The Precision Capital Thesis: Why Focus Beats Scale in Seed Investing
There is a version of venture capital that looks like a numbers game. Deploy small checks across a hundred companies, accept that 90 will fail, and construct a portfolio large enough that the one or two breakout winners return the fund many times over. The logic is seductive. If venture returns are driven by power law dynamics — and they are — then maximising exposure to the tail seems like the rational strategy.
At KnownWeil Capital, we have built our firm around a deliberate rejection of this model. Not because the logic is wrong — the power law does govern venture returns — but because we believe the spray-and-pray approach misidentifies what generates access to the best outcomes at the seed stage. This essay explains our Precision Capital thesis: why a focused, high-conviction portfolio with deep founder partnership produces superior returns to broad deployment, and what the data from our own experience suggests about why this is so.
The Scale Illusion
The scale model in seed investing is predicated on a specific assumption: that deal access is fungible. If you can see a hundred seed deals, any of them might be the one that returns 100x, so the optimal strategy is to maximise the number of bets. This assumption breaks down in practice for a specific reason: the best founders are not uniformly distributed across the deal flow of all funds.
Consider the empirical pattern in venture returns at the seed stage. A 2023 analysis of seed fund performance data from Cambridge Associates found that the top quartile of seed funds generated a median net IRR of 28.4% over the 2015–2020 vintage years, compared to 11.2% for the median fund. The gap between top-quartile and median performance at seed is wider than at any other stage of venture investing. This dispersion is not random. It reflects the degree to which seed fund performance is concentrated in a small number of firms that have developed genuine proprietary access to exceptional founders — access that broad deployment strategies cannot replicate.
Exceptional founders choose their investors. The best seed-stage founders — the ones who have already demonstrated extraordinary capability in their domain, who have the specific technical insight or market access that makes their company structurally defensible — receive multiple term sheets. They select investors based on a combination of capital terms, reputation, and the specific value they believe the investor can add. A fund deploying 50 checks per year into companies identified through a broadly distributed sourcing process is not competing for the same founders as a fund making 6–8 investments per year based on deep relationships and referral networks built over a decade.
What Precision Capital Means in Practice
Our Precision Capital model has three structural commitments that distinguish it from the scale approach.
First, portfolio concentration. We deploy a maximum of $5M per company at the seed stage, and we make 6–10 new investments per year. This gives us a portfolio that is large enough to benefit from diversification but small enough that every partner has meaningful time to spend with every portfolio founder. A partner at a fund making 40 seed investments per year cannot have substantive ongoing relationships with 40 founding teams. The math does not work. A partner at KnownWeil can have a meaningful relationship with 15–20 companies across the active portfolio — close enough to provide genuine operational support, open enough to receive honest information about the company's challenges.
Second, diligence depth. We spend significantly more time on each potential investment than the average seed fund. Our target is 40–60 hours of partner time per deal before we make an investment decision, including reference calls with former colleagues, customers, and other investors who have worked closely with the founder. For a fund making 200 investments per year, 40 hours per deal is arithmetically impossible. For us, it is a standard that every investment meets. This depth of diligence does not eliminate risk — no amount of diligence can accurately predict which seed-stage companies will achieve product-market fit — but it significantly improves the quality of the conviction we develop and reduces the frequency of investments made on insufficient information.
Third, post-investment partnership. Our model assumes that the relationship between KnownWeil and a portfolio company does not end at the wire. We expect to be involved in the 12–18 months following the seed investment more intensively than at any subsequent stage, because the decisions made in those months — about hiring, product strategy, initial customers, and the narrative used to raise the Series A — have a disproportionate impact on the company's trajectory. Portfolio founders have access to a partner's time on a weekly basis if they want it. We connect them to potential customers, help them navigate hiring decisions, and prepare them for Series A conversations with the best firms in Europe and the United States.
The Counter-Argument: Diversification and Its Limits
The strongest argument against our approach is the diversification argument. If venture returns are driven by a small number of extreme outliers, and if those outliers are unpredictable, should not a rational investor maximise the number of bets to ensure they are in the right place when the unpredictable outcome occurs?
The diversification argument is compelling as a theoretical matter but breaks down when you examine the empirical distribution of returns at the seed stage. The most important finding from academic research on venture capital returns — including the work of Harris, Jenkinson, and Kaplan at the University of Chicago — is that fund performance is not primarily driven by being present in a large number of investments. It is driven by a small number of very large investments in which the fund holds a meaningful position. Owning 1% of a company that returns 100x produces the same gross return as owning 10% of a company that returns 10x, but the latter outcome requires substantially less diversification to achieve.
At the seed stage, funds that deploy broad portfolios typically own 1–3% of the companies they back. Funds that deploy concentrated portfolios with higher conviction typically own 6–10%. The power law governs both approaches, but the concentrated fund captures a larger share of the returns from each of its winners. When the outcome distribution at seed has a sufficiently fat tail — and the evidence suggests it does — the concentrated approach with higher ownership outperforms the diversified approach with lower ownership even if the absolute number of winners is smaller.
"We would rather own 8% of 12 exceptional companies than 1% of 100 average ones. The power law rewards position size as much as portfolio breadth."
The Sourcing Question
The most common objection to the precision model is the sourcing question: how does a fund that makes only 6–10 investments per year ensure it is seeing the best opportunities? The worry is that a narrow deal funnel will mean missing the most important companies.
Our experience has been that the sourcing challenge facing concentrated funds is real but manageable, and that the solutions are very different from those available to broad-deployment funds. Concentrated seed funds cannot rely on volume as a sourcing strategy. They must develop deep referral networks that generate high-quality, pre-filtered introductions. For KnownWeil, this means intensive cultivation of relationships with two groups: outstanding Series A and B-stage investors who refer their best seed-stage deal flow to us as a preferred co-investor and warm-up fund, and the portfolio companies themselves, whose founders invariably know other exceptional founders in their networks.
The quality differential between sourcing strategies matters enormously at the seed stage. A cold inbound from a founder who discovered us through our website carries fundamentally different signal than a referral from a Series A partner who has spent 10 hours with the founder and believes they are exceptional. We invest almost exclusively from the latter category. This means our deal volume is lower than that of funds that actively seek broad inbound flow, but the conversion rate from first meeting to investment is significantly higher — roughly 1 in 12 for referred opportunities, compared to industry averages of 1 in 50 or 1 in 100 for broad-funnel funds.
Portfolio Construction and Reserve Strategy
One dimension of the precision model that receives less attention than it deserves is reserve strategy — the allocation of capital to follow-on investments in existing portfolio companies versus new investments. Our reserve ratio is more conservative than many seed funds: we maintain reserves of approximately 1.5x our initial investment for each portfolio company, which we deploy selectively into companies that have demonstrated clear product-market fit signals between their seed and Series A rounds.
The reserve strategy is a direct expression of the precision philosophy. We are not attempting to own a slice of every company in the portfolio through to the growth stage — the dilution from subsequent rounds makes that mathematically unattractive. We are attempting to deploy follow-on capital into the subset of portfolio companies where our conviction has been validated by evidence, increasing our ownership in the companies that are demonstrably working while allowing our ownership in the rest to be diluted without follow-on investment.
This approach requires genuine discipline. The temptation to support struggling portfolio companies with additional capital — either from loyalty to the founder or from reluctance to recognise a loss — is one of the most common forms of value destruction we observe in seed fund portfolios. Reserve capital is not a rescue fund. It is an optionality instrument that we exercise only when the company has crossed the threshold from uncertain to compelling.
Why Now Is the Right Time for This Approach
The case for precision capital has strengthened considerably over the past three years for a specific structural reason: the cost of starting a technology company has fallen dramatically, while the cost of building a category-defining company has risen.
Open-source AI models, cloud infrastructure, and developer productivity tools have reduced the cost of building an initial product to a fraction of what it was in 2015. A two-person founding team can build and ship a functional B2B SaaS product in six months for less than €100,000 in operating costs. This compression of initial development costs means that the seed stage is now occupied by a much larger number of companies than it was in previous years. The number of seed rounds closed globally in 2024 was approximately 14,000, up from roughly 6,000 in 2018, according to PitchBook data.
But the cost of building a company from seed-stage concept to Series B-stage defensibility has risen in parallel. Enterprise sales cycles have lengthened as buyers face AI vendor fatigue. The talent required to build differentiated AI products — research engineers, ML infrastructure specialists, applied scientists — commands compensation packages that were unimaginable in the pre-ChatGPT era. And the competitive pressure from well-funded incumbents moving down-market into the seed-stage opportunity has intensified in virtually every technology category.
This combination — more companies at the seed stage, higher cost to reach the Series A threshold — creates the conditions that favour the precision model. In a market where capital is abundant and the quality of seed deals is highly variable, the funds that will generate superior returns are those that combine rigorous selection with active post-investment support to help the best companies navigate the increasingly expensive journey from first product to repeatable revenue.
A Note on What We Are Not
Precision Capital is not a claim to infallibility. We will lose money on more than half of the companies we back. That is the statistical reality of seed-stage investing, and any fund that suggests otherwise is either selecting its portfolio from a later stage than it claims or is presenting cherry-picked performance data.
What we are claiming is a different set of trade-offs from the broad-deployment model. We will see fewer deals. We will pass on some companies that turn out to be exceptional. We will own a small, high-conviction portfolio rather than a large, widely distributed one. In exchange, we will have deep relationships with the founders we back, meaningful ownership positions that allow us to capture substantial returns from our winners, and a reputation for partnership quality that makes the best founders choose us over funds offering more capital on easier terms.
That trade-off has served our investors and our founders well over the six years we have been executing against it. Our ambition is to refine it further, not to abandon it for a model that offers breadth at the expense of depth. Precision is not a constraint on our ambition. It is the source of it.