Analysts new to early-stage investing tend to worry about bias at the obvious moments: when a founder is unusually charismatic, when a sector is overhyped, when a deal comes from a trusted source who has never been wrong. These are the moments that feel dangerous. They invite vigilance.
The research suggests something less comfortable. The five cognitive biases most consistently documented in investment decision literature do not typically appear when analysts are on guard. They appear at the moments investors feel most fluent, most experienced, most confident in their pattern recognition. The bias strikes not when the situation feels uncertain, but when it feels familiar.
This matters because awareness is not a sufficient defence. You cannot notice a bias that activates precisely when you feel least likely to be biased. The defences that work are not attitudinal: they are structural. They change what information is available at which point in the evaluation process, before the confident pattern-recognition System 1 thinking that Daniel Kahneman documented in Thinking, Fast and Slow has already closed the question.
Consider the scenario: a seed-stage analyst at a European fund has evaluated thirty companies in twelve months. A founder walks into a first meeting with a background that closely mirrors the fund’s last three successful investments: technical co-founder, enterprise SaaS, previous employer was a recognisable tier-one company. The analyst feels experienced. The pattern is familiar. System 1 is running. The questions asked are the right questions, but the weighting given to the answers has already been shaped by a template that may or may not apply here. The bias is not visible in the meeting notes. It is visible in the outcome.
The Five Biases: Where They Appear and What to Do About Them
1. Confirmation Bias
Definition: the tendency to search for, interpret, and recall information in a way that confirms a pre-existing hypothesis.
In the VC context, confirmation bias typically activates after the initial positive impression, not before it. The analyst who has decided a company is interesting will, in due diligence, disproportionately weight information that supports that view and disproportionately discount information that challenges it. Reference calls become exercises in selecting for confirming testimony. Market sizing analysis anchors on the optimistic scenario. Questions in follow-up meetings are framed in ways that make it easy for the founder to confirm rather than challenge the analyst’s thesis.
From the investor network: a seed-stage associate at a Benelux fund described a process in which the fund’s investment thesis for a particular sector had been developed over eighteen months. When a company appeared that perfectly matched the thesis, the due diligence process was, in practice, an exercise in documentation rather than interrogation. The sector knowledge that made the analyst feel most qualified to evaluate this deal was the same knowledge that made them least likely to consider evidence that challenged the thesis.
Structural countermeasure: the pre-mortem. Before beginning due diligence on a positively screened deal, the analyst writes a one-page memo assuming the investment has failed three years from now, and identifies the most plausible reasons why. This forces the generative question: what would the counter-evidence look like? It activates System 2 evaluation before System 1 has completed its pattern-matching closure.
2. Similarity Bias (Homophily)
Definition: the tendency to favour people who share similar backgrounds, experiences, or characteristics.
In investment research, Gompers, Mukharlyamov, and Xuan documented this pattern rigorously. Their paper “The Cost of Friendship,” published in the Journal of Financial Economics in 2016 (NBER Working Paper 18141), examined VC syndication decisions and found that investors were significantly more likely to co-invest with partners who shared the same ethnic, educational, or career background. This homophily measurably reduced the probability of investment success, and the detrimental effect was most prominent for early-stage investments. The study attributed the cost of affinity primarily to poor post-investment decision-making by syndicates with high background similarity: less challenge, less diversity of perspective, fewer corrective signals reaching the decision-makers.
At the individual evaluator level, similarity bias means that founders who attended the same university, came from the same employer, or share cultural background with the analyst receive a materially more generous reading of ambiguous signals. This is not a conscious favouritism. It is a pattern-recognition shortcut that has been documented consistently enough to be treated as a structural risk rather than an individual failing.
Structural countermeasure: evaluation rubrics completed independently before discussion. If multiple people evaluate a founder, each scores the pre-registered evaluation dimensions before any group conversation. Group discussion then examines where scores diverge, not where they converge. Convergence in a homophily-prone environment is a data point to investigate, not a reassurance.
3. Narrative Bias (Storytelling Over Evidence)
Definition: the tendency to over-weight coherent stories relative to base rates and statistical evidence.
The survey by Gompers, Gornall, Kaplan, and Strebulaev (NBER Working Paper 22587, 2016) of 885 institutional VCs found that the management team was consistently rated as the most important factor in investment decisions, ranked above product, market, and business model. This is a reasonable prior: team quality matters. What the finding also reveals, when read alongside the behavioural literature, is that team evaluation is highly susceptible to narrative bias. A founder who tells a coherent, compelling story about why they are uniquely positioned to solve this problem activates a cognitive process that Kahneman identified as System 1 pattern satisfaction: the story makes sense, the pieces cohere, the analyst feels they understand. The statistical question, which is whether this specific type of founder actually predicts outcomes at better than base rate, does not get asked. It is pre-empted by the narrative’s coherence.
From the investor network: an analyst at a seed fund in the Nordics described a process in which two founders with almost identical storytelling quality were evaluated differently. One had a polished narrative about a prior failure and what it taught them; the other communicated the same underlying resilience through technical depth and unpolished directness. The first founder advanced to partner meeting; the second was screened out at analyst level. The decision was later reviewed and reversed, but only because a partner happened to read both notes side by side.
Structural countermeasure: base rate anchoring before founder meetings. For each deal, the analyst documents the empirical base rate for the outcome being predicted: what percentage of companies at this stage, in this sector, with this team profile, have reached Series A? What does the distribution look like? This does not override qualitative judgment. It gives System 2 a prior to anchor to before the story begins.
4. Anchoring
Definition: the tendency to rely too heavily on the first piece of numerical information encountered when making subsequent quantitative judgements.
In early-stage investing, anchoring most commonly operates through valuation. The first number in a conversation, whether proposed by the founder, referenced from a comparable round, or generated by a quick mental calculation, exerts disproportionate influence on the final figure. This is not a valuation article, but the mechanism extends beyond valuation: anchoring also affects how analysts assess traction metrics. If a founder opens with the month in which growth was highest, that month becomes the reference point. Subsequent months are evaluated against that peak rather than against a realistic baseline.
A well-documented real-world pattern: in seed rounds, founders who present a deck with a specific valuation figure in the opening slides receive first-meeting counteroffers that cluster around that number regardless of whether the analyst had formed an independent view before seeing the deck. Analysts who form a preliminary valuation range before reading the deck produce counteroffers from a distribution centred on a materially different point.
Structural countermeasure: pre-meeting valuation estimation. Before reviewing a deck or meeting a founder, the analyst writes down the valuation range they would consider for a company at this stage, in this sector, with this level of traction, given the fund’s ownership targets. The number is not shared. It is a prior. Any offer that deviates significantly from that prior requires a written explanation of the new information that justified the deviation.
5. Escalation of Commitment (Sunk Cost Bias)
Definition: the tendency to continue investing resources in a course of action because of past investment, even when evidence suggests the course is not working.
Academic research on VC decision-making has documented sunk cost effects in follow-on investment decisions. A study published in the Journal of Corporate Finance (2024) found evidence that past capital investment influenced follow-on round decisions, with the bias operating not only at the individual level but at the organisational level within investment committees. For early-stage analysts, this bias appears most acutely after a company has consumed three to four months of evaluation time, relationship investment, and internal advocacy. Passing on the deal at that point is genuinely harder than it would have been at the beginning, for reasons that have nothing to do with the quality of the company.
The practical manifestation is that deals which should have been declined at initial screening are occasionally funded because the cumulative time investment made passing feel disproportionately costly. The decision is not, at that point, being made about the company. It is being made about sunk time.
Structural countermeasure: stage-gate criteria set before the evaluation begins. At each stage of evaluation, the criteria required to advance to the next stage are pre-specified and written down. Advancement decisions are made against those pre-specified criteria, not against the overall investment of time. If the criteria for advancing to IC are not met, the deal does not advance. The time already spent is not a relevant input.
The Bias Audit: A Practical Table
| Bias | When It Strikes | The Signal You Are in It | The Structural Defence |
| Confirmation | After positive first impression, during due diligence | Due diligence feels like validation, not investigation | Pre-mortem memo before diligence begins |
| Similarity | First meeting; founder background assessment | Meeting felt unusually smooth; conversation was unusually easy | Independent rubric scoring before group discussion |
| Narrative | During and immediately after founder pitch | You find yourself explaining the investment to yourself in the founder’s own language | Base rate documentation before any founder meeting |
| Anchoring | First moment a number is stated | Your internal estimate shifted upon hearing the founder’s figure | Pre-meeting valuation estimate, written before deck review |
| Escalation | After three or more months of active evaluation | Declining now feels wasteful rather than correct | Stage-gate advancement criteria set at evaluation start |
The Implication
Awareness of these five biases is necessary but insufficient. The well-documented problem with awareness as a defence is that these biases are most active precisely when the analyst feels most competent, most on familiar ground, most certain they are evaluating clearly. The confidence that signals expertise to System 2 is the same confidence that permits System 1 to close questions before they have been properly opened.
What outperforms awareness is process architecture: decision checkpoints that require specific outputs before advancement, criteria that must be specified in advance, independent assessments that are completed before group discussion shapes them. These interventions do not make analysts better at using their intuition. They create conditions in which the intuition is tested against evidence before it solidifies into a conclusion.
The one change to implement first: add a single pre-meeting requirement to your evaluation workflow. Before any founder meeting on a deal you have positively screened, write two paragraphs. The first states the base rate context for this type of company reaching the next milestone. The second states the specific piece of evidence that would cause you to pass. If you cannot write the second paragraph, you have already confirmed the deal and are no longer evaluating it.