Bank Account Portability and the Case for Registry Switching

Summary

• The Current Account Switch Service, launched in September 2013, made the receiving bank responsible for a standard seven-working-day move. It transfers balances and payment arrangements, closes the old account and redirects transactions addressed to old details. The customer no longer has to negotiate every dependency separately.
• Its guarantee matters as much as its speed. If a switching failure produces charges or interest, those direct costs are refunded. European payment-account rules similarly define provider roles, business-day duties, information requirements and responsibility for direct financial loss caused by non-compliance.
• CASS is not full account-number portability. The customer's bank details normally change. The Financial Conduct Authority nevertheless identified the machinery that full portability would require: shared routing, a history of current and previous numbers, prevention of premature reuse, mandate transfer and balance movement.
• The evidence does not show that easier switching alone creates a competitive market. Awareness, confidence, inertia, product comparison and account opening still matter. The lesson is narrower and stronger: an exit right becomes real when an institution can execute it predictably.
• NRS can advocate adaptation of that mechanism: RIRs and providers authorised by the competent registry framework would run a recipient-led switch, signed state export, scheduled cutover, current-provider pointer, authenticated notices, continuity controls, remediation duties and independent dispute review.
• The neutral switch function must not become a new monopoly. Its interfaces, records, audits and succession arrangements should allow the coordinator itself to be replaced while preserving one authoritative registration state.

The competition problem was fear of the move

Bank customers do not remain with an incumbent only because its product is best. They may stay because departure appears capable of breaking ordinary life. Salary arrives at an account number known by an employer. Rent, tax, utilities, loan payments and subscriptions leave through standing orders and direct debits. Friends, customers and counterparties retain old details. A switch can therefore look less like buying a new product than rewiring a dependency network while it is live.

The UK government described low switching as a barrier to competition when it announced CASS in September 2013. The diagnosis was structural. A challenger could offer a better account and still fail to win the customer if the expected cost of moving relationships exceeded the expected benefit. Incumbency earned value from anticipated disruption.

That distinction matters for number registration. A network may dislike its registry's fees, service, governance or risk posture and still remain because exit appears to endanger registration history, reverse DNS, RPKI state, contacts, transfer recognition and proof relied upon by counterparties. The incumbent does not have to refuse every change. Dependency itself supplies discipline.

Competition policy sometimes answers this problem with disclosure: publish prices, explain alternatives and ask consumers to compare. Information helps only when a chosen alternative can be reached safely. CASS addressed the missing execution. It did not merely tell customers that other banks existed. It created a common route from one to another.

Registry reform needs the same order of operations. Announcing that operators are free, that registry relationships are voluntary or that alternative institutions may someday appear means little while the act of leaving remains undefined. Exit is not a belief. It is a sequence with an authorized initiator, required data, clocks, cutover, rollback and remedy.

September 2013 turned a promise into a service

The Treasury's launch account described four practical assurances. A full switch would complete within seven working days. The customer could choose the switch date. Payments sent to old details would be redirected. A guarantee would protect the customer against financial loss caused by a problem during the switch.

Each assurance answers a distinct reason not to leave. The time limit bounds uncertainty. The chosen date lets the customer avoid a known mortgage payment or payroll event. Redirection catches counterparties that do not update in time. The guarantee tells the customer that process failure is not simply another risk transferred to the weaker party.

The service also covered nearly the entire current-account market at launch. That reach was indispensable. A switching rule followed by only challengers would discipline challengers, not incumbents. Common participation makes the customer's right independent of which large provider currently holds the account.

Today the service describes itself as free, independent and operated by Pay.UK, a not-for-profit company rather than a retail bank. More than 12 million switches have been completed according to its current public count. Those figures are not proof that every customer shops around or that bank concentration disappeared. They show that a common exit mechanism can persist at market scale without one retail provider absorbing all the others.

The institutional achievement is easy to miss because the customer experience is intentionally uneventful. The customer opens the new account, chooses a date and gives authority. The mechanism handles the complexity. Good portability makes the extraordinary legal and technical fact of provider replacement feel ordinary.

The receiving provider owns completion

CASS places the new bank or building society in charge of the switch. That allocation of responsibility is more than customer convenience. It corrects a bargaining defect.

If the old provider controlled initiation, it could delay, introduce friction or use the departure request as a retention opportunity. Even without misconduct, the customer would have to coordinate two institutions with opposite commercial incentives. A receiving-provider-led model gives the actor that wants the new relationship a reason to complete the move and one channel through which the customer can see progress.

The old bank still performs essential acts. It supplies payment information, transfers the closing balance, stops payments at the agreed point and closes the account. But it performs those acts inside a shared process. Its cooperation is a duty attached to market participation, not a concession granted after reviewing whether the customer has a good reason to leave.

The 2014 European Payment Accounts Directive adopts the same basic logic. After the consumer authorizes the switch, the receiving payment-service provider initiates it. The transferring provider must deliver specified information and perform specified actions within business-day limits. The instruments define who does what rather than inviting the two firms to improvise around the customer.

NRS should campaign for the recognised registry framework to make the receiving provider responsible in the same way. A network seeking to change service should not have to secure the incumbent's political agreement before it can begin, but the RIRs and recipients qualified by the competent authority—not NRS—must authenticate and execute the transition. The recipient then has both the incentive and duty to drive completion.

This does not remove the incumbent from validation. It removes the incumbent's power to decide whether exit is institutionally desirable.

A deadline allocates power, not only time

Seven working days is not a mystical optimum. The FCA later found that reducing the period to five days was unlikely to deliver significant consumer benefit. Customers cared more about a switch occurring on the selected date without errors than about an incremental reduction in elapsed time.

That finding improves the analogy. The important feature of a deadline is not speed for its own sake. It is the conversion of delay into an observable failure. Without a clock, an incumbent can request another document, defer a review, wait for a committee or leave the customer unsure whether silence means rejection. With a clock, every entity knows when an answer, export, validation or activation is due.

A registry switch needs several clocks rather than one theatrical target. The receiving provider should confirm a complete request. The incumbent should produce a signed state export. The coordinator should report conflicts. The holder should have a period to review material changes. The selected cutover should occur within a published window. A failed activation should trigger rollback or repair within another fixed period.

Different resource conditions may justify different timings. A small ASN record with no hosted RPKI may be simpler than a large address portfolio with reverse zones, delegated customer records and active certification. A lawful freeze or contested corporate-control change may pause the ordinary track. Complexity should change the declared class and deadline, not return the process to unbounded discretion.

Clocks must also stop honestly. If a provider can mark a request incomplete using vague demands, the deadline becomes decorative. Required evidence, valid rejection grounds and the authority to review a pause must be published. Time discipline works only when control of the clock is itself constrained.

The move is a bundle of dependencies

CASS is often summarized as moving an account. Operationally, it coordinates several things. It moves the available balance, transfers standing orders and direct debits, redirects incoming and outgoing transactions addressed to the old details, and closes the old account. The receiving institution does not receive the incumbent bank as a whole. It receives the state required to continue this customer's ordinary payment life.

That is the correct level at which to design registry switching. A number-resource registration is not one row containing a name. It is a bundle of relied-upon states: holder identity, authorized contacts, historical changes, resource status, dispute notation, reverse-DNS delegation, registration-data publication, RPKI certificates and route-origin authorizations where used, and evidence needed for future transfers or financing. Some functions are public; others require restricted access and careful custody.

A portability mechanism should define a minimum state package for each service. The incumbent would export it in a signed, versioned form. The recipient would validate completeness against a public profile. The holder would see a human-readable comparison before activation: what remains identical, what credentials change, which security entities require reissuance, which services will be referred temporarily and which claims remain disputed.

The bundle should be no larger than continuity requires. Portability is not a reason to copy unrelated membership history, private deliberations or every document an institution has collected. Data minimization reduces privacy risk and prevents the incumbent from turning undocumented holdings into a migration veto.

Nor should a missing optional service block the whole move. If the holder has no reverse zone or RPKI object, the export should say so in a verifiable way. A standard package makes absence legible without inviting invented requirements.

Redirection protects the customer from everybody else's delay

The original CASS design offered 13 months of payment redirection. Later UK implementation material described 36 months, and current Pay.UK information says transactions are redirected for three years after the switch date or longer under the governing agreement. The period changed because the underlying problem is persistent: counterparties do not all update together.

Redirection separates customer exit from perfect coordination by every payer and payee. An employer may hold the old details. A dormant client may revive an invoice after two years. A direct-debit originator may update slowly. The switch remains successful because stale information is translated into current delivery.

Full account-number portability would remove much of that translation need by preserving the details themselves. The FCA's 2015 assessment found that consumers and especially small businesses valued that possibility. Businesses would avoid changing invoices and stationery, notifying customers or signaling distress through new bank details. Yet CASS shows that even partial continuity can remove a large practical obstacle if redirection is reliable.

For number registration, the old administrator cannot remain the permanent forwarding point. That would preserve the dependency portability is meant to end. The equivalent should be a bounded, signed referral service combined with updates to common discovery mechanisms. A query sent to the former provider could return the new qualified provider and current state version without retaining authority to alter the record. Relying systems should receive authenticated notices and refresh from the current source.

The referral period must be long enough for caches and counterparties to update, but it cannot be the only path. If the incumbent fails, the neutral switch record and independent discovery must still identify the successor. Banking redirection teaches continuity; it also shows why the translation layer must not belong exclusively to the party being left.

The guarantee is a map of responsibility

The Current Account Switch Guarantee says that charges or interest incurred because something goes wrong in the switch will be refunded. The promise is narrower than unlimited compensation, but it does crucial work. It names a class of loss and makes the process answerable for it.

Article 13 of the Payment Accounts Directive follows the same logic. A payment-service provider must refund without delay direct financial loss, including charges and interest, caused by its failure to perform its Article 10 obligations. Exceptions and national legal requirements remain. The provision does not pretend every consequence can be valued automatically. It prevents assigned duties from becoming costless promises.

Registry agreements often move in the opposite direction: broad operational discretion sits beside narrow liability. The institution can delay or alter consequential state while the network bears outage risk, legal cost, customer loss and asset uncertainty. A portability right without a liability statement would repeat that asymmetry. The incumbent could miss export deadlines, the recipient could mishandle a security transition, or the coordinator could publish conflicting state, while each points to another actor.

The switching guarantee advocated by NRS should identify direct remedies by failure class, with enforceability supplied by RIR agreements, provider contracts, arbitration or applicable law. A late incumbent export could shift reasonable revalidation cost to the incumbent. A recipient's activation error could require immediate restoration and reimbursement of defined repair expense. A coordinator's inconsistent state could trigger emergency rollback, independent incident review and service credits funded by the coordinator.

Wider business losses would remain subject to contract and applicable law. The guarantee's value lies in making the first layer automatic and visible. When every actor knows which failure it pays to repair, reliability is no longer merely a moral aspiration.

Information rights make the guarantee usable

A customer cannot enforce a process they cannot understand. The Payment Accounts Directive requires information about the transferring and receiving providers' roles, the timing of each step, fees, information requested from the consumer and alternative dispute resolution. These are not decorative disclosures. They define whether the customer can identify a breach.

Registry switching needs an equally legible rights statement. Before authorizing a move, the holder should receive the resource set, current provider, proposed provider, services included, cutover class, expected interruption if any, fees, evidence required, security changes, dispute status, rollback conditions and review route. The language should identify what portability does not do: it does not decide ownership, extinguish a court order, validate an unauthorized transfer or guarantee that every network will accept every route.

During the switch, one status view should show which actor currently holds the next duty and when it is due. If a request is paused, the holder should see the exact ground and evidence needed to cure it. After completion, the holder should receive a signed receipt linking the old and new state versions and confirming which provider is current.

This visibility disciplines all sides. The holder cannot plausibly deny an authorized cutover. The recipient cannot hide an incomplete migration. The incumbent cannot convert a missed duty into an unexplained review. The coordinator cannot change history silently.

Portability is strongest when it leaves an evidence trail that a court, insurer, lender, counterparty or independent reviewer can understand without depending on the incumbent's narrative.

CASS is not account-number portability

The analogy becomes misleading if CASS is described as something it is not. A full CASS switch normally closes the old account and opens a new one with different bank details. Transactions aimed at the old details are redirected. The customer preserves payment continuity, not the literal identifier.

The distinction was central to the FCA's 2015 review. It defined account-number portability as changing provider while retaining the same account details. Sort codes and account numbers help identify and locate an account. Keeping them would remove the need to notify counterparties and reduce fear that incoming payments would disappear.

The FCA found meaningful stated interest. Thirty-five percent of surveyed consumers and 40 percent of surveyed small businesses said portable details would make them more or much more likely to switch. Those responses indicate perceived value, not guaranteed future behavior. The report also found concerns about cost, complexity, fraud and the historical tie between sort codes and providers or branches.

This limitation strengthens the case for registry switching. IP addresses and AS numbers are already supposed to remain stable through many changes in commercial relationship and network operation. Renumbering is not the intended price of replacing a recordkeeper. The desired outcome is closer to full identifier portability than to redirection from an abandoned identifier.

Banking therefore offers two lessons, not one. CASS shows how to standardize migration even when identifiers change. The account-portability study shows the additional shared machinery needed when the identifier must remain the same. NRS advocacy needs both ideas: a disciplined switch operated by authorised institutions and a persistent resource identity maintained through the recognised registry system.

The FCA described the missing common layer

The FCA identified four capabilities likely to be required for full account-number portability. Incoming payments must be routed after a customer moves, including after multiple moves. Current and previously issued account numbers must be recorded so providers do not reallocate them prematurely. Payment mandates must transfer. Existing balances must move.

It then outlined two broad architectures. One would add centrally operated routing and information functions while banks retained their existing systems. The other would create a deeper shared utility for back-office functions, leaving providers to differentiate products, pricing, websites and applications.

The first architecture contains the most direct NRS lesson. A provider does not need to surrender all operations to a central institution for portability to work. It needs to expose a common switching interface and accept a shared answer to one bounded question: which qualified provider currently serves this identifier? Customer-facing services and institutional choices can remain plural.

For an IP prefix or ASN, the shared function should be thinner still. It should not run the network, set commercial use, price the resource or decide routing policy. It should record an authenticated provider transition, preserve one current state version and distribute enough information for qualified registration and security services to follow.

The deeper utility model is a warning. Centralizing every registry function could make portability easy by making providers cosmetic. It could also create a powerful common operator whose failure or discretion affects everyone. The correct design centralizes only the transaction invariant and evidence needed to prevent incompatible authority. Everything that can remain independently operated should remain so.

The common layer should know that the pointer changed. It should not become the owner of what the pointer identifies.

The anti-duplication rule is simple to state

RFC 7020 describes registration accuracy as ensuring that IP addresses and AS numbers are not allocated to more than one party at the same time. Portability must preserve that invariant. The answer is not to copy a registration into a second provider and hope users decide which version to trust.

Each resource set should have one current service-provider pointer and a monotonically increasing state version. A switch creates a pending transaction linked to the current version. At the agreed cutover, the coordinator validates the expected prior state, activates the new provider and increments the version as one indivisible operation. The old provider's record becomes historical. It may refer queries, but it cannot publish a competing current version.

If two recipients attempt overlapping switches, the second transaction waits or fails with a visible conflict. If the resource set must be divided, the holder first authorizes an explicit split into non-overlapping children. If evidence of holder control changes during the process, the transaction pauses for defined review rather than allowing both claims to activate.

This is how identifier continuity and provider competition coexist. The identifier does not duplicate because the authority to serve it changes by serialization. The market contains several providers; the record contains one effective provider at a time.

The rule should be enforced by independently verifiable evidence. Signed receipts, version history and replicated audit records allow relying parties to confirm that a claimed provider state descends from the last accepted state. The coordinator supplies ordering, not truth by proclamation. If its publication conflicts with the signed history, the conflict is detectable.

Portability changes the administrator, not the holder

A bank switch usually involves closing one contract and opening another. A number-registration switch should be more precise. It changes the provider responsible for maintaining and publishing registration services; it does not by itself transfer the resource to a new holder.

The holder identity, resource boundaries and chain of control remain. A corporate merger, sale, court-directed transfer or correction may require a separate rights transaction. Combining provider substitution with holder transfer would make error difficult to isolate and give incumbents an excuse to treat every departure as an ownership adjudication.

This separation should appear in the request. A service switch states: the recognized holder remains X; the resource set remains Y; the current provider is A; the requested provider is B. A holder transfer states a different proposition and invokes different evidence. The two may occur near each other, but each must have its own authorization, decision and review trail.

The distinction also limits NRS. An authorised portability coordinator can verify that the requester controls accepted credentials and that the current record identifies the represented organisation. NRS can monitor whether that process is fair but cannot authenticate a switch or determine contested ownership.

The receiving registry earns the relationship by serving the same holder accurately. It does not acquire the resource. The incumbent loses the service relationship, not the historical fact that it once maintained the record. The coordinator records the transition; it does not become the superior holder.

This three-way separation is the constitutional core: identifier, holder and provider are related, but none should be silently converted into another.

A concrete registry-switch transaction

The mechanism can be stated without institutional mysticism.

First, the holder selects a qualified receiving provider and authenticates through credentials strong enough for the consequence. The request lists resources, current services and authorized representatives. The recipient checks completeness and opens one switch transaction.

Second, the neutral coordinator locks overlapping provider changes for that resource set and notifies the incumbent. The lock prevents conflicting switches; it does not stop routing, ordinary record access or unrelated maintenance.

Third, the incumbent has a fixed period to provide the signed state package or identify a permitted hold. Valid holds are narrow: conflicting holder authority, a specific lawful order, active fraud evidence or a technically material mismatch. General policy disagreement, unpaid unrelated membership fees or institutional opposition to portability are not valid grounds.

Fourth, the recipient validates the package and presents differences to the holder. Required security transitions are prepared in advance. The last verified state remains in service.

Fifth, the holder confirms the cutover. At the scheduled time, the coordinator activates the recipient pointer, increments the version and publishes authenticated notices. The recipient begins current services; the incumbent becomes a referral and history source only.

Sixth, automated checks test registration data, reverse DNS, RPKI publication where applicable and access to correction channels. If a critical check fails, predefined rules either complete a bounded repair or restore the last verified state.

Finally, the transaction closes with signed receipts, timing results, incident details and responsibility for any direct repair cost. This is not merely data movement. It is a rights transaction that leaves evidence.

RPKI and reverse DNS need planned cutovers

Bank payment redirection can tolerate a period in which old details point to new details. Routing-security and reverse-DNS services have different failure modes. They need function-specific migration plans inside the common transaction.

For reverse DNS, the switch must preserve the served zone and the chain through which it is delegated. Nameserver changes can be prepared before activation, checked from independent vantage points and rolled back if delegation fails. Historical provider control should not be required after the new delegation becomes effective.

RPKI is more sensitive. Certificates, repositories, manifests, revocation information, route-origin authorizations and relying-party refresh behavior interact. A migration cannot be reduced to copying files. The holder and recipient need a planned issuance and publication sequence that avoids an unintended interval in which valid routes appear unauthorized. The exact method depends on the trust arrangement and deployed standards, but the service obligation is clear: provider substitution should not become a weapon against live routing.

The common coordinator does not need to issue every certificate or host every zone. It needs to know which service transition is required, whether prerequisites passed, which state is current and when rollback is still safe. Qualified providers implement the specialized functions through open, tested interfaces.

This is another reason to classify switches. A registration-data-only transition can be quicker than one carrying active RPKI and reverse-DNS dependencies. The customer should see the difference before authorizing the date. Uniform rights do not require pretending that all technical moves have identical risk.

Disputes must travel without becoming vetoes

Portability will attract the objection that a holder could leave to escape a dispute. That risk is real and manageable. CASS does not make debt disappear merely because a current account closes. Account obligations, credit decisions and legal claims remain governed by their own instruments. A switch changes the service arrangement, not history.

A registry switch should carry dispute status, relevant orders and a signed chain of prior decisions into the new state. The recipient cannot delete a valid notation because the holder dislikes it. A court-directed freeze remains effective within its scope. A contested corporate-control request can pause until an independent decision identifies who may instruct the provider.

The incumbent, however, must not manufacture a dispute by objecting to exit. A valid hold requires identified evidence and a review route. The provider cannot transform an ordinary fee complaint, policy disagreement or reputational allegation into indefinite control over registration.

During review, the last verified operational state should normally continue. The purpose of a hold is to prevent an unauthorized change, not to punish the network by breaking existing service. If the dispute concerns only which qualified provider should serve the record, an independent interim custodian may maintain continuity while the merits are decided.

This structure treats disputes as data and legal questions rather than institutional property. They survive the move where they should, but they do not make the incumbent immortal.

Switching quality is not the same as market activity

The FCA found CASS worked well for customers who used it, yet awareness and confidence were lower than expected. Switching rose after launch but remained low. The CMA continued to identify inertia, comparison difficulty, concentration and stable market shares. A technically successful exit rail did not compel customers to leave or guarantee that better products existed.

That is an important limit for NRS advocacy. Portability would not prove that many qualified registry providers will emerge, that networks will compare them well or that every provider will offer materially different service. It would not settle the legal status or economic treatment of number resources. It would not by itself make registration accurate.

It would change incentives. An incumbent that knows a holder can leave faces a different relationship from one that knows departure requires renumbering or institutional permission. A challenger can invest in service because winning a customer is operationally possible. A holder can threaten exit credibly during a dispute. Insurers and lenders can price provider failure separately from destruction of the resource identity.

The proper measure is therefore not port counts alone. Measure completion time, failed exports, invalid holds, rollbacks, service interruption, security incidents, correction outcomes, provider concentration and the share of holders who know the right exists. Then examine whether prices, service quality and governance behavior change.

Portability is competition infrastructure. It is not competition by decree.

Neutrality must be engineered around the switch operator

Pay.UK's position outside retail banking helps CASS. A customer does not ask the old bank to operate the entire common service, and the new bank does not gain authority over all future switches. But independence is not a permanent property bestowed by corporate form. A common switch operator can itself become critical infrastructure.

NRS should highlight that danger before reliance forms. The competent RIR/IANA framework should require the portability record to be continuously replicated to independent custodians. Interfaces and transaction formats should be public. Providers should be able to test against more than one implementation. Audit logs should be tamper-evident and available to authorised reviewers.

Governance should separate rulemaking, operation, audit and dispute review. A vendor operating the transaction service should not decide whether a competitor qualifies. A registry provider should not control the coordinator's board in proportion to the customers it risks losing. Holder representatives and technical operators should see performance evidence without gaining access to private data they do not need.

Most importantly, the coordinator must have a successor procedure. Procurement, migration rehearsal and emergency substitution should occur before failure. The state required to continue switching cannot be held in a proprietary format or under credentials known only to one company.

The neutral layer is justified because it prevents competing providers from writing incompatible histories. Its authority ends there. It orders transitions; it does not allocate commercial rights, govern routes or license acceptable business models.

A phased introduction can test the right without weakening it

Portability need not begin with every resource and service condition. It can begin with a minimum class while stating the complete right from the start.

The first phase could cover uncontested holder-preserving moves for resources with verified contacts and no active freeze. Two or more qualified providers would implement the same export, transaction and referral interfaces. Test cases would include repeated switches, partial resource sets, stale credentials, provider outage, overlapping requests and rollback.

The second phase would add reverse-DNS transitions and richer registration data. The third would add planned RPKI migration after independent validation demonstrates that relying-party behavior and security-entity continuity are understood. Complex portfolios and emergency provider failure would follow once ordinary switching produces stable evidence.

Phasing must not become a permanent excuse. Each excluded class needs a published reason, owner and date for review. A provider cannot declare all consequential customers too complex to move. The point of early limits is to learn where the process breaks, then expand the right with evidence.

The service should publish aggregate results from the first day: median and tail completion times, pause reasons, incumbent response rates, rollback causes, erroneous activations, fraud attempts and customer complaints. A right that cannot be measured will drift back toward discretion.

Bank switching became credible through repetition. Registry switching should do the same, but with open technical proof and a stronger anti-monopoly design.

Fees should follow service, not punish departure

CASS is free to the switching customer, while participating institutions fund the shared arrangements. European payment-account rules require any switching fees to be reasonable and related to actual cost. The important design choice is that the incumbent cannot set an exit price equal to the value of the customer it is losing.

Registry portability needs the same discipline. The incumbent may recover the ordinary, published cost of producing a verified export if that cost is not already included in service fees. The recipient may charge for onboarding and specialized migration work. The coordinator may recover the narrow cost of ordering and auditing the transaction. None should levy a percentage of the resource's market value, a penalty for leaving a region or an open-ended legal-review charge controlled by the incumbent.

Charges also need allocation when a provider causes rework. A holder should not pay twice because an export was incomplete, a deadline was missed or the coordinator published inconsistent state. The switching guarantee should move those direct repair costs to the responsible actor.

Transparent fees support entry as well as exit. A prospective provider can price migration services when the required work is known. A holder can compare the full cost before authorizing. The coordinator can publish cost distributions and investigate outliers. Portability becomes less credible when price is discovered only after the incumbent knows the holder is committed to leaving.

What the banking analogy cannot decide

A current account is a regulated financial contract. A bank performs identity, sanctions, fraud, credit and prudential checks before accepting the customer. Deposits and payment instructions move through national and international payment systems under detailed law. Number resources occupy a different legal and operational environment.

An IP prefix is announced through BGP by autonomous networks. An ASN identifies an autonomous system in routing. RPKI can express route-origin authorization but does not settle every underlying legal claim. Registration providers operate across jurisdictions, and no single consumer regulator supplies one universal remedy. A switch may affect customers and counterparties that never joined the provider contract.

These differences mean NRS cannot copy a seven-day timetable, a compensation phrase or a centralised utility and declare the problem solved. It must advocate resource-specific authentication, security transitions, distributed discovery, cross-border legal handling and independent adjudication, while RIRs, IANA-facing services and authorised providers implement them.

The analogy also cannot prove that NRS should be a coordinator at all. It supports a mechanism, not a brand. NRS's achievement would be persuading authorised institutions to build a credible, replaceable switch service and publishing independent evidence about whether it works.

What banking can decide is narrower. Mature markets can require incumbents to cooperate in customer departure. They can assign the receiving provider to lead. They can publish clocks. They can preserve continuity across stale dependencies. They can attach direct financial consequences to failure. None of that requires preserving one provider's monopoly.

That institutional fact is enough to change the number-registry debate.

The right is continuity without allegiance

The present registry arrangement often treats continuity as a reason to preserve the institution. Banking switching treats continuity as a reason to make the institution replaceable. The account must keep paying salaries and bills, so the providers cooperate in a controlled departure. Criticality strengthens the switch duty; it does not cancel it.

Number registration should follow the same direction. Uniqueness must continue. Accurate records must continue. Reverse DNS and routing-security services must have safe transition paths. Existing routes and customers must not be collateral damage. Those requirements justify stronger portability controls than an ordinary consumer account, not weaker exit.

The concrete NRS contribution should therefore be an advocacy specification for a Registry Switch Service, with six promises made enforceable by RIRs, authorised providers, arbitral bodies or law: the receiving provider leads; every actor has a deadline; the holder receives a complete state comparison; one atomic cutover preserves one current provider; failures trigger restoration and assigned remediation; disputes receive independent review without destroying the last verified operational state.

Around those promises sit the safeguards: strong authorization, narrow holds, signed history, bounded referrals, service-specific cutovers, public metrics, replicated coordinator state and a tested successor.

Standardized migration does not make every provider equal. It makes provider quality contestable. Deadlines do not solve every dispute. They prevent silence from becoming power. Liability statements do not compensate every consequence. They stop institutional failure from being free. A neutral coordinator does not own the identifier. It protects the order in which providers change.

The result is continuity without allegiance. A network keeps the identity on which its customers and infrastructure rely. A registry competes to serve that identity. The common layer preserves uniqueness. No incumbent receives immortality merely because switching was once inconvenient.

That is the case bank portability makes for number registration: protect the relationship that must continue, and make the institution providing it possible to leave.

Evidence and analytical limits

This analysis relies on the UK government's 2013 launch record, current CASS and Pay.UK service materials, the FCA's 2015 operational and account-portability review, the European Payment Accounts Directive, UK implementation materials, CMA competition findings, RFC 7020, current NRS public positions, Lu Heng's portability and registry-continuity arguments, and the still-unfinished ICP-2 review.

Official bank and payment-system materials establish dates, assigned roles, service features, reported performance and legal duties. They do not independently prove that switching caused every market change. The FCA and CMA evidence is therefore used to preserve the negative result as well as the positive one: easier execution improved the process, while inertia, awareness, comparison and product structure continued to limit switching.

NRS and Lu Heng materials supply a stated institutional direction, not evidence that a recognized cross-RIR portability service currently exists. The proposed Registry Switch Service is a derived design. The 2024 ICP-2 consultation principles discussed continuity, record sharing and successor operation, but the wider RIR Governance Document remained under revision in July 2026. No current cross-RIR portability right is inferred from a proposal.

The banking analogy is limited to mechanisms: receiving-provider control, standardized data transfer, deadlines, redirection, guarantee, liability allocation, public information and neutral coordination. It does not equate bank deposits with IP addresses, payment routing with BGP, or a national consumer remedy with global number-resource law. Any implementation would require independent technical testing, explicit contracts, lawful dispute handling and evidence that the switch preserves rather than weakens uniqueness and live-network continuity.