MoatOS 6.0 — How We Score Moats
Every moat claim on Moat & Margin comes out of one instrument: MoatOS, our moat-identification system, now in version 6.0. This page explains what it measures, how it refuses to be fooled, and what its outputs mean — so that when a research note says a company scores well on price discretion, you know exactly what that sentence cost to produce.
The core principle
A moat exists only if a company can sustain returns on invested capital above its cost of capital because competitors cannot replicate its advantage — even with time, capital, and intent.
A moat is proven when competitors try and fail — not when the company succeeds.
Growth is not a moat. High margins are not a moat. A famous brand is not automatically a moat. Each of those can exist for a while without any structural protection at all. MoatOS is built to tell the difference.
Before anything is scored: the archetype
Moats come in types, and each type should leave a different fingerprint. Every company is first tagged with a primary archetype — distribution-led, brand-led, vertical-integration-led, industrial-OEM-led, network-effects-led, regulated/licence-led, or process/data-led. The tag is a testable expectation: it predicts which dimension should lead the score and prescribes an archetype-specific stress test (a brand must show a realised price premium; a vertically integrated producer must show margin stability through input-cost spikes). If the evidence ends up pointing somewhere else, the analysis must explain why before proceeding — that mismatch is often the first sign of a misread moat.
The seven dimensions
MoatOS scores seven distinct mechanisms by which returns persist. Each is scored separately, on its own evidence:
| Dimension | What it measures |
|---|---|
| D1 — Network effects | Does each additional user make the product more valuable to every other user? |
| D2 — Switching costs | How expensive — in money, risk, or disruption — is it for a customer to leave? |
| D3 — Cost advantage | Can the company structurally produce at lower cost than any challenger — through scale, integration, or a process rivals have tried and failed to copy? |
| D4 — Price discretion | Does the company control its price, rather than having price imposed on it? (Both directions count — see below.) |
| D5 — Intangibles & cornered resources | Brand trust, IP, exclusive licences, locked-up supply — counted only where they change customer behaviour. |
| D6 — Efficient scale | Is the market too small to reward a second entrant at efficient scale? |
| D7 — Counter-positioning | Does its business model force incumbents to choose between responding and protecting their own economics? |
An eighth dimension — D8, Ungoverned Risk — is tracked but never added to the score: structural threats largely outside the company's control, such as regulation, export restrictions, or concentrated input dependencies, which can breach a moat without any competitor succeeding. D8 shapes the discount rate, the duration view, and position sizing — not the moat score. Capital allocation is likewise deliberately unscored: it preserves moats rather than creating them.
Price discretion cuts both ways
The classic evidence for pricing power is raising prices without losing volume. But some of the strongest moats in the world never raise prices: deliberate price restraint, funded by structural cost leadership, that earns superior returns anyway while competitors bleed — the discount-retailer playbook. Earlier frameworks (ours included) structurally failed that entire archetype. MoatOS 6.0 scores price discretion in two modes — willingness-to-pay (raise prices, hold volume) and willingness-to-sell (hold prices down by choice, earn superior returns regardless) — and a company must convincingly pass one. Fail both, and no other strength can lift the verdict past narrow.
The evidence rules
Dimension scores are easy to inflate, so 6.0 binds them with four rules:
- One fact, one dimension. Every piece of evidence is allocated to the single dimension whose mechanism it demonstrates. The same distribution network cannot count four times.
- Outcomes are not mechanisms. High ROIC, fat margins, and market share are what a moat must explain — never what proves it. Outcome-only evidence is hard-capped, however authoritative its source. Without this rule, a moat framework is just a momentum screen with extra steps.
- Mechanism needs a footprint. A claimed advantage that leaves no measurable residue in the financials is a hypothesis, not a moat. Story without footprint caps the score.
- Every wall must be a wall. Any highly scored dimension must state, in one sentence, the benefit and why competitors specifically cannot copy it.
The narrative evidence feeding those scores is held to a fail-closed standard: verbatim management and filing quotes, character-verified against the source document, with locator attached. A quote that cannot be verified is dropped, never paraphrased into existence. Where a dimension has no direct evidence, the note says so — "no direct evidence of switching costs found" is a published finding, not a gap we paper over.
The gates
- False Moat Filter — accumulated disqualifying evidence forces a NO MOAT verdict outright.
- Barrier Gate, asked twice. Replication: can a well-funded competitor copy this advantage within five years? Bypass: can they make it irrelevant without copying it — by changing the basis of competition entirely? The worse answer governs. Attackers don't always climb the wall; sometimes they move it.
- Price Discretion Gate — dual-mode, as above. Not overridable.
The gates are the point. Most companies we examine fail them — which is what you'd expect if moats are actually rare.
What the output looks like
A full MoatOS run produces a dimension-scored profile aggregated dominant-source style (a moat is one primary mechanism plus support, not seven mediocre scores), a gated verdict — no moat / narrow / wide — a moat trend (widening, stable, or eroding, with the driving dimension named), a duration view in years, and the archetype tag. Verdict bands never include buy or sell: moat classification and investment action are separated by design, and the valuation side lives in ValuationOS.
The whole system runs deterministically against a locked interpretation spec, with a standing acceptance test: two independent runs on the same company, from the same evidence set, must land within half a point on the final score. The most recent verification pair landed 0.02 apart. Same evidence, same framework, same score — that is the standard behind every number we publish.
MoatOS is a research instrument, not an advisory service. Its outputs are educational analysis — never buy/sell recommendations or price targets. See the ground rules.