The Stranger Tax

Founding teams that included at least one co-founder the other members had never previously worked with raised fifty percent more crowdfunding capital than teams composed entirely of people who already knew each other, and were significantly more likely to be defunct or inactive when researchers followed up two years later. The study, published in the Journal of Business Venturing and built on data from more than three thousand Kickstarter companies, also found no evidence that the stranger co-founders contributed more varied or deeper skills than founders who already had relationships with their partners. The network advantage was measurable, the skills advantage was a fiction, and the failure arrived on schedule.

This is the structure of the stranger tax: you pay it not at the beginning, when everything looks like opportunity, but in the operational middle of the company, when execution requires the kind of relational infrastructure that takes years to develop and cannot be manufactured by shared enthusiasm about the market.

What the Early Signal Is Actually Measuring

The stranger brings something a long-time collaborator cannot easily replicate, which is the appearance of externally validated choice. When two founders who have known each other for a decade announce a venture together, the market reads it as what it is: people who already trust each other working on something they believe in. When a founder brings in a partner they met at a conference eight months ago, someone with a different network and a different pedigree, the market reads it as broader relevance, as cross-pollination, as a wider tent. The initial raise reflects that reading, which is why the fifty percent advantage in early capital shows up so cleanly in the data.

What it does not reflect is whether these two people can disagree productively in month seven, or hold each other accountable through a quarter when the numbers have stopped moving, or navigate a structural conflict in the agreement they papered over when they were still excited about starting. Those capabilities are built from shared history, from watching someone perform under pressure, from knowing how they behave when they are scared and when they are wrong. None of that can be established in the first year of a stranger partnership, and none of it shows up in a crowdfunding campaign.

The researchers were explicit about what the strangers did and did not bring. They brought wider networks. They did not bring more varied or deeper skills. And the founders in the teams that failed reported, in their own words, that the cause of delays and failures was difficulty of collaboration within the team. They selected for reach and got coordination failure in return, which is not a surprising outcome when you understand the mechanism, but is a pattern that founders continue to produce because the early signal is so legible and the cost is so deferred.

When Execution Demands What Selection Ignored

A founding partnership in its first six months is solving a fundamentally different problem than the same partnership in month eighteen. In the early phase, the problem is access: access to capital, attention, distribution, and credibility. A stranger co-founder helps with that because their networks do not overlap with yours, their reputation complements yours, and their presence in the partnership reads to the market as diversified conviction. The partnership is performing outward at this stage, and strangers perform well outward.

In month eighteen, the problem is internal. Who makes the call when the two of you disagree about the product? Who says the thing that needs to be said when the company is struggling and saying it might damage the relationship? Who holds accountability without defaulting to blame when a commitment was missed? These conversations require relational infrastructure that takes years to build, and a founding team of strangers enters that phase with almost none of it. They have the network benefit of their first six months and the relational deficit of their entire history together, which is to say, almost no shared history at all.

The founders who built something durable together typically describe their partnerships as boring to explain: they knew each other for years, had worked through something difficult already, had seen each other be wrong and recover from it. That does not make for a compelling stage talk and it does not produce a fifty percent increase in early backers, but it is the structure that predicts whether the company survives the build.

The Selection Error the Market Rewards

The deeper problem is that the market trains founders to repeat this error. Stranger partnerships raise faster in the early phase, which produces a feedback signal that reads as confirmation: the partnership is working. Investors respond. The room responds. The founder interprets that response as evidence of a good selection decision, rather than evidence of a good network-expansion decision, which are not the same thing and rarely produce the same outcomes.

The question that was never asked during selection, because the raise went so well, is whether this partner has a track record of delivering through friction, of operating with integrity when external pressure disappears, of telling the truth in low-stakes moments before it becomes necessary in high-stakes ones. These are the signals that predict whether the company gets built, and they are almost entirely absent from the early fundraising phase that founders use as their primary form of partnership validation.

onSpark is built around the evaluative infrastructure that makes this assessment possible before the partnership is formed, because the data is unambiguous that the moment after the deal closes is too late to discover what the selection process should have found.

The stranger who walked in with the wider room was the easier choice. They made the raise feel like proof. The partner worth having required more scrutiny, less excitement, and a longer evaluation window, which is exactly what founders in a hurry eliminate first, and exactly what the research shows determines whether the partnership survives the build.