Fair Isaac, Layer 1: Moat Analysis

MOAT VERDICT

Field	Verdict	Why
Strength	Exceptional	Coordination lock, generic brand, zero marginal cost, infrastructure role
Security	Low	A coordinator exists. FHFA can reassign the standard administratively
Substitution	Permitted, not yet adopted	VantageScore approved since 2025. Negligible share of actual closings
Competitive test	Held nineteen years, now retested	Bureau owned rival failed on capability, then was granted permission
Pricing power, demonstrated	Exceptional	Royalty $0.60 to $10.00 in five years, no volume loss
Pricing power, current authority	Actively contested	Regulator named $0.99. Competitor free. Senate and FTC engaged
Condition	Partially impaired in conforming and FHA mortgage. Intact elsewhere	Collective lock became individual switching costs in that channel only
Erosion, price vector	Accelerating sharply	Free bundling sets dual pull cost to zero
Erosion, standard vector	Not yet begun	Freddie Mac VantageScore volume $10 million. A rounding error
Register	1 triggered, 1 partially triggered, 4 clean	4a fired July 2025. Damage conditions 1, 2 and 6 clean
Failure mode	Administrative reassignment of the standard	Already fired. The reject gate rather than a future risk
Tier	Rejected. Zero weight	Foundational layer impaired. Would score Tier C regardless
Decision	Rejected. Not monitored. No Layer 2 at any price	Mandate requires an intact moat. Damage disqualifies regardless of price

In one line: an exceptional business whose moat was structurally damaged by government action in its most valuable channel, and which is therefore outside this mandate despite reporting the strongest economics in its history.

How the tier is set. The tier is derived mechanically from three fields, so that it cannot be adjusted after the fact: Security, Condition, and current pricing authority. Strength is excluded because everything surviving the screen scores exceptional and it therefore does not discriminate between names. Demonstrated pricing power is excluded because it is historical. Each of the three is scored on a three step scale: high, moderate or low for security; intact, partially impaired or impaired for condition; unconstrained, constrained or actively contested for pricing authority.

No degradation on any of the three is Tier A at full weight. One or two fields degraded a single step is Tier B at two thirds weight. Three fields degraded, or any single field degraded two steps, is Tier C at one third. Impairment of a foundational or reinforcing layer is a rejection regardless of what the tier would otherwise be.

Condition is scored on the foundational and reinforcing layers only. Damage to a layer this framework classifies as optional is recorded but does not move the tier, because the thesis by construction does not require that layer. That rule carries an obvious hazard, since it creates an incentive to reclassify a damaged layer as optional after the damage occurs. The guard is procedural rather than analytical: a layer may be treated as optional only where it was classified that way in a revision written before the damage, never in the revision that reports it.

Two independent routes produce the same outcome here, which is worth recording because agreement between them is a check on the framework rather than a restatement of it. The reject gate fires directly, because the impaired layer is foundational. And the tier, computed as though the gate did not exist, would be C: security is degraded two steps to low and pricing authority two steps to actively contested, either of which alone is sufficient. A name that fails on the gate and would also score the lowest tier is not a marginal exclusion.

Decision rule applied. This mandate takes positions only in businesses whose moat is intact. A moat that is structurally damaged fails at Layer 1 regardless of how strong the current economics are, how attractive the price becomes, or how slowly the damage is expected to convert into outcomes. Condition 4a is triggered, the coordination lock in conforming and FHA mortgage has been substantially converted into ordinary switching costs by an actor outside the company's control, and that is disqualifying on its own.

The economics being excellent is not a mitigating factor and is not treated as one. Scores revenue growing sixty percent, mortgage scores growing one hundred and twenty seven percent, and margins expanding are exactly what a damaged mechanism looks like before the damage transmits, which is why the rule is drawn on the mechanism rather than on the output.

Re entry conditions. The rejection is not permanent by construction. Any one of the following would justify writing a fresh Layer 1 from scratch rather than amending this one.

Restoration of exclusivity, meaning permission for alternative models withdrawn in conforming and FHA lending, whether by a subsequent administration, by statute, or through the pilot being abandoned. This would rebuild the coordination lock rather than merely pausing its erosion.

Demonstrated inertness over a full cycle: by mid 2030, five years after the first permission, a competing score is still pulled alone in a negligible share of conforming and FHA closings, the score component pricing settlement has stabilised, and Scores segment margin has held in the eighties throughout. The competitive section below establishes that this outcome has a documented precedent in another industry and takes roughly five years to confirm, which is where the date comes from.

A change of subject, meaning software platform revenue grows to the point where Scores is no longer the thesis. The company would then be a different business with a different moat, requiring assessment on its own terms rather than reinstatement on these.

Absent one of those, the name stays out and is not tracked.

WHAT THE COMPANY DOES Fair Isaac was founded in 1956 by an engineer and a mathematician to sell analytic decision systems to lenders, an idea roughly two decades ahead of the computing capacity required to implement it well. The general purpose, bureau based FICO Score that now carries the company's identity arrived in 1989. The relevance of the origin is not sentiment. The company's competence was never scoring. It was the industrialisation of repeatable decisions under uncertainty, and the score is one exceptionally successful expression of that competence rather than the competence itself.

The business reports in two segments best understood as two companies sharing a balance sheet.

Scores is the licensing of the FICO Score. B2B, the sale of scores to lenders through the three national bureaus for mortgage, card, auto and personal loan origination and for account management, is the economic engine. B2C, reaching consumers through myFICO.com and through partnerships with issuers, is smaller and slower and strategically important out of proportion to its revenue, for reasons set out under the brand layer. Scores has grown to roughly two thirds of group revenue on mortgage repricing, from closer to half a few years ago.

Software comprises the FICO Platform and the legacy applications preceding it: decision management, originations, customer management, fraud detection, optimisation. Platform annual recurring revenue compounds in the thirty to fifty percent range with dollar based net retention well above one hundred and thirty percent, while non platform ARR declines. The segment nets to mid to high single digit growth, so the strategically important number and the reported number are different numbers.

How the money is actually made. The score is a mathematical model, and once built and validated the cost of producing another one is the cost of a computation. The company does not own, store or process the underlying credit file. The bureaus do. FICO supplies the algorithm, the bureaus supply the data and the delivery, and until recently the bureaus also set the price to the end lender and captured a substantial share of the economics.

That is a royalty business with essentially no cost of goods, in which the binding constraint on revenue is not capacity but the ability to set and collect a price through a channel one does not own. FICO's economics are therefore determined almost entirely by pricing authority, and its strategic history over the last five years reads as a campaign to reclaim that authority from the channel.

The channel problem. The binding input is access to the consumer credit file, controlled by three companies who are simultaneously FICO's distributors and, through joint ownership of VantageScore Solutions, the sponsors of its only competitor at scale. That arrangement has produced litigation before and is inherently unstable. The direct licence programme is the response: resellers in the mortgage tri merge channel contract with FICO directly rather than buying through bureau markup, which converts an opaque intermediated price into a published one and gives FICO a direct relationship with the parties who consume the score.

Where the revenue concentrates. Mortgage is a minority of score pulls and a majority of recent revenue growth, because per score pricing has been raised most aggressively there and a single loan file generates multiple scores across three bureaus. Card and auto are far higher in unit volume, far lower in unit price, and more stable. Account management pulls, in which existing portfolios are rescored periodically, are the least visible and among the most recurring. B2C sits outside the lending decision entirely and functions as brand infrastructure.

Two consequences. The company is more exposed to mortgage than the headline mix suggests, because mortgage carries disproportionate revenue per unit and disproportionate political salience. And the growth runway is not unit growth in the United States, where penetration is near complete, but price realisation, international expansion, and the platform. That runway is more dependent on the company's own pricing decisions and therefore far more visible to parties who might object to it. It is not a coincidence that the channel carrying the growth is the channel where the moat's composition has changed.

THE MOAT

The question is whether there is a durable reason Fair Isaac can charge what it charges, grow as it grows and earn what it earns, and whether that reason holds for fifteen to twenty five years. The layers are ranked, not listed flat: two foundational, three reinforcing, one optional, plus governance. Throughout, hold the distinction between mechanism and consequence. Pricing power, margins and returns are consequences and evidence; the mechanisms are the demand driver, the coordination lock, the consumer brand and the channel position.

Foundational: the permanent demand driver Lending exchanges certain present value for uncertain future value, and the lender always knows less than the borrower. Every credit system in recorded history has developed some device for compressing that asymmetry into a decision, whether guild membership, personal acquaintance, collateral, character letters or ledgers of prior conduct. This is adverse selection in the Akerlof sense, and the standardised score is its industrialised solution. Credit bureaus emerged independently in multiple national economies within decades of impersonal consumer lending appearing, which indicates the demand is a structural property of the activity rather than an artefact of any regulatory regime.

The need for a quantified, portable, third party assessment of consumer credit risk is judged permanent with high confidence, for as long as consumer lending at scale exists. The form that assessment takes is emphatically not permanent. It has already changed once, from human judgment to statistical scoring, and could change again, from a single reference number derived from bureau tradelines to a continuously updated assessment drawn from bank transaction data, or to a model with no interpretable single output.

The foundational claim is therefore limited: demand is anchored in a permanent feature of lending, conditional on the single reference number remaining the operative form of that assessment. Everything below inherits that conditionality.

Foundational: the coordination lock, and what happened to it This is the layer that does the work, and it is routinely mis described as switching costs. The distinction is not pedantic, because it is the entire reason the moat has been exceptional rather than merely good, and it is the thing that changed in 2025.

Switching costs are individual. A customer bears a private cost to leave, the cost is quantifiable, and it falls a little every year as tooling improves, vendors compete and familiarity spreads.

A coordination lock is collective. A lender wishing to underwrite on a different score would still have to sell the loan into a secondary market referencing FICO, have it rated by methodologies calibrated to FICO, service it under agreements referencing FICO, and explain the decision to a regulator and a borrower who both expect FICO. The score is not a product the lender consumes. It is the unit of account in which the lender's counterparties transact.

The difference shows up in behaviour rather than magnitude. Switching costs erode continuously. Coordination locks do not erode at all, because they are binary and self enforcing until something breaks them, which is why they produce pricing power ordinary switching costs never produce.

The mechanism also does three things a superior competing model cannot. It makes the incumbent's accumulated calibration a shared asset of the whole system, since decades of loan performance are expressed in FICO units and cannot be restated without cost. It converts every contract, guideline and covenant naming the score into a small distributed switching cost borne by someone other than FICO. And it forces a challenger to win the coordination of many independent parties rather than to win on model quality, which is a different and considerably harder problem.

The asymmetry that defined it. Coordination locks are broken by coordinators, and in this system a coordinator exists. The FHFA, the GSEs it conserves, and the FHA can move a large part of the chain by administrative decision. That asymmetry, durable against private competition and materially less durable against public decision, was always the defining structural fact about this business, and in 2025 it stopped being theoretical.

Reinforcing: the consumer facing brand The substitution test asks what the end user's second best alternative is. For most infrastructure software the answer is a genuine menu, because the end user is a procurement department indifferent to the vendor's name.

Here the end user of a mortgage is a borrower who has been told for thirty years that a specific three digit number determines the terms of their life, who monitors it through free consumer products, and who will not accept without explanation that a different number applies. No other credit score has achieved genericisation, and "my credit score" is understood by most American consumers to mean this one.

The scope needs stating precisely, because the strong version overstates it. Consumer familiarity does not prevent a lender using a different model internally and does not constrain the many credit decisions involving no consumer disclosure. What it does is raise the political and explanatory cost of substitution at exactly the moment substitution would be most visible, and give the company a channel to the consumer that does not run through the bureaus.

It is reinforcing rather than foundational because it is a position produced by decades of the foundational mechanism operating. It is also the layer a bureau sponsored competitor can least easily replicate, since it took thirty years and a consumer education effort nobody is now positioned to repeat. Its relative weight has risen since 2025, which is worth reading as both a strength and a warning: reinforcing layers become more valuable when the foundation weakens, and also more exposed.

Reinforcing: accumulated time and validated performance Two arguments here, frequently run together.

The predictive one: the company has sold analytic decisions to lenders for seventy years and the bureau score has been the operative standard for more than thirty, surviving the savings and loan crisis, the 2008 collapse and its aftermath, numerous alternative scoring ventures, and two prior bureau attempts at displacement. That record is the best available predictor the model survives the next challenge, though the inference is judgment rather than evidence.

The causal one, which is stronger: age is a literal ingredient of the product, because a score's value to a lender is a function of how much loan performance has been observed against it. A scale with three decades of through the cycle validation across tens of millions of loans is not merely older than a challenger, it is more informative, and the gap closes only with the passage of the same time. This is the one input a well capitalised competitor cannot purchase, and why regulatory validation moves as slowly as it does.

One qualification specific to the current situation. The Enterprises publishing historical scores for competing models is a partial, third party funded attack on exactly this advantage. It does not manufacture through the cycle performance data, so the causal claim survives. What it removes is the practical obstacle the causal claim was previously enforcing on the company's behalf.

Reinforcing: the reclaimed channel A standard without a price is a public good. The layer converting the standard into economics is pricing authority, and that authority runs through a channel owned by counterparties with an active interest in suppressing it.

Through 2025 the tri merge channel was intermediated, opaque and margin capturing. By 2026 a majority of the largest resellers had signed direct agreements and the price to the reseller was published rather than inferred, which materially reduces the risk that the bureaus expropriate the standard's economics by controlling the last mile.

The honest counterpart is that this is a commercial arrangement rather than a governance structure. It can be renegotiated, litigated, or overtaken by regulation of the price itself. It converts "the bureaus currently permit us to price" into "we price and the bureaus distribute," which is a real improvement, but a contractual one, and contracts expire.

The layer is now performing two contradictory jobs. It remains the mechanism by which FICO reaches resellers directly and prices in public. It is also the vehicle through which the company delivered its concession under political pressure, since FICO Score 10T at $0.99 plus a $65 funding fee is a direct licence product. The same architecture that reclaimed pricing authority from the channel is now the instrument through which that authority is being renegotiated with a regulator instead.

Optional the software platform A scenario dependent hypothesis rather than a moat layer, and the thesis does not require it.

With high confidence, the platform is growing quickly off a small base, with net retention indicating genuine expansion within accounts rather than repriced renewals, addressing buyers the Scores business already reaches. With moderate confidence, a decision platform embedded across originations, servicing, fraud and collections plausibly creates a second independent lock of a conventional enterprise software kind. With low confidence, treated as scenario rather than forecast, it eventually becomes large enough that impairment of the score's economics would no longer be a thesis level event.

Two counterforces keep it optional. The segment nets to mid single digit growth because platform expansion is partly offset by legacy decline. And this is a genuinely competitive market against large, well capitalised decisioning vendors, in which FICO has neither a coordination lock nor a consumer brand, which are precisely the two advantages making the Scores business exceptional.

Governance, as it bears on the moat A conventional single class, widely held US listed structure with no controlling shareholder. Nothing structurally prevents a future management trading long term standard integrity for a nearer term print. What exists instead is a long tenured chief executive whose strategy has been consistent for over a decade, and a business whose economics are so favourable the temptation to compromise them has been low.

Both are conditions rather than structures. The absence of a governance guardian is a genuine structural weakness and is not mitigated by the current management having behaved well. The strategy for a decade has been legible and singular, which makes key person risk real and concentrated.

There is a specific reason this matters more here than the generic complaint about single class structures. The pricing decisions that drove the recent revenue acceleration are also what produced the regulatory response, and a governance structure with no counterweight to quarterly incentive is a structure that will keep making that trade. The dilution risk in this business is not overuse of the standard but overextraction from it, and nothing in the governance arrangement guards against it.

WHAT CHANGED IN 2025 AND 2026 Three legs, of which the third moves fastest.

Permission. In July 2025 the FHFA director announced by social media post that Fannie Mae and Freddie Mac would immediately permit VantageScore 4.0 for conforming loans, with tri merge retained. In April 2026 the FHFA revised GSE selling policies to permit current use of VantageScore 4.0 and future use of FICO Score 10T, HUD adopted both models for FHA insured mortgages, and a limited rollout began with twenty one large lenders approved.

Historical data. The Enterprises have published historical score data for both models. This is the detail that matters most while receiving the least attention, because publishing historical scores removes the single largest technical obstacle to migration: a lender cannot recalibrate a pricing grid or rewrite investor documentation against a scale with no performance history. A third party is retiring that obstacle on the lenders' behalf at no cost to them, which is what the dismantling of a coordination lock looks like when it happens. It appears nowhere in share data. It appears as the steady removal of the reasons share was locked.

The price war. It began before the April announcement and moves orders of magnitude faster than any migration of share. The provocation was FICO's own: 2025 wholesale royalty set at $4.95 from $3.50, then 2026 pricing set at $10.00 for what the company describes as the identical product. Across five years the royalty moved from roughly $0.60 to $10.00, sixteen fold at a compound rate near one hundred percent.

The response came from the parties owning the competing model. TransUnion cut VantageScore 4.0 to $4 in October 2025 and to $0.99 in March 2026. Experian moved to free indefinitely with a commitment to remain at least fifty percent below FICO. Equifax moved to $4.50 with free bundling. All three now supply VantageScore 4.0 at no charge to any lender that also buys a FICO score.

That last clause is the most consequential commercial fact in this file. Dual pulling is now free, which sets to zero the cost barrier to normalising a competing scale inside pricing grids and underwriting habit, and thereby converts a large future decision into a small one. It required no regulatory action to accomplish.

Why the collective lock became an individual one Consider what a conforming lender now faces if it wants to use VantageScore 4.0. Not the requirement that the whole chain move at once, but that it republish historical scores against the new scale, recalibrate pricing grids, rewrite investor documentation and retrain underwriting staff. Every one of those is a cost borne by that individual lender, which is the definition of the ordinary switching cost this section opened by distinguishing itself from.

The moat did not weaken by a degree so much as change material, from something that does not decay into something that does.

Substantially converted, not converted, and the qualifier is doing necessary work rather than hedging. Residual coordination survives inside conforming: tri merge means an alternative still ships through the same three bureaus, model risk review still applies, mortgage insurers and correspondent investors still have to accept the scale, and loans destined for private securitisation still need rating agency convergence that has not occurred. What the decisions dissolved was the largest single component, the acceptance of the ultimate buyer, which in conforming and FHA lending is the view that mattered most. In jumbo and non agency, where private investors and rating agencies must converge independently and no coordinator exists, the collective lock survives largely unchanged.

Two qualifications keep that residual from being read as durable. Tri merge, the largest item on the list, is itself under organised attack, with the Mortgage Bankers Association pressing for a single report and score above a defined threshold. And its practical force has already been undercut commercially, because free bundling means routing through three bureaus imposes no incremental cost on carrying a second scale. A barrier that costs nothing to satisfy is not doing the work the word implies.

Why market share cannot settle the question The common defence is that FICO's share of conforming closings remains overwhelming. Accurate, and close to uninformative.

A coordination lock breaks at the top and shows up in share years later, because migration itself takes years. Intact share is what you would observe in year one of a genuine structural break, and also what you would observe if nothing were happening. It cannot distinguish between them.

Applying this document's own rule, share is a consequence, and citing it as evidence about the mechanism is the same error as citing operating margin as evidence of the moat rather than of its effects.

Two vectors, moving at different speeds Erosion here can proceed along two independent paths. Pricing authority, meaning what FICO can charge for the standard. And the standard itself, meaning whether FICO's score remains the referenced unit. They are not the same, do not move together, and a company can lose the first entirely while keeping the second.

The price vector is moving quickly, driven by a competitor that can be given away free, by report level cost inflation making the visible royalty politically intolerable, and by a political process that escalated in March 2026 into Congressional oversight and a competition authority referral, venues the original actor does not control.

The standard vector has not started. At the April 2026 announcement the FHFA director put Freddie Mac's VantageScore volume at $10 million of loans, which against a conforming market measured in hundreds of billions is indistinguishable from zero. Resellers were reported through late 2025 as not yet offering FICO's direct programme. And FICO successfully doubled its headline royalty in the middle of all of it, which is not the behaviour of a company that has lost the ability to price.

Both facts are true. Collapsing them into one judgment permits two opposite errors: complacency by pointing at share, alarm by pointing at price. The competitive section below shows that this exact split has a precedent, and that in the precedent the standard survived while the pricing leverage did not.

What remains intact Card, auto, personal loan and account management are untouched and constitute the large majority of pulls by volume, and their coordination locks run through private counterparties with no coordinator empowered to reassign the standard. The consumer brand was never a collective mechanism and is not dissolved by a permission grant. Thirty years of through the cycle validation remains non replicable. And the mechanism of change was administrative rather than statutory, leaving it reversible by a subsequent administration.

The net assessment: the moat's composition in its highest revenue per unit channel has shifted substantially from a structural lock toward a bundle of inertia, brand and calibration advantage. That bundle is formidable and has held through a full year of permitted competition. It is not the same asset, because what it replaced was the only non decaying component.

PRICING POWER Treated separately because the demonstrated record and the current authority now diverge sharply, and a single judgment on "pricing power" would report the first and miss the second.

Demonstrated: exceptional, and among the strongest in the listed universe. The wholesale mortgage royalty moved from roughly $0.60 to $10.00 across five years, a sixteen fold increase at a compound rate near one hundred percent, on a product the company itself describes as unchanged. Volumes did not fall. Scores segment operating margin sits near eighty eight percent. That is not a pricing decision a business with substitutes gets to make, and it is the clearest quantitative evidence that the coordination lock was real.

Current authority: actively contested, and deteriorating fastest of any item in this file. The same pricing that demonstrates the moat is what attracted the response. A regulator has publicly named a target price of $0.99. Three bureaus supply the alternative at between $0.99 and free. A Senate oversight letter in March 2026 recited the price history back as evidence and argued the dominance was cemented by a regulatory framework rather than won on merit, alongside a referral to the competition authority.

The structural problem underneath the price war. The bureaus do not need VantageScore to earn a return. Their return comes from the credit report, which is where the 2026 increases landed, with resellers describing roughly a forty three percent rise in the total price and the trade body warning of forty to fifty percent. FICO's royalty is around fifteen percent of a tri merge bundle running from eighty to over one hundred dollars.

So FICO faces a competitor whose owners are indifferent to that competitor's margin and better off if it wins at zero. The contest cannot be settled by matching price. This is materially worse than facing a rival who must earn a return, and it is the channel problem this document describes throughout, now in its most direct form. The reading is interpretive, since no party has stated it, but the observed pricing behaviour is difficult to account for otherwise.

One reading in the other direction. FICO Score 10T is available through the direct licence programme at $0.99 per score plus a $65 funding fee. The headline reads as ninety percent capitulation; the structure does not. A tri merge pull at $0.99 costs roughly three dollars against roughly thirty at the $10 rate, but the $65 attaches only when a loan funds, converting a per pull royalty into a per closing royalty. Whether that is dilutive depends on pull through rates and pulls per file, neither disclosed. A company with no pricing authority left does not get to restructure the basis on which it charges.

The scale point that bears on all of it. FICO being fifteen percent of the bundle cuts two ways. It is a defence, in that the company is not the principal driver of the cost consumers face. It is also an exposure, because FICO is the visible, nameable, single source component of a bundle whose total price has become politically intolerable, and a regulator seeking a headline reduction reaches for the nameable part first.

THE COMPETITIVE LANDSCAPE

This section runs longer than the framework usually requires, for a specific reason. The competitive question here is no longer whether anyone can build a better model. That was answered definitively over nineteen years, and the answer is set out first below. The live question is what happens to a private standard when a public coordinator decides to make it competitive, and the useful evidence for that does not come from credit scoring at all. It comes from other reference standards that faced the same event, and there are enough of them, with enough elapsed time, to constitute evidence rather than analogy.

Every entry below earns its place by testing a named claim and returning a result.

Case	What it tests	Result
VantageScore	Can capability alone displace a coordination lock	No. Nineteen years, negligible penetration
LIBOR	Can a coordinator replace an embedded reference standard	Yes. Roughly six years, and it worked
Nielsen	What happens when an industry body certifies rivals	Standard survives. Pricing leverage degrades
Credit rating agencies	Does removing regulatory references dislodge an incumbent	Largely no. References proved hard to replace
The three bureaus	Can the channel expropriate the standard's economics	In progress. Rival now supplied free
In house lender models	Does a better model displace the reference score	No. It does a different job
Cash flow underwriting	Is there a richer data source	Yes, but not portable, standardised or referenced

The one direct attack, and what it proved VantageScore was launched in 2006 by the three companies that control the underlying data, distribute the incumbent product, and have every commercial incentive to displace it. It is not a startup and never was. Two decades and multiple model generations later it had achieved wide use in account management, marketing and consumer education, and negligible penetration of the decisions that matter economically.

Superior or equal model quality, unlimited data access and control of distribution all proved insufficient. That is the single best piece of competitive evidence for the thesis, and it isolates precisely what the challenger lacked. It was never capability. It was permission.

The controlled experiment ran nineteen years and the incumbent won it decisively. In July 2025 the missing input was supplied, and the live experiment began.

Three precedents from other reference standards Because the contest is now about permission rather than capability, the informative evidence comes from elsewhere. Three standards have faced a coordinator deciding they should no longer be exclusive, and they produced three different outcomes, which is more useful than one.

LIBOR proves replacement is possible. It was the reference rate embedded in hundreds of trillions of dollars of notional contracts, and no single party could switch away from it unilaterally, which is the same structure described in the moat section above. Regulators decided to replace it after manipulation findings, an industry committee selected an alternative, and the transition ran from the 2017 announcement to cessation of the principal settings in 2021 and 2023. It worked. A coordination lock of far greater scale than FICO's was dismantled by coordinators on a timescale of years.

Three disanalogies limit how far that carries. LIBOR was discredited by a scandal, its underlying transaction base had shrunk so far that the rate was substantially expert judgment rather than measurement, and the replacement was driven by coordinated action across multiple national regulators. None of those applies here. LIBOR therefore establishes possibility rather than likelihood, and it is worth holding in mind mainly as a corrective to the argument that embedded standards cannot be moved.

Nielsen is the closest analogue and the most useful case in this file. It is a private measurement standard functioning as an industry currency, sold to parties who transact against it, embedded in contracts, with no statutory basis and no regulator. In 2021 the Media Rating Council, an industry body acting on complaints from a trade association of networks, suspended Nielsen's national and local television accreditation. A major network issued a request for proposals to more than seventy measurement companies and stated it would likely work with more than one. Alternative providers emerged and pursued certification.

That sequence is structurally the same event as a regulator permitting an alternative credit score: a coordinator declaring that the incumbent is no longer the only approved unit of account, followed by certified rivals and an industry conversation about operating with several currencies rather than one.

Five years of outcome data now exist, and both halves matter. Nielsen's national accreditation was restored in 2023. It remains the currency. One prominent alternative was granted accreditation in early 2026 and another withdrew from the accreditation process entirely a few months later, and industry observers openly question whether the alternatives were ever intended to replace the incumbent rather than to sit alongside it. The standard survived, because the coordination problem below the coordinator is real and because running multiple currencies is expensive for everyone who has to reconcile them.

But the pricing leverage did not survive intact. In 2024 a major customer allowed its Nielsen contract to expire in a pricing dispute and moved to an alternative provider before the two sides came to terms. That negotiation could not have happened in the same way before a certified alternative existed. The mere availability of a credible substitute changes what a customer can threaten, whether or not the threat is ever executed.

The lesson is the FICO base case with evidence behind it: keep the standard, lose part of the rent. And it is precisely the outcome the invalidation register in this file is least able to detect, which is why the gap is named explicitly in that section rather than left implicit. It is also where the five year clock in the re entry conditions comes from, because five years is roughly how long the Nielsen question took to resolve into a legible shape.

Credit rating agencies show that references can be stickier than the coordinator expects. After the 2008 crisis, legislation directed federal agencies to remove references to credit ratings from regulation and to substitute alternative measures of creditworthiness. Agencies found workable substitutes difficult to construct, removal was partial and slow, and the incumbent oligopoly survived largely intact with its position in the plumbing broadly preserved.

The reason that case cuts less favourably here than it first appears is the reason it worked. There was no certified, funded, distributed alternative standing ready to be referenced instead. Here there is one, it has existed for twenty years, its owners control the distribution, and they are currently giving it away.

The channel, which is the live front The three bureaus occupy an unusual position as simultaneously channel, competitor sponsor and the party with most to gain from FICO's disintermediation. Their structural advantage is ownership of the data, the one asset FICO does not have and cannot build. Their weakness is that they are three, they compete with each other, and coordinated action among them attracts competition authorities.

The mechanism now visible changes what a price war here means. The bureaus do not need VantageScore to earn a return, because their return comes from the credit report, and that is where the 2026 cost increases landed. Against that backdrop, pricing the competing score at $0.99, or zero, or free alongside a purchased FICO score, costs its owners very little and improves their position further if it displaces a royalty they currently have to pass through.

The consequence is structural rather than tactical. FICO faces a competitor whose owners are indifferent to that competitor's margin and better off if it wins at any price including zero, which means the contest cannot be settled by matching price. That is a materially worse configuration than facing a rival who must earn a return, and it is the same channel problem this document describes throughout, now in its most direct form. The reading is interpretive, since none of the parties has stated it, but the observed pricing behaviour is difficult to account for otherwise.

The competitors that are not really competitors

In house and proprietary lender models are often mistaken for a threat. Every large lender already runs custom scorecards alongside the bureau score, and this is a permanent feature of the market rather than an encroachment: the custom model refines the decision while the standard score enables the transaction with third parties. Collapsing that distinction is the most common error in bearish analysis on this company, and holding it clearly resolves a great deal of apparent contradiction.

Cash flow and open banking underwriting is the most credible long horizon challenge. Direct access to bank transaction data offers a richer, more current and less gameable signal than the bureau tradeline, and the infrastructure to obtain it now exists. Its weakness is that it is not portable, not standardised, not consented at population scale, and not referenced by any secondary market documentation, which makes it a threat to the form of the assessment rather than to the incumbent within it.

Model based underwriting that outputs a decision rather than a score would dissolve the need for an interpretable common number altogether. It is constrained heavily, and probably durably, by adverse action explanation requirements and fair lending law, which favour interpretable, documented, validated models. That regulatory constraint currently protects the incumbent and could in principle be relaxed.

The competitor that is not a company The most underrated threat is a settlement rather than a rival: a scenario in which the reference role is deliberately made competitive by policy, with several approved scores, published prices and mandatory interoperability. That preserves the demand for a score while destroying the rent attached to being the score.

Standard competitive analysis, which asks whether anyone can build a better model, is structurally blind to this, because model quality is not the contested variable. The 2025 and 2026 decisions are the first two steps along exactly this path, and the Nielsen case above is what the destination looks like when it is reached without a scandal to accelerate it.

WHY THIS WAS NOT COMPETED AWAY

The central inputs are an installed coordination equilibrium, thirty years of through the cycle validation, and a consumer brand. None of the three can be purchased, which is why the direct attack documented above failed for nineteen years despite being mounted by the parties best equipped to mount it.

Corporately, the business has never been seriously threatened with capture, and the reason is arithmetic rather than governance. There is no controlling family and no dual class structure. What protects it instead is a valuation that has for years reflected the quality of the franchise, an aggressive buyback that has shrunk the share count while leaving the balance sheet levered and book equity negative, and the near certainty that an acquisition of the credit scoring standard by any financial or data incumbent would attract antitrust review of a kind few acquirers would attempt.

The moat is therefore protected from imitation by mechanism and from capture by circumstance. The second is a weaker form of protection and should be recorded as such, and neither form of protection addresses the thing that actually happened, which was neither imitation nor capture but reassignment.

THE FINANCIAL FINGERPRINT Included as evidence of the mechanisms rather than as financial analysis.

Revenue compounded at a high single to low double digit rate for a decade, with recent acceleration into the high thirties on mortgage repricing, while operating margin expanded almost continuously. Non GAAP operating margin has reached the mid sixties, at or near the top of the entire large capitalisation listed universe. Scores segment operating margin runs near eighty eight percent, among the highest disclosed segment margins in public markets. Return on invested capital is not meaningful, because invested capital is negligible and book equity negative; the relevant framing is that the business requires essentially no capital to grow. Free cash flow conversion closely tracks net income.

The benchmarks against which erosion is measured objectively rather than by feel: Scores segment margin sustained in the eighties, group operating margin above sixty percent, free cash flow conversion near net income, and B2B Scores unit volume and revenue per unit both trending intact.

One caution about relying on those four. All of them lag, because they measure the moat's output, and the change identified in this document occurred upstream of every one of them. Financial benchmarks confirm erosion after the fact. They do not detect it.

One structural feature is noted rather than judged. The balance sheet carries net debt in the mid two times range with negative book equity, deliberately, as the residue of the buyback programme. For a business with near zero marginal cost and no acquisition ambition that is defensible capital policy. It pairs poorly with a revenue base exposed to a single administrative decision, because leverage removes the option to be indifferent to a prolonged stressed period, and that pairing has become materially less comfortable since 2025.

ALTERNATIVE EXPLANATIONS

An institutional test of any moat claim is whether competing explanations fit the data equally well.

A mortgage origination cycle contributes but cannot be the whole story, because Scores revenue grew strongly through historically depressed origination volumes, which requires price rather than volume as the driver, and the margin structure long predates any recent cycle.

Price increases that will prove unsustainable is testable, in that an unsustainable price produces volume loss or substitution. Through the first year of permitted competition, unit growth continued alongside price realisation, and the company cut its headline price on one product while segment revenue accelerated.

A monopoly rent about to be regulated away is not inconsistent with the data. It is compatible with all of it, which is why it is retained as the principal live alternative rather than dismissed as a courtesy. It has gained substantial weight, and the honest thing to record is by how much. When first written it rested on a forecast. It now rests partly on events: two permission grants, a price war taking the competing product to free, a Senate oversight letter, a referral to the competition authority. That is no longer a prediction about political action. It is political action in progress, with the forecast element reduced to how far it goes.

Software growth cannot explain group results, since the segment grows in the mid to high single digits and is the smaller and less profitable half.

The preferred explanation is a standard setting position with near zero marginal cost, protected by a coordination lock and a consumer brand, whose price had been suppressed by an intermediated channel and was normalised toward the value delivered. It is the only account consistent simultaneously with the margin history, the price elasticity across multiple increases, the resilience of volumes through permitted competition, the nineteen year failure of a well resourced direct attack, and the multi decade persistence of the margin structure. But the gap between it and the monopoly rent explanation has narrowed materially, and a reader who weights the alternative more heavily is no longer disagreeing with the evidence, only with our reading of how the political process resolves.

ASSUMPTIONS AND INVALIDATION Six assumptions carry the thesis.

#	Assumption	Status
1	Consumer lending at scale continues and requires quantified, ex ante risk assessment	High confidence
2	Lenders prefer an external, common, portable standard for the reference decision	High confidence
3	The infrastructure above lending continues to reference an external score by name	High near term. The transmission mechanism by which 4 fails
4	The referenced score continues to be FICO specifically	The central uncertainty. Contested since 2025
5	FICO retains pricing authority at the wholesale level	Actively contested
6	Distribution remains economically viable and is not foreclosed by the data owners	Under structural renegotiation

The conditions under which the moat case is judged false, with current status.

1. Revenue per score declines materially and durably without a compensating unit increase. Not triggered, and the reading is more favourable than commentary suggests: the 2026 royalty was doubled and Scores revenue accelerated. The qualifications are that the $0.99 plus $65 structure changes the basis of charging in a way not yet observable, that the competing product is free when bundled, and that the mortgage volume currently carrying the segment is cyclical while any negotiated price architecture would be permanent.

2. A competing score is pulled alone rather than alongside in a rising and material share of conforming or FHA originations. Not triggered, with no public evidence of material single pull substitution.

3. Secondary market documentation, GSE guidelines and rating methodologies migrate to generic or dual score language at scale. Partially triggered. GSE and FHA policy now name two approved models, historical data has been published for both, and Classic FICO has been reframed from the standard to an approved standard.

4a. Administrative permission for an alternative score in a systemically referenced channel, converting the collective coordination lock in that channel into individual switching costs. Triggered in July 2025 and extended in April 2026.

4b. Statutory reassignment of the standard, or a rulemaking fixing price or mandating interoperability. Not triggered.

5. Mainstream origination practice demotes the bureau derived score to a secondary input behind direct bank transaction data. Not triggered.

6. Scores segment operating margin trends durably out of the eighties for reasons other than mix. Not triggered, with margin currently expanding.

The previous version of this register set condition four's bar at statutory action while the mechanism actually operating was administrative, which meant a register calibrated to the wrong instrument and capable of staying clean while the standard was progressively reassigned. Splitting it into 4a and 4b is the correction.

One gap remains and is named rather than left open. Every condition is drawn around the standard, and condition 1 is the only one touching price. That was defensible when the threat was displacement. It is less so now that the fastest moving vector is pricing authority, because a settlement in which FICO keeps the standard and loses most of the rent attached to it would leave five of six conditions clean while removing a large part of the economics.

The competitive section gives that gap a name and a precedent. It is the Nielsen outcome, it is the most likely single destination on current evidence, and this register would not catch it.

SUMMARY Foundational: the permanent demand driver. The lender's irreducible need to price information asymmetry, bounded by the fact that the need is permanent while the form of the assessment is not.

Foundational: the coordination lock, now partial. The score as unit of account across originator, servicer, rating agency, investor and regulator. Fully intact in card, auto, personal loan and account management. Substantially converted into ordinary switching costs in conforming and FHA mortgage by administrative decision in 2025 and 2026, with residual coordination surviving through tri merge, model risk review, insurer acceptance, and the absence of rating agency convergence outside the agency channel.

Reinforcing: the consumer brand. Genericisation of the category noun, a second lock not running through the bureaus, now carrying more load than it previously had to.

Reinforcing: accumulated time. Predictively, survival predicts survival. Causally, through the cycle calibration is a non replicable ingredient. The second argument has been partially attacked by third party publication of historical scores for competing models.

Reinforcing: the reclaimed channel. Now the layer under active contest rather than the one most likely to be contested next, and simultaneously the mechanism that reclaimed pricing authority and the vehicle through which it is being renegotiated with a regulator.

Optional: the software platform. Explicitly not load bearing.

Governance provides no guard. Single class, widely held, no controlling shareholder. Discipline here is a condition rather than a structure, and the specific discipline missing is restraint in extraction, which is what produced the regulatory response in the first place.

The competitive record is the part of this file that has changed most in the current revision. The one direct attack was mounted by the parties controlling the data and the distribution and achieved negligible penetration across nineteen years, which proves capability was never the binding constraint and permission always was. Permission has now been granted. Of the three comparable reference standards that have faced the same event, one was replaced outright after a scandal, one saw regulatory references prove too hard to remove, and one, the closest analogue, kept the standard and lost part of the pricing leverage. That third outcome is the most likely destination here on current evidence, and it is the one this file's invalidation register would fail to detect.

The central risk has two halves. The thesis is conditional on the reference role remaining singular rather than becoming contestable by design, and that change has begun in one channel. It is also conditional on the rent attached to the standard surviving, and that is being contested now, in public, by a regulator naming a target price, by three bureaus who can supply the alternative free because they earn their return elsewhere, and by processes outside the reach of the agency that started it. FICO could keep the standard and lose most of the economics, and five of six invalidation conditions would remain clean while it happened.

Both halves share one feature no operational excellence, model superiority or pricing skill can hedge. The counterparty is not a competitor but a policy, and the eighteen months from a price increase to an antitrust referral measures how quickly a policy counterparty moves once it decides to.

Decision: rejected. The moat is structurally damaged in conforming and FHA mortgage. Under this mandate that is disqualifying regardless of price, and the name is not monitored pending one of the three re entry conditions specified at the top of this document.

A note on what these documents are and are not. The research published on this site is written for the author's own investment process and published in that form. Each document reflects the analysis, assumptions and judgment of the author at the date of writing and nothing more. Nothing on this site constitutes investment advice, a recommendation tailored to any reader, or an offer or solicitation to buy or sell any security. Valuation figures, scenarios and price levels are illustrative analyst assumptions unless explicitly sourced to company reports or other primary data. Investing involves risk, including the possible loss of capital, and past performance is no guarantee of future results. Any reader considering an investment should do their own work and consult a qualified financial adviser who knows their situation. The author may hold, or may come to hold, positions in the securities discussed.