When One Person Tracks Purchase Results, the Record Leaves Too

← Back · ACT III · Chapter 15: The Value Lattice System

When value depends on one named person, it leaves when that person does. A standard keeps working after they go.

A research note posted to SSRN in May this year argues that when AI agents do most of the operational work, human judgment should be “concentrated at named coherence points rather than distributed across functional hierarchies.” It presents itself as an extension of the organizational-design canon, building on Mintzberg, Galbraith and Laloux. On the question of value, it points the wrong way.

The proposal puts authority in a person. You name the point, you put someone in it, and from then on the hard calls go there. That feels like discipline. What it builds is a single point of failure.

Value that depends on a named holder lasts only as long as that holder’s attention, tenure and standing. Move the person and the value moves too.

A Named Point Is a Point of Failure

Governance that concentrates judgment at a person creates a dependency the enterprise cannot see on its balance sheet. The value produced under that person is real, but the custody of it is only ever borrowed. When the named holder leaves, is reorganized, or is simply overloaded, the judgment that governed the outcome leaves with them, and nobody is left managing the outcome.

The pull toward one point is easy to understand. A messy set of requirements gets boiled down to a few clean ones, and the clean ones are handed to a single owner. Each step makes the picture simpler, and at the end there is one point of failure with a title on the door.

That is why a reorganization can reopen the question of what value means. The point moves, the authority goes with it, and so does the definition of what counts, because custody was put in a chair.

A Standard Governs by Default

A standard does not ask who holds it. The definition of value, the terms under which it is claimed, and the test it must pass are written into the instrument, not into a person’s judgment. Anyone can read it, apply it, and be checked against it. No one has to be present for it to bind.

Appointing someone to keep things coherent works until that person changes jobs. A written standard needs nobody in particular in the seat next year, because a document isn’t affected by who leaves or arrives.

Buyer-Governed, Not Point-Governed

The deeper question is who the standard answers to. Named-point governance answers upward, to whoever drew the org chart. A value standard answers to the buyer, because the buyer is the party whose outcome the value is meant to serve. Custody sits with the buyer’s claim, not with an internal node appointed to judge it.

Concentrating judgment at a named point keeps the definition of value inside the org chart, where it moves whenever the chart does. A buyer-governed standard puts it in a document that can be checked, and moving a person doesn’t move it.

A vendor that delivers gets something out of this too. The test it will be judged by stays put when its sponsor on the buyer’s side moves on, so a result it has already earned doesn’t have to be argued again with someone new.

The instinct to name a point and concentrate judgment there is understandable and, in an enterprise run by AI agents, tempting. But it rebuilds the oldest failure in value governance: authority you can relocate, and therefore lose. A standard removes the point, so value stays governed because the reference holds, not because someone is standing on it.

References

Anil Prakash Singh, “Coherence as an Extension of the Organizational Design Canon”, SSRN research note, written May 15, 2026, posted May 21, 2026 (abstract 6770624).

Keep reading: ACT III · Chapter 15: The Value Lattice System