Formulas are copyable in an afternoon. A consistent record is not.

Starting from the concession

Inspectability and defensibility pull in opposite directions, and inspectability won.

A system that insists every formula be traceable by the person it governs has already given up on the formulas as a defensible asset. That trade was made deliberately and the reasoning is set out elsewhere: an operator honours a constraint they can trace and quietly overrides one they cannot. The consequence is that a competent builder could reconstruct the calculation layer from the published descriptions in a fairly short time. What they would then possess is an empty instrument, and the gap between an empty instrument and a useful one is where the actual durability sits. That gap is measured in periods rather than in engineering effort, which is why it cannot be closed by building faster.

What the record consists of

Three assets, all of them time-denominated.

The first is a run of periods captured under unchanged branch definitions, which is what makes attribution meaningful — a branch's expectancy is only comparable across time if the branch meant the same thing throughout. The second is a benchmark calibrated to that specific mix rather than to a generic assumption, so deviation readings describe this operation. The third is a promotion history: the changes proposed, tested, compared against baseline, approved or rejected, which converts a set of current rules into a set of rules with reasons. None of the three can be assembled quickly, and none can be bought.

Whose asset it is

The record belongs to the operator, which is an unusual place for a moat.

This matters more than it first appears. The compounding asset does not sit with whoever built the workbooks — it sits in the operator's own file, accumulating on their machine, describing their account. That has a pleasant consequence and an uncomfortable one. The pleasant version is that the value of the system to a given operator rises with every period they run it, independent of anything the architecture does. The uncomfortable version is that a new operator's first quarter is the weakest the system will ever be for them, and no amount of sophistication in the instruments compensates for having three months of history.

The switching cost runs both ways

An asset built on stable definitions is damaged by changing them.

A record whose value comes from consistency is, by construction, hostile to revision. Redefining what counts as a Trend Partial does not merely affect future periods; it severs comparability with everything before, and the operator has traded a genuine asset for a marginal improvement in categorisation. This is why branch definitions are treated as near-immutable while thresholds and coefficients are treated as tunable, and it is also a real constraint on the operator rather than a feature — the system becomes less able to accommodate a fundamental change in how they think about their own strategy the longer it has been running.

• The calculation layer is published and therefore not defensible; that was chosen.
• The compounding asset sits with the operator, not with the vendor.
• Consistency-derived value makes definition changes genuinely expensive.

What this predicts

The system should feel thin early and disproportionately useful later.

If the account above is right, the experience of adoption has a specific shape, and it is not a flattering one to advertise. Early periods produce readings with wide uncertainty, a benchmark not yet fitted to the operator's real mix, and diagnostics that cannot distinguish variance from anything because there is no history to compare against. The instruments look expensive relative to what they return. The curve is steep afterwards, and an operator who abandons in month three has paid the full cost of the record while collecting almost none of its value.

The key idea

Durability that comes from time cannot be sold, only accrued.

This is the least commercially convenient version of the argument and the most accurate one. There is no configuration of the workbooks that shortcuts a two-year evidence base, no premium tier that supplies one, and no acceleration available to a buyer willing to spend more. What is being purchased is the apparatus and the discipline for accumulating something that only accumulates. That framing sets expectations correctly at the point of sale, which matters, because the most common way this value is lost is an operator stopping just before it starts to arrive.