Why Smart Teams Still Struggle at Scale

I have had the privilege of building teams from zero — literally founding a function, hiring the first researcher, and scaling to a multi-office operation supporting over $125 billion in assets. Along the way, I have watched brilliant people struggle not because they lacked talent, but because the organization around them was not built to let talent compound.

The uncomfortable truth is that intelligence does not scale. Systems do. And the gap between a smart team and a high-performing one at institutional scale is almost entirely about the systems — of governance, communication, and accountability — that surround the people.

The Small Team Illusion

When a team is small — three or four people, maybe five — everything works through proximity. Decisions happen in conversations. Context is shared implicitly. Everyone knows what everyone else is working on because the surface area is manageable.

This is intoxicating. Small teams move fast, produce outsized results, and develop a sense of invincibility. Founders and early leaders internalize this experience and assume it will continue as the team grows. It almost never does.

At around seven to ten people, the first cracks appear. Information asymmetry grows. Two people work on overlapping problems without realizing it. Decisions that used to be made in a hallway conversation now require a meeting, but nobody has formalized when and how those meetings happen. The team is still smart — often smarter, because you have been hiring well — but output per person starts to decline.

By the time you reach fifteen or twenty, the small team model is actively working against you. The implicit coordination that was your superpower has become your bottleneck.

The Three Failure Modes

In my experience, smart teams at scale fail in three predictable ways.

The silo trap. In financial institutions especially, the natural organizational lines — risk, technology, portfolio management, operations — create gravity wells that pull teams apart. Each group optimizes for its own metrics. Risk builds models that technology cannot deploy. Technology builds platforms that risk does not trust. Portfolio managers make decisions without the analytics that quant teams are producing in parallel.

I have seen this pattern at every firm I have worked with. The solve is not removing organizational boundaries — those exist for good reason. The solve is creating forcing functions for cross-functional alignment. Regular touchpoints where the CRO's priorities, the CIO's questions, and the technology team's roadmap are explicitly reconciled. Without these, even the most talented individuals end up building in isolation.

The data governance vacuum. Smart people with bad data make confidently wrong decisions. This is perhaps the most dangerous failure mode because it is invisible until something breaks.

When I joined HPS to help build the quant function, one of the first things we addressed was data governance — not because it was glamorous, but because nothing else could work without it. Accuracy and access speed improved by 70 percent, and that single investment unlocked everything downstream: better models, faster reporting, more trustworthy analytics. Without it, we would have had a team of talented researchers producing analyses that no one could rely on.

Data governance is unsexy. It is also non-negotiable. Any leader who skips this step because the team is "smart enough to work around it" is building on sand.

The accountability gap. In small teams, accountability is natural. Everyone sees what everyone produces. At scale, accountability requires explicit structure — clear ownership of outcomes, defined metrics, and a culture where missed targets are discussed openly rather than obscured by complexity.

The hardest version of this problem is cross-functional accountability. When a risk model underperforms, is it because the model was wrong, the data was stale, or the implementation was flawed? In a siloed organization, each team points to the others. In a well-governed one, there is a clear framework for diagnosing and resolving these questions.

Building the Connective Tissue

The answer to all three failure modes is what I call connective tissue — the processes, governance frameworks, and cultural norms that allow individual brilliance to compound rather than collide.

Shared context rituals. Weekly or biweekly sessions where every function presents its current priorities, blockers, and dependencies. Not status updates — priority alignment. The goal is not information sharing for its own sake; it is ensuring that when the CRO asks a question on Friday, the quant team, the data engineers, and the portfolio analytics group are already pointed in the same direction.

Data as a first-class asset. Treat data governance with the same rigor you apply to investment governance. Define ownership, quality standards, access protocols, and escalation paths. Invest in it early and continuously. The ROI is invisible in the short term and transformative over years.

Outcome-based ownership. Assign individuals or small teams to outcomes, not tasks. Instead of "build a default prediction model," the mandate becomes "reduce realized loss rates by 20 basis points within 12 months." This forces cross-functional collaboration because no single team controls every lever.

Mentorship as infrastructure. At scale, the leaders you develop matter more than the leaders you hire. When we built the quant credit function, internal promotions — people who grew from junior researchers into team leads — became our most reliable source of institutional knowledge and cultural continuity. A 20 percent internal promotion rate is not just a retention metric; it is a sign that the system is developing talent, not just consuming it.

The Compounding Effect

When these systems work, the results are nonlinear. A team of five, properly connected and governed, can support a $125 billion platform with the responsiveness of a startup and the rigor of an institution. Risk reporting that once took days collapses to hours. Models that once sat in notebooks reach production. Insights that once stayed trapped in one analyst's head become organizational knowledge.

The irony is that the best-performing teams at scale often look effortless from the outside. That effortlessness is not the absence of structure — it is the presence of so much structure that the structure itself becomes invisible.

Intelligence gets you in the door. Systems determine whether you stay.