Doing things that don't scale isn't a startup hack. It's a leadership principle.

People treat "do things that don't scale" as a temporary startup trick, to be abandoned once the company grows. It is the opposite: it is where the leader touches reality, learns what is real and builds trust before automating. The unscalable is not a phase. It is a permanent discipline.

The idea has become popular that, at the start, a company should do things that don't scale — serve customers manually, one by one, with effort that would not be sustainable at scale. It is treated as a startup trick: an initial phase to be overcome as soon as the company grows and can automate. That reading is incomplete and misses the most important thing. Doing things that don't scale is not a temporary hack; it is a permanent principle of leadership. The unscalable is not a phase to abandon; it is a discipline to keep.

Start by asking why the unscalable works at the start. When you serve each customer manually, you are not just getting that customer; you are learning. You see up close what they want, what bothers them, how they react — rich, direct, unfiltered information that only manual contact provides. The effort that does not scale is, in fact, an instrument of learning: it puts you in direct contact with reality, before any system, metric or automation interposes itself between you and what is true. The unscalable is expensive in effort and very rich in learning.

Here is why this should not be abandoned when you grow. The temptation, when growing, is to automate everything, replace manual contact with systems, metrics and processes. That brings efficiency, but it has a hidden cost: each layer of automation distances the leader from direct reality. Metrics summarize, systems filter, processes mediate. The leader who only sees reality through those layers loses the direct contact the unscalable provided — and loses, with it, the ability to perceive what the layers do not capture. The unscalable should not be abandoned when growing; it should be deliberately preserved, precisely to keep the contact with reality that scale tends to erase.

Here is the mechanism that makes this a principle of leadership, and not just a tactic. Leading well requires understanding reality — what customers want, what works, what is broken. Scale, with its layers of automation and metrics, distances the leader from that direct reality, giving them a summarized and filtered picture. Deliberately doing some things that don't scale — serving some customers personally, doing some part of the work manually, keeping some direct contact — is what keeps the leader connected to the reality the layers hide. It is not nostalgia or inefficiency; it is the discipline of not letting scale blind the leader to what is real.

See the asymmetry between what scales and what does not, in terms of learning. What scales — systems, metrics, automation — is efficient at executing, but poor at teaching; it summarizes and filters, hiding the texture of reality. What does not scale — manual, direct, personal contact — is inefficient at executing, but very rich at teaching; it exposes the texture the summary erases. That is why the leader needs both: the scalable to execute at scale, and the unscalable to keep learning what the scalable does not reveal. Abandoning the unscalable when growing is to trade all the learning for efficiency, and to go blind in the name of scale.

There is also a dimension of trust that the unscalable builds and the scalable does not. Direct, manual, personal contact builds a relationship and a trust that automated systems do not build. When the leader does something that does not scale — serves personally, solves a problem directly, is present — they build a trust no automation reproduces. That trust is an asset scale tends to erode, replacing the personal relationship with the automated transaction. Keeping some unscalable is keeping alive the building of trust that scale, alone, would let die.

This connects to threads we had been pulling. Learning speed as an advantage: the unscalable is a source of direct learning that scale does not provide. The company that goes blind by distancing itself from reality: the automation that distances the leader from the real is the leadership version of that blindness. The unscalable is the antidote against the blindness scale brings — the way for the leader to keep contact with the reality systems summarize and hide. It is a principle of leadership precisely because it preserves what scale, left alone, would make the leader lose.

It is fair, in balance, to recognize that not everything should be unscalable, and that scale is necessary and valuable. A company that refuses to automate, insisting on doing everything manually, does not grow and does not survive. The point is not to reject scale in favor of the manual, but to recognize that the unscalable has a value — of learning and of trust — that should not be entirely sacrificed when you automate. The wise leader scales what should be scaled and deliberately preserves some unscalable, not out of inefficiency, but to keep the contact with reality and the trust only the unscalable provides. The balance is not all manual nor all automated, but scaling without losing the direct contact the unscalable keeps.

For the leader and the investor, this suggests distrusting total automation as a sign of maturity. A company whose leader has completely distanced themselves from direct reality, seeing everything only through metrics and systems, may look mature and efficient, but is blind to what the layers hide. A leader who, even in a large company, deliberately preserves some unscalable contact with reality — some customers served personally, some part of the work done by hand — keeps a connection to the real that makes them a better leader. The question about a leadership stops being 'has it automated everything?' and becomes 'has it preserved some direct contact with reality, or has it let itself be blinded by scale?'.

The rule of this moment: doing things that don't scale is not a trick to be abandoned when you grow, but a discipline to be kept, because the unscalable is where the leader touches reality, learns what is true and builds trust. Scale brings efficiency and blindness in equal measure; the unscalable is the antidote against the blindness. The leader who abandons all the unscalable when growing gains efficiency and loses contact with the real, and that contact is what distinguishes whoever leads from whoever merely administers summaries.

Doing things that don't scale is not a startup hack. It is a leadership principle. The unscalable — personal attention, manual effort, direct contact — is where you learn what is real and build trust, before automation interposes itself, and the leader who preserves it keeps a contact with reality that scale tends to erase. Mark this essay not as tactical advice for startups in their early days, but as a permanent principle of leadership — the lesson that scale, left alone, blinds, and that deliberately doing something that does not scale is how the leader keeps their eyes open to the reality systems summarize, filter and hide.

Leo Bentier