Early-Stage Venture Experiment Model
Back experiments that can become enormous rather than demanding false certainty
- Difficulty
- Advanced
- Time to result
- ~ongoing to results
- Steps
- 6
- Confidence
- 97%
Treat an early-stage startup as an experiment within a power-law portfolio. Because many investments are expected to return less than their cost, each accepted opportunity must be capable of producing an exceptional outcome. Start with the size and character of the potential market, then assess whether the founders can discover product-market fit by iterating around their initial product. Suspend disbelief long enough to model what success would become and whether these people could create it, while still checking whether the experiment is intelligently designed. Avoid detailed discounted-cash-flow models when inputs are almost entirely invented. As customers, revenue, retention, and operating history accumulate, progressively replace possibility-based judgment with metrics, multiples, and conventional valuation.
Origin
Extracted from Masters in Business, where Benedict Evans contrasted public-equity analysis with venture investing in startups consisting of founders, a presentation, and uncertain market hypotheses.
Core principles
- 01Most experiments may fail without invalidating the portfolio
- 02Every investment must possess exceptional upside
- 03Early-stage evaluation emphasizes markets and founder adaptability
- 04Product-market fit is discovered through iteration
- 05Financial precision becomes useful only after operating evidence appears
How to run it
- 1
Define the experiment
State what the startup is testing about a customer need, product, distribution channel, or market structure. Distinguish the core hypothesis from incidental implementation details.
Pro tip Write down what a negative result would teach before investing.
Watch out An incoherent experiment is not redeemed merely by having large theoretical upside.
- 2
Test the upside
Estimate whether a successful version could become large enough to matter under a power-law portfolio. Focus on order of magnitude rather than a falsely precise terminal value.
Pro tip Ask whether one extraordinary success could pay for many failed experiments.
Watch out A merely plausible small business may not fit a venture portfolio.
- 3
Assess the founders
Evaluate whether the founders can learn, recruit, adapt, and discover the market opportunity adjacent to their initial idea. Early-stage investing is partly a bet on their search capability.
Pro tip Look for evidence that the team has already revised its beliefs in response to reality.
Watch out Do not assume the original pitch will remain the final product or market.
- 4
Suspend disbelief
Temporarily set aside the obvious failure reasons and describe the company if it works. Then judge whether the success mechanism and team are credible enough to warrant the experiment.
Pro tip Separate “unlikely” from “impossible.”
Watch out Suspending disbelief does not mean ignoring fraud, incoherence, or untestable claims.
- 5
Design learning milestones
Identify the evidence that would show movement toward product-market fit, such as retention, usage, willingness to pay, or repeatable acquisition. Use milestones to guide iteration.
Pro tip Favor milestones that resolve the largest uncertainty first.
Watch out Do not treat vanity growth as proof of product-market fit.
- 6
Shift toward metrics
As the company generates reliable operating data, increase the weight given to unit economics, growth quality, multiples, and cash-flow analysis.
Pro tip Match valuation rigor to the maturity and reliability of the evidence.
Watch out Continuing to value a mature company only on possibility can conceal deteriorating economics.
In the wild
A seed investor avoids manufacturing a ten-year DCF. Instead, the investor asks whether the workflow could become essential across a very large market, whether the founders have unique insight, and what retention milestone would demonstrate product-market fit. Funding is staged around tests of adoption and repeat usage.
→ The decision rests on market scale, founder learning ability, and evidence-producing milestones rather than fabricated precision.
A team discovers that its intended consumers rarely use the product, while small businesses adopt one feature heavily. Rather than abandoning the company in a wholesale pivot, the founders iterate around that adjacent demand and test pricing, retention, and distribution.
→ The company uses evidence to discover a better product-market fit while preserving its core capability.
Common mistakes
Building a seed-stage DCF
Detailed cash-flow models for a company without dependable customers or revenue mostly formalize assumptions that cannot yet be known.
Rejecting every implausible idea
Transformative outcomes often look implausible because they require new behavior, infrastructure, or markets.
Ignoring experiment quality
Accepting failure as normal does not excuse a test that cannot generate useful evidence or was poorly executed.
Is it for you?
Best for
It is best for seed-stage investors and innovation teams managing portfolios of high-upside experiments.
Not ideal for
It is not ideal for valuing mature businesses with stable cash flows and comparable operating histories.
From the transcript
“everything you do has to be capable of being amazing”
“what you have to do is sort of suspend disbelief and not think, here are all the reasons why this might not work, But what…”
“Silicon Valley is it's a machine for running experiments, and most of the experiments won't work, and that's the plan.”
From the episode
Benedict Evans Talks About the Tech Bubble
Benedict Evans