The triangulation chain, rung by rung

A company saying it is moving up the value chain costs nothing, carries no penalty if it never happens, and reliably lifts a share price for a quarter or two. So we do not treat the statement as evidence. We treat it as the first rung of a ladder, and the question that matters is how far up the ladder the claim has actually climbed. The triangulation chain is that ladder: stated, approved, funded, permitted, protected. Each rung is a higher standard of proof than the one below it, and — this is the part that makes it work — each is recorded with a different counterparty who gains nothing from flattering the company’s story.

Why five rungs rather than one test

The reason a single test fails is that every individual signal has a cheap counterfeit. A management team can describe a transformation in warm, specific language for years. A board can approve a strategic direction that commits nobody to anything. Capital expenditure can rise because a furnace needed replacing. A patent can be filed on something nobody will ever build. Taken one at a time, each of these is noise. Taken in sequence, they are expensive to fake, because faking rung four requires having already faked rungs one through three in a way that is consistent across documents filed months apart with regulators, lenders, and a patent office that does not care about the equity story.

The sequence also has a natural direction. Money does not get spent before somebody approves the spend; a permit is not sought for a facility nobody intends to fund; intellectual property is not defended around a product line that does not exist. Because the rungs are ordered, the gap between where a claim sits and how long it has been sitting there becomes readable. That gap is the single most useful output of the whole exercise.

The rungs

Stated. Somebody with authority has said it, in a document that carries liability. A line in the business section of an annual report, a direct answer on an earnings call, a strategy slide filed as an exhibit. The bar here is only that the claim is specific enough to be falsified later — “we are shifting toward higher-value products” is not a statement, it is a mood. “We expect specialty to exceed half of segment revenue” is a statement, because in three years it will be either true or false.

Approved. A body other than the executive making the claim has ratified it. Board approval of a capital programme, a shareholder vote on an asset purchase, a joint-venture agreement signed and filed. Approval is where the first real constraint appears: the claim now exists in a document that a director’s name is attached to.

Funded. Money is committed, and the commitment is attributable to the project rather than merely simultaneous with it. This is harder to read than people expect, because maintenance and growth capital are not separate line items under US GAAP — you get one capital-expenditure figure and are left to work out what it bought. The readable forms are a disclosed project cost, a drawn facility with stated use of proceeds, a material definitive agreement filed on a current report, or a construction-in-progress balance that grows in step with a named plant. Absent one of those, treat a rising capex number as consistent with the claim rather than as proof of it.

Permitted. An external authority has granted whatever consent the activity requires, and that authority has its own reasons for saying no. Environmental and construction permits for a plant; a regulatory approval or registration for a product; a certification or qualification for a component going into somebody else’s system. This rung is valuable precisely because the granting body is indifferent to the share price. It is also where timelines go to die: permit queues are slow, and a transition that has been “awaiting approval” for three years is a different thing from one that cleared its consents last quarter.

Protected. Once the higher-value position exists, something has to stop it being competed away. Patents and their continuations, regulatory exclusivities, customer qualifications that take years to win and are painful to switch away from, long-term supply agreements, and process know-how that is not in the public domain. Without this rung a company can genuinely complete a climb and still end up with commodity economics, because the differentiation it built turns out to be copyable. This is the rung that connects the chain to the moat in our vocabulary, and it is the one most often skipped by analysts who stop counting once the plant is built.

Reading the chain on a real filing

The practical version of this is a table you fill in per claim, not per company. A business with three announced transitions may have one at rung five and two stalled at rung two, and averaging them destroys the information.

Rung	Where the evidence lives	What counterfeits it
Stated	Annual report business section, earnings-call transcript, filed strategy exhibit	Vague direction-of-travel language with no falsifiable claim
Approved	Board or shareholder authorisations, joint-venture and definitive agreements	A strategy endorsement that commits no resource
Funded	Disclosed project cost, construction in progress, use-of-proceeds language, material agreements	A capex line that rose for unrelated maintenance reasons
Permitted	Environmental and construction consents, product registrations, customer qualifications	An application filed, which is not an approval granted
Protected	Granted patents and continuations, exclusivities, long-term contracts, switching costs	Filings with no grant, or protection on something adjacent to the actual product

Two habits make this reliable. The first is to date every entry. A rung reached is only meaningful against when the claim was first made, and a dated ledger is what lets you say “approved in year one, still unfunded in year four” rather than “approved and unfunded”, which sounds much better. The second is to record the document, not your summary of it — the specific filing and the specific passage. A chain built from recollection drifts toward the story the company tells; a chain built from citations does not.

What the chain cannot do

It does not tell you whether the transition is a good idea. A company can climb all five rungs into a market that turns out to be smaller than it believed, or arrive with a differentiated product nobody needs. The chain measures execution reality, not strategic judgement, and pairing it with an honest read of the end market is not optional.

It is also poor at speed. Some genuinely fast transitions — a licensing deal, an acquisition of an already-qualified business — compress three rungs into a single quarter and can look suspiciously abrupt against a framework built for plant-scale timelines. And it says nothing about price: a fully triangulated, fully protected climb that the market has already re-rated is a fine business and a poor opportunity, which is why the chain sits alongside the valuation half of the migration signature rather than replacing it.

Finally, it depends entirely on disclosure quality. In markets where segment reporting is coarse and project-level detail is scarce, the funded rung in particular becomes difficult to evidence — one reason our coverage is sequenced market by market rather than applied uniformly.

The bottom line

Announcements are free; the chain makes them pay rent. By asking where a specific, dated claim sits across five independently recorded rungs, you replace the question “do I believe this management team?” with the much more answerable “how far has this particular claim actually travelled, and how long did it take?” That reframing is most of the work. The rest is the credibility ledger that records what the same team did with its last set of promises, and the full method that puts both alongside the financial baseline everybody already has.