COVID, not peer-reviewed in the version read, and the outcome is opens. |
| Clingingsmith & Shane 2018, *Management Science* ([full text](https://www.davidclingingsmith.com/s/training-aspiring-entrepreneurs-to-pitch-experienced-investors-evidence-from-a-field.pdf)), doi:10.1287/mnsc.2017.2882 | field experiment; 271 aspiring founders in 4 elevator-pitch competitions; accredited-investor judges | Pitch training (by video) raised use of best practices by about a third of a SD. With experienced investors it helped only high-quality ideas and hurt the rest: training "helps investors to distinguish between high- and low-quality venture ideas". | **Moderate**: randomized with real investors, but students with ideas, not startups. |
| Clingingsmith & Shane 2017, preprint ("Let Others Go First") ([full text](https://osf.io/download/vafg4/)) | field experiment; 271 participants, 50 judges, 897 ratings in the same competitions | Pitching first or second cut judge interest by about 2 points (about 0.36 SD, p<0.01) against the average slot; slot 4 scored 4.18 points above slot 1. Holds for all but the weakest 30% of pitches. | **Moderate**: order was randomized, but unpublished and the same student sample. |
| Brooks, Huang, Kearney & Murray 2014, *PNAS* ([full text](https://pmc.ncbi.nlm.nih.gov/articles/PMC3970475/)), doi:10.1073/pnas.1321202111 | lab experiment; field: 90 pitches in 3 competitions; experiments: 521 MTurk adults, 194 lab participants | The same pitch narrated by a male voice was chosen for funding by 68% versus 32% with a female voice (p<0.0001). In the field, men were 60% more likely to win. | **Weak** for investor behaviour: the experiments use non-investors; only 20 women-led pitches in the field. |
| Shane, Drover, Clingingsmith & Cerf 2020, *Journal of Business Venturing* ([full text](https://www.davidclingingsmith.com/s/Founder-Passion-and-Investor-Interest.pdf)), doi:10.1016/j.jbusvent.2019.105949 | lab experiment; 15 people in an fMRI scanner watching pitch videos | High displayed passion raised neural engagement 39% and stated interest 26%. Abstract: high-passion founders "increase ... investor interest in the venture by 26%". | **Weak**: n=15, non-professional investors, staged pitches. |
| Chen, Yao & Kotha 2009, *Academy of Management Journal* ([abstract](https://api.openalex.org/works/doi:10.5465/amj.2009.36462018)), doi:10.5465/amj.2009.36462018 | lab experiment; not given in the abstract; VCs in a lab experiment and a field study | Abstract: both studies "consistently showed that preparedness, not passion, positively impacted decisions to fund". No effect size given. | **Moderate**: real VCs and two designs, but abstract only and pre-2015. |
| Chan & Park 2015, *Journal of Business Venturing* ([abstract](https://ideas.repec.org/a/eee/jbvent/v30y2015i5p732-748.html)), doi:10.1016/j.jbusvent.2014.12.002 | lab experiment; not given in the abstract; screening evaluators, quasi-field study plus experiments | Product images helped partially; red hurt: "a consistent negative effect of red on favorable screening decisions". No magnitudes. | **Weak**: abstract only, no magnitudes, business plans rather than decks. |
| Lee & Huang 2018, *Organization Science* ([abstract](https://api.openalex.org/works/doi:10.1287/orsc.2017.1172)), doi:10.1287/orsc.2017.1172 | lab experiment; not given in the abstract; field evaluations plus an experiment | Social-impact framing shrank the gender penalty for women-led ventures through perceived warmth; the effect "did not apply to men". | **Weak**: abstract only, no n or magnitudes; the outcome is an evaluation. |

## Observational, with an outcome

| Study | Design and n | Finding | Strength |
|---|---|---|---|
| Huang & Pearce 2015, *Administrative Science Quarterly* ([full text](https://cpb-us-e2.wpmucdn.com/sites.uci.edu/dist/9/311/files/2015/11/HuangPearceASQAngels2015.pdf)), doi:10.1177/0001839215597270 | observational, with outcome; 28 interviews; experiment with 135 angels; 90 angels rating 90 recorded pitches, followed 4 years | A positive gut read of the entrepreneur predicted home runs four years later, but not survival or growth, and funding only marginally (p<0.10). Only 6 home runs in the sample. | **Weak** as a predictor: 6 home runs, and the raters were not the companies' real investors. |
| Kanze, Huang, Conley & Higgins 2018, *Academy of Management Journal* ([abstract](https://api.openalex.org/works/doi:10.5465/amj.2016.1215)), doi:10.5465/amj.2016.1215 | observational, with outcome; 189 startups at TechCrunch Disrupt NYC Startup Battlefield, 2010-2016, plus an experiment | Each extra prevention-focused Q&A question (risk, loss) went with about $3.8M less raised; in the experiment, answering it with a growth-focused answer raised funding. Abstract: each such question "significantly hinders" fundraising. | **Weak to moderate**: field correlation plus an experiment; the $3.8M figure is from a [secondary summary](https://gap.hks.harvard.edu/we-ask-men-win-and-women-not-lose-closing-gender-gap-startup-funding). |
| Hu & Ma 2025, *Journal of Finance* ([full text](https://songma.github.io/files/hm_video.pdf)), doi:10.1111/jofi.13471 | observational, with outcome; 1,139 one-minute application videos to YC, MassChallenge, 500 Startups, Techstars and AngelPad, plus an MBA experiment | One SD more positive delivery raised the chance of selection 3 pp on an 8.5% base (+35%), robust to team controls. Among those selected, the more positive did worse later. Abstract: delivery works by leading investors "to form inaccurate beliefs". | **Moderate** for selection: observational, large n, tracked outcomes. |
| Pollack, Rutherford & Nagy 2012, *Entrepreneurship Theory and Practice* ([abstract](https://api.openalex.org/works/doi:10.1111/j.1540-6520.2012.00531.x)), doi:10.1111/j.1540-6520.2012.00531.x | observational, with outcome; 113 funded pitches from Shark Tank (2009) and Dragons' Den (2005-2010) | Preparedness went with more funding, through perceived legitimacy: "Cognitive legitimacy, in turn, was positively related to amount of funding received". | **Weak**: funded pitches only, edited TV, abstract only. |
| Warnick, Davis, Allison & Anglin 2021, *Journal of Business Venturing* ([abstract](https://api.semanticscholar.org/graph/v1/paper/DOI:10.1016/j.jbusvent.2021.106109?fields=title,abstract,year,venue)), doi:10.1016/j.jbusvent.2021.106109 | observational, with outcome; 489 funding pitches scored by facial-expression software | Inverted U for happiness, anger and fear (some beats none or a lot); sadness hurts; more changes of expression go with more funding (frequency of change "promotes funding"). | **Weak**: abstract only, correlational, context and magnitudes unknown. |
| Balachandra, Briggs, Eddleston & Brush 2019, *Entrepreneurship Theory and Practice* ([abstract](https://api.crossref.org/works/10.1177/1042258717728028)), doi:10.1177/1042258717728028 | observational, with outcome; 185 pitches at university elevator-pitch competitions (2007-2008) | Being a woman did not lower interest; stereotypically feminine behaviour cut the chance of reaching the final by 13 pp, for men and women alike. Abstract: investors are "biased against the display of feminine-stereotyped behaviors". | **Weak**: university competition; the 13 pp figure is from a [secondary summary](https://gap.hks.harvard.edu/dont-pitch-girl-how-gender-stereotypes-influence-investor-decisions). |
| Maxwell, Jeffrey & Lévesque 2011, *Journal of Business Venturing* ([abstract](https://cemi.com.au/__static/d131b455b58151434ebbe127488e80d3/06011-maxwell-jeffrey-levesque-2011%282%29.pdf?dl=1)) | observational, with outcome; 150 entrepreneur-investor interactions on Dragons' Den (Canada) | Angels do not add up many attributes; they reject at the first fatal flaw. Eight causes: adoption, product stage, protection, customer engagement, route to market, market potential, relevant experience, financial model. | **Weak to moderate**: real angels, but edited TV, and only a one-page summary was read. |
| Parhankangas & Ehrlich 2014, *Journal of Business Venturing* ([abstract](https://ideas.repec.org/a/eee/jbvent/v29y2014i4p543-564.html)), doi:10.1016/j.jbusvent.2013.08.001 | observational, with outcome; not given in the abstract; nascent ventures seeking angels in the New York area | Angels favour moderate positive language, self-promotion of innovation, supplication and attacks on competitors, plus high opinion conformity: "the moderate use of positive language". | **Weak**: abstract only; written proposals, not spoken pitches. |
| Kirsch, Goldfarb & Gera 2009, *Strategic Management Journal* ([abstract](https://api.openalex.org/works/doi:10.1002/smj.751)), doi:10.1002/smj.751 | observational, with outcome; 722 funding requests to one US VC | Having a business plan, and most of what it says, is "weakly associated with VC funding decisions"; the VC learns that information through other channels. | **Moderate**: real decisions, but one VC in the dot-com era, abstract only, plans not decks. |
| Kaplan, Sensoy & Strömberg 2009, *Journal of Finance* ([abstract](https://api.openalex.org/works/doi:10.1111/j.1540-6261.2008.01429.x)), doi:10.1111/j.1540-6261.2008.01429.x | observational, with outcome; 50 VC-backed companies from first business plan to IPO | The business line stayed stable while management turned over. Abstract: investors "should place more weight on the business". | **Moderate** on what persists, silent on pitches: survivors only, abstract only. |
| Howell 2020, *Journal of Financial Economics* ([full text](https://www.nber.org/system/files/working_papers/w23874/w23874.pdf)), doi:10.1016/j.jfineco.2019.10.009 | observational, with outcome; 4,328 ventures in 87 competitions across 17 US states, 2007-2015 | Winning a round (regression discontinuity on judge rank) raised the chance of outside funding 9-13 pp on a 24% base, even with no prize money. Abstract: "judge ranks strongly predict venture success". | **Strong** (quasi-experimental), but a local effect near the cutoff; it does not say which pitch element earned the rank. |
| McKenzie & Sansone 2019, *Journal of Development Economics* ([full text](https://documents1.worldbank.org/curated/en/968231513116778571/pdf/WPS8271.pdf)) | observational, with outcome; >1,000 winners and >1,000 non-winners of Nigeria's YouWiN! competition, 3-year follow-up | Judge scores did not predict survival, jobs, sales or profit three years on (R² < 0.04); machine learning did no better. Abstract: scores "are uncorrelated with business survival". | **Moderate** (observational, large n). Small firms, written plans, judges who are not VCs. |
| Gius 2025 (online 2024), *Strategy Science* ([abstract](https://api.openalex.org/works/doi:10.1287/stsc.2024.0169)), doi:10.1287/stsc.2024.0169 | observational, with outcome; 67 venture competitions | The more judges disagree on a startup, the likelier it succeeds, above all for unusual proposals; the business model polarizes most. Abstract: "the more venture competition judges disagree ... the more likely the startup is to succeed". | **Weak to moderate**: abstract only, no magnitude. |
| Baek & Hegde 2026, NBER working paper 35063 ([full text](https://www.nber.org/system/files/working_papers/w35063/w35063.pdf)) | observational, with outcome; ~750,000 US startups linked to 329 accelerators | Raw: accelerated startups raise $0.53M more and are 29 pp likelier to raise at all. Corrected for selection, 60 to 80% of accelerators add negative value (the introduction says 60-80 percent; figure 5, about 80%), and 65% of startups joined one of those: "a small right tail generates large gains". | **Moderate**: very large n, but a working paper not yet peer-reviewed, and it does not isolate demo day. |

## Surveys, meta-analysis and platform data

| Study | Design and n | Finding | Strength |
|---|---|---|---|
| Clark 2008, *Venture Capital* ([abstract](https://api.openalex.org/works/doi:10.1080/13691060802151945)), doi:10.1080/13691060802151945 | survey; 24 UK angels rating 3 live pitches | Presentation factors drove scores and interest most, yet angels explained their decisions only by substance; they "appeared to be unaware of" the effect. | **Weak**: 24 angels, 3 pitches, abstract only. |
| Gompers, Gornall, Kaplan & Strebulaev 2020, *Journal of Financial Economics* ([full text](https://www.nber.org/system/files/working_papers/w22587/w22587.pdf)) | survey; 885 institutional VCs at 681 firms | Team rated important by 95% and most important by 47% (53% at early stage); business model 10%, product 13%, market 8%. Only 10% of deals come inbound from the company; about 30% through professional networks, 20% from other investors. | **Moderate**: large, but self-reported; Clark 2008 and Hu & Ma 2025 show investors under-report how much delivery weighs. |
| Roccapriore, Crook & Cardon 2021, *Academy of Management Proceedings* ([abstract](https://api.openalex.org/works/doi:10.5465/ambpp.2021.10280abstract)), doi:10.5465/ambpp.2021.10280abstract | meta-analysis; 87 studies, 45.4M investor decisions | Objective signals (preparedness, experience) and subjective ones (passion, coachability) both help, but "objective signals lead to 1.7 times as many successful funding outcomes". Investor social capital matters more than investor experience. | **Weak to moderate**: conference abstract; the pooled decisions are mostly crowdfunding (our estimate, from the scale). |
| Dropbox DocSend 2026, pre-seed deck guide ([full text](https://www.docsend.com/blog/what-to-include-when-building-your-pre-seed-pitch-deck/), unverified); DocSend's [research overview](https://www.dropbox.com/resources/docsend-pitch-deck-research) does open | platform data; not disclosed | Pre-seed decks: 4 min 10 s per view, business model 83 s, product 77 s, team 46 s, market 39 s, traction 37 s. | **Weak**: vendor data, method not published; reading time is not a decision. |
| Papermark 2026, Fundraising Report ([full text](https://www.papermark.com/fundraising-report)) | platform data; 24,541 decks, 358,672 investor views, 184 countries, 2024-2026 | 4.0 min per view; rounds that closed got about 36 min of total reading versus about 18 typical; the team slide is read longest (5.7 s per view); the biggest drop-off is between slides 1 and 2; fewer than half of views reach the end; 9 to 16 pages engage best. | **Weak**: vendor report; hot deals get read more, so causality likely runs backwards. |

## What it adds up to

- **Team information moves interest; metric traction does not move the average investor.**
  Randomized emails: team raised clicks, traction no effect except for inexperienced investors
  (Bernstein 2017; the outcome is a click, not a check). Elite schooling in a cold-email subject
  line: +0.7 pp opens (Zhang 2022; a working paper, run during COVID).
- **Third-party validation works in combination**: certification plus a prominent customer,
  +72% interest (Bapna 2019; stated interest, n not given).
- **Delivery predicts selection, not returns.** One SD more positive delivery goes with a 3 pp
  higher chance of selection on an 8.5% base, and among those selected the most positive do
  worse (Hu & Ma 2025). With VCs, preparedness moves decisions and passion does not (Chen 2009;
  abstract only). Angels score delivery but explain themselves with substance (Clark 2008; 24
  angels).
- **Training only helps a good idea** in front of experienced investors (Clingingsmith 2018;
  student founders with ideas, not startups).
- **Order may matter:** pitching first or second cost about 0.36 SD of judge interest, in one
  unpublished preprint with the same student founders (Clingingsmith 2017).
- **Q&A framing matters:** each risk-focused question goes with less money raised, and answering
  with growth helped in the experiment (Kanze 2018; the field result is a correlation).
- **Angels reject at the first fatal flaw** rather than adding up scores (Maxwell 2011; edited TV,
  one-page summary read).
- **Judges' scores predict weakly and inconsistently.** Winning a competition round raises the
  chance of outside funding 9-13 pp (Howell 2020); in Nigeria, scores predicted nothing three years
  out (McKenzie 2019); judge disagreement predicts success (Gius 2025; abstract only).
- **Written plans weigh little** in a VC's decision (Kirsch 2009; one VC, dot-com era).
- **60 to 80% of accelerators add negative value** after correcting for selection, and a few add a
  lot (Baek 2026; a working paper not yet peer-reviewed).
- **Survey and platform data:** team is the top factor for 47% of VCs and only 10% of deals
  arrive inbound (Gompers 2020; self-reported); a deck gets 3 to 4 minutes per view, and the
  largest drop-off is between slides 1 and 2 (DocSend, Papermark; neither publishes its method).

## Where it contradicts common advice

- "Passion sells": it sells to non-professionals and at selection, but does not predict returns
  (Hu & Ma 2025); VCs respond to preparedness (Chen 2009, abstract only). Roccapriore 2021 points
  the same way, but by our estimate from its scale the pooled decisions are mostly crowdfunding.
- "Traction is king": randomized, it did not move the average investor's clicks (Bernstein
  2017), while third-party validation moved stated interest (Bapna 2019). Mixed.
- "Bet on the jockey": surveys say team (Gompers 2020, self-reported); data on survivors says
  business (Kaplan 2009, abstract only).
- "Practise the pitch": pays only when the idea is good (Clingingsmith 2018, student founders).
- "Go first at demo day": the first slots lost, in one unpublished preprint with student
  founders (Clingingsmith 2017).

## What no study here measures

The type of opening (hook, story, statistic), the market-size slide, a quantified problem, slide
order (DocSend asserts one without a method), live demo versus video, the ideal length of a
spoken pitch, and anything specific to crypto or hackathons. Treat advice on those as opinion.