Public-Evidence NSV Research Protocol · v2.0

How Signal scores a company.

Signal assesses a company from the outside, using public evidence only. This page is the scoring method in full: what each dimension measures, how bands are assigned, and how the durability verdict is computed rather than asserted.

A Signal report is only useful if a different analyst, given the same evidence, would reach the same bands. Everything below exists to make that possible.

Relationship to Aisocrat’s broader methodology

Signal is a public-evidence research protocol derived from the Aisocrat NSV Core. It shares that foundation with Compass and Engine but uses different dimensions, evidence standards and outputs, because it addresses a different unit of analysis: Compass assesses a decision episode from the inside, Signal assesses a company from the outside. The Compass and Engine diagnostic architecture is proprietary and is not described here. Signal bands and verdicts are not equivalents of Compass diagnostic results.

Six dimensions, three pairs

The forces

Traction · Friction

How much push, how much drag. Magnitude and breadth only — nothing about whether it lasts.

Observation

The character

Traction Quality · Friction Quality

What is each force made of? Does the growth compound, or is it rented? Does the friction teach, or only cost?

Interpretation

The source

Economic Locus · Claim Integrity

Who pays for the motion — the business, or someone else? And can the company’s account of itself be relied on?

Structural

The pairing is the point. A force and its character are independent: traction can be enormous and largely rented, friction can be severe and almost entirely productive. Collapsing the two produces a score that means nothing — “strong traction” tells you a company is big, not that it will still be big in three years.

The dimensions

What each one asks, and what it refuses to ask.

01

Traction

How much unprompted demand is this company absorbing, and across how many independent surfaces?

Size, growth shape and breadth. Growth that is anomalous for the revenue scale already reached counts for more than large absolute numbers growing at category rate. Traction says nothing about whether the demand lasts — that is scored separately.

Very strong · Strong · Moderate · Weak · Negligible

02

Traction Quality

If capital, discounting and paid distribution stopped for four quarters, how much of this demand would still be here?

Two tests. The subsidy-withdrawal test separates demand that survives without continuous payment from demand that does not. The price-war test reads direction under pressure: share gained while prices are falling is the one kind of demand that cannot be bought.

Very strong · Strong · Mixed · Weak · Rented

03

Friction

How much force is currently working against this company, and where does it come from?

Six standing categories: financial, competitive, legal and regulatory, organizational, concentration, and political. Current drag only — a resolved lawsuit is history, not friction.

Severe · Strong · Moderate · Low · Minimal

04

Friction Quality

At the end of a period of friction, does the company hold a capability it did not have before — or a hole where something used to be?

Each friction item is productive, neutral, or destructive. One rule overrides commercial rationale: friction that removes the ability to sense problems or stop the company is always destructive, whether the capability is lost through attrition or through redefinition.

Strong · Mixed · Weak · Destructive

05

Economic Locus

What share of forward motion does the business generate itself, and is that share rising or falling?

The only dimension reported as two values — a level and a direction. Direction always refers to the internal share. Primary measures: external capital consumed per dollar earned, whether margin gains come from engineering or from pricing, and whether absolute burn is falling while revenue grows.

Internal-driven · Internal-emerging · Mixed · External-dominant · External-dependent

06

Claim Integrity

When the company’s account of itself has to be corrected, which direction does the correction travel — and does the company deliver it, or does someone else?

Three tests: direction of correction, mode of arrival, and whether superseded claims are withdrawn or left standing. A voluntary revision against the company’s own interest is the strongest positive signal available here. This scores the company, not the quality of our evidence.

Strong · Mixed · Strained · Weak · Broken

Deliberately not scored

Resources. Balance sheet, compute, headcount and brand are inputs to Economic Locus, not a score in their own right. A large war chest is not strength; it is what an external locus is made of.

Evidence quality. How much we can see is carried by the per-dimension confidence bands, not by a dimension. Confusing it with the company’s own honesty is what Claim Integrity exists to prevent.

Valuation. Never scored. It is context and a watch item. Signal assesses the company, not the price.

The durability verdict

Computed, not asserted.

Score all six dimensions first. Only then open the rule table. Tests apply in the order below and the first one met is the verdict. The order is deliberately terminal-first: a company can post excellent Economic Locus and Traction Quality readings while its coherence is coming apart, and the destructive tests must resolve before the constructive ones get a chance to.

1

Collapsing

Economic Locus external-dependent AND the external source is withdrawing — or a single terminal exposure is crystallising.

2

Fragmenting

Friction Quality is Weak or Destructive — the destructive share dominates. Applies regardless of how strong Traction is.

3

Compounding

Traction Quality is Strong or better AND Economic Locus is Internal-emerging or Internal-driven.

4

Coasting

Traction is Strong or better AND Economic Locus is Mixed, External-dominant or External-dependent.

Every report publishes this table with its own results, plus why the verdict is not one band higher, why it is not one band lower, and what specific observable change would move it. A verdict stated without its nearest alternatives is not falsifiable.

Evidence standards

No band is scored above the confidence of its evidence.

High

Audited or filed figures, or a company statement reconciled against an independent primary source.

Medium-high

Consistent reporting from two or more established outlets, or a single primary document.

Medium

A single credible source, or an unreconciled company statement.

Low

Inference from indirect indicators. Disclosed, and avoided for banding wherever possible.

Where figures are disputed across sources, reports cite the most conservative well-sourced figure and disclose the range — never the most favourable number in either direction. Third-party aggregator estimates are downgraded, and where estimates of the same quantity vary by more than a factor of two they are not used for banding at all.

The same foundation, a different job

Signal scores companies from the outside. Compass starts from the same core question and applies it to the decision you are actually stuck on, returning a structured brief in about ten minutes. Different protocol, different output — free, no account.

Run Compass →

Signal is a directional monitoring tool. It is not an audit, a legal finding, an investment recommendation, or a substitute for structured due diligence. Dimension definitions, band vocabularies and verdict rules are versioned; any change to them requires a version increment and a comparability note in every report that spans the change. Back to the NSV core.