The Say-Do Gap: What 7,800+ VC Investment Theses Reveal About How Investors Actually Evaluate Startups

If you ask a VC what they look for in a startup, they will say "team." Gompers et al. (2020) surveyed 885 institutional VCs and found 95% cited management team as the single most important factor.

We tested this claim against data. Using 7,800+ real investor thesis statements, 249 scored evaluations, and — most critically — 162 real VC deal memos from 15 ANZ VCs explaining exactly why they invested, we compared what VCs say they prioritize against what actually predicts their investment decisions.

The results are striking. And the new VC memo data changes the picture.

What Do VCs Actually Evaluate? Stated vs Revealed Preferences

Product execution is the #1 predictor of evaluation outcomes in both AI scoring and human expert assessment. But it is mentioned in only 4.1% of investment theses — the least-stated of the four major evaluation dimensions.

Why Do VCs Say One Thing and Do Another?

Thesis statements are LP-facing marketing, not evaluation rubrics. VCs write about market opportunity to attract limited partners. When they sit down with a deck, they evaluate what has been built — product maturity, moats, defensibility, and traction.

Team is assessed in meetings, not decks. A team slide is a paragraph of bios and a LinkedIn URL. Product is screenshots, architecture, metrics, and competitive analysis. The deck medium favours product evidence over team narrative.

Traction is assumed, not articulated. Only 0.8% of theses mention traction or revenue. VCs do not write "we want growth" in their thesis because it is obvious. But traction is the #3 predictor in practice (r=0.803).

Does VC Evaluation Differ by Geography?

The say-do gap is not uniform. Analysis of 4,956 investors across 12 global tech hubs reveals systematic differences in evaluation culture:

• Melbourne is the most team-focused hub globally — 24.1% of theses mention founders (vs 4.3% in Berlin)
• Berlin and London are the most product-focused (39.1% and 37.7%)
• Sydney balances team and market equally (15.6% and 16.9%)
• Australian regional investors are uniquely market-focused (15.7%) — likely because the AU domestic market is small, so VCs need founders to demonstrate global TAM

How Does Evaluation Differ by Sector?

What drives evaluation also depends on what you are building:

• Fintech: Traction is #1 (r=0.900). VCs want to see revenue.
• Consumer: Traction AND financials dominate (r=0.929, r=0.778). Unit economics are everything.
• Healthcare: Market and product dominate (r=0.916, r=0.923). Team barely registers (r=0.503). The science matters more than the scientist.
• Tech (general): Most balanced — product (0.855) and team (0.806) are closest to equal.

What Should Founders Do Differently?

Your pitch deck is a product document, not a team document. If your strongest slide is "Our Team," you are leading with the weakest signal in the medium that matters most for getting the meeting.

Lead with what you have built. Lead with what is working. Lead with the moat. The team assessment happens when they meet you — not when they read your deck.

NuScore evaluates your deck across 7 dimensions and shows you a ranked score waterfall of exactly which signals moved your score. See where product, team, and traction land for your specific deck.

See your score waterfall at nuvc.ai →

Methodology and Data Sources

This analysis draws on three independent data sources:

• 7,800+ investor thesis statements from the NUVC investor database, analysed for keyword frequency across 9 evaluation dimensions. Active investors only, thesis length > 20 words.
• 197 AI-scored pitch decks evaluated by the VCGrade 2.0 hybrid scoring engine (LLM judge + rule engine) across 5 dimensions. Pearson correlations computed against overall score.
• 110 human-scored accelerator applications from an Australian accelerator competition (2024), scored by experienced program evaluators across 6 dimensions on a 0-100 scale. Anonymised for publication.

Statistical methods: Pearson correlation (r), coefficient of determination (r²), point-biserial correlation for outcome analysis. Geographic analysis uses keyword extraction from thesis free-text, bucketed by investor location. Sector analysis uses industry classification from pitch deck metadata.

NEW: What 172 Real VC Deal Memos Reveal (March 2026 Update)

We extracted structured investment reasoning from 162 real deal memos published by 15 ANZ VCs. Each memo was analysed for 7-dimension scores, thesis type, key signals, and accepted red flags.

Finding: Conviction Is the #1 Predictor (r=0.598)

When we correlate each dimension against the VC's overall investment enthusiasm, conviction — the gut-level "this is a winner" signal — correlates most strongly at r=0.598. Traction is #2 (r=0.544). Team is #5 (r=0.328). This is the clearest evidence that VCs invest on conviction first, then rationalise with team/market narratives.

Finding: Team-First Deals Score Lower

37% of deals were classified as "team-first" investments. But team-first deals average a lower overall score (7.64) than tech-moat (7.75) or multi-thesis (7.93) deals. Serial founders only score +0.1 above baseline. The best investments check multiple boxes simultaneously.

Finding: Financials Are Universally Ignored

Across ALL thesis types, VCs score financials 5.0-5.2/10. At growth stage, the correlation drops to r=0.117 — nearly zero. Stop perfecting your financial model for seed. Spend that time on product depth and traction.

Finding: "It's Competitive" Is Not a Dealbreaker

18% of accepted red flags relate to competition. VCs almost never reject on competition alone — "crowded but big" is perfectly acceptable.

Reference: Gompers, P. A., Gornall, W., Kaplan, S. N., & Strebulaev, I. A. (2020). How do venture capitalists make decisions? Journal of Financial Economics, 135(1), 169-190. Bernstein, S., Korteweg, A., & Laws, K. (2017). Attracting early-stage investors: Evidence from a randomized field experiment. The Journal of Finance, 72(2), 509-538.