BIAN in Practice, Part 2: From Service Landscape to Clear Ownership
Part 2 of BIAN in Practice: use a lending journey to connect Service Domains with applications, authoritative records and accountable owners—without turning the reference map into a mandatory deployment blueprint.
Perspectives · BIAN in Practice · Part 2
Previously: BIAN Architecture: A Common Language for a Global Banking Landscape introduced the reference framework. This second article moves into a practical question: how does a bank use that reference to make responsibility clearer?
A transformation workshop has reached the application-mapping stage. The loan origination team says it owns the loan journey. The core banking team says it owns the loan. The credit team owns the approval policy. CRM maintains the relationship, while an integration platform connects the systems.
Every statement may be correct within its own scope. The difficulty appears when the programme asks who owns an offer’s state, who may establish the lending arrangement and who resolves a request whose outcome is uncertain.
BIAN becomes useful when the reference map helps resolve those decisions. The aim is a traceable relationship between banking responsibilities, implemented behaviour and accountable owners.
Read the landscape as a map of responsibilities
BIAN describes its Service Landscape as a reference structure organising Service Domains. Its portal describes the domains as a peer set, each with a specific business function or purpose, and offers tooling for scenarios that bring relevant domains together. [1] [2]
The landscape’s groupings help navigation. They should not automatically become the bank’s organisation chart, database structure or deployment hierarchy. A Service Domain, a business process and an application answer different questions.
| Concept | Question it helps answer | Mapping caution |
|---|---|---|
| Business capability | What does the bank need to be able to do? | A bank’s capability catalogue and BIAN artefacts require an explicit mapping. |
| Service Domain | Which defined banking responsibility is being examined? | Use the selected release’s purpose and scope, not a similar-sounding name. |
| Business scenario | How do responsibilities collaborate for an outcome? | One journey can involve several domains and alternative paths. |
| Application | Which deployed product or component implements behaviour? | One application can cover several responsibilities; one responsibility may span systems. |
| Deployment unit | What is released and operated together? | A logical boundary does not mandate a separate microservice. |
| Accountable owner | Who approves meaning and resolves disagreement? | The bank must assign this role; the reference model cannot appoint it. |
Begin with a bounded journey
A useful starting point is one outcome with known friction. “Modernise lending” is too broad for an initial mapping exercise. “Clarify how an approved consumer-loan offer becomes an established lending arrangement” gives the workshop a more manageable scope.
State the trigger, start and end conditions, exceptions and exclusions. Decide whether the exercise covers a new customer or an existing one, a secured product or an unsecured one, and whether it ends before or after disbursement. Different choices can change the responsibilities involved.
Then describe the current behaviour in ordinary banking terms. Record the decisions and state changes before assigning domain names. This prevents the map from becoming a search for labels that fit an already preferred product architecture.
A worked example: one lending journey, several meanings of complete
Fictional scenario: a bank offers a consumer loan through a digital channel. An origination platform coordinates the application, specialist services provide assessments, an authorised bank decision accepts the credit exposure, and the core establishes the relevant lending records. Customer communication and query-only reporting follow.
This allocation is illustrative. It is not a complete official BIAN scenario or a mandatory mapping. The point is to expose where teams need shared definitions.
| Journey responsibility | Illustrative implementation | Evidence the workshop needs |
|---|---|---|
| Manage the customer’s offer/application procedure | Origination workflow and channel | Offer identity, current state and rules for progressing it |
| Produce and consider assessment inputs | Assessment services and authorised decision workflow | Input provenance, decision authority, conditions and exceptions |
| Record relevant accepted terms | Agreement/documentation components | Which terms were accepted, by whom, and any outstanding conditions |
| Establish the product’s operational records | Core platform through a controlled interface | Authoritative product reference and verified establishment outcome |
| Communicate progress and completion | Communication service | Which business event permits each message |
| Provide replicated information for reads | ODS and reporting consumers | Source, replication freshness and reconciliation; no business transaction execution |
The same word—“complete”—could mean documents received, offer accepted, operational records established or funds disbursed. The mapping workshop should replace that ambiguity with states whose evidence and owners are explicit.
Use the specification to challenge the first mapping
BIAN’s public repository publishes release-specific semantic API artefacts. For a worked reference, the release 14.0.0 Customer Offer specification includes operations associated with the offer procedure, disclosures, underwriting and product initialisation. [3] [4]
That detail matters. A workshop should not assign every occurrence of the word “underwriting” to a separate service simply because a local team uses that label. Read the selected definition and relevant operations, then determine how the bank’s actual responsibilities relate to them.
Equally, an offer procedure’s product-initialisation activity does not, by its name alone, prove that a particular local API commits a loan record. The bank must identify the implemented behaviour and authoritative outcome. Reference semantics guide the investigation; application names do not settle it.
For each candidate mapping, record whether coverage is full, partial, shared or uncertain, with a short rationale. Marking uncertainty is useful: it creates an owned question for business and technical specialists to resolve before procurement or implementation depends on the interpretation.
Build a mapping register that can support a decision
A coloured landscape diagram is useful for orientation. A practical register should also retain the evidence needed to explain each mapping.
| Register field | Why it matters |
|---|---|
| Journey and responsibility | Defines what is being mapped and the relevant exclusions |
| Reference release, artefact and source | Makes the interpretation reproducible |
| Purpose match and exclusions | Explains why the reference fits and where it does not |
| Application/component and interface | Connects the logical view to the implemented estate |
| Coverage and evidence | Distinguishes demonstrated behaviour from a supplier or team claim |
| Business, data and technical owners | Separates approval of meaning from stewardship and operation |
| Authoritative state or record | Identifies where a particular fact becomes established |
| Gap, decision and acceptance condition | Turns uncertainty into work with an owner |
For example, an origination vendor may demonstrate offer tracking while relying on the core for product establishment. That can be a valid split. The register should show it clearly rather than classifying both products as complete, interchangeable implementations of “lending”.
A bank can use several systems to implement a responsibility while retaining one accountable owner for its agreed meaning. Record the systems of record for specific facts and lifecycle stages; a blanket label such as “customer master” or “loan owner” is often insufficient.
Separate orchestration from the authority to change a record
In the example, the origination workflow may request product establishment and track progress. The core may commit the authoritative operational record. A credit decision may originate in a separate authorised process. These are related responsibilities, but their permissions and evidence differ.
The integration platform can authenticate callers, validate messages, route requests and support reliable handoffs. Assigning it an architectural role does not automatically make it the owner of credit decisions or account state.
The ODS remains a query-only store populated through replication. It can offload reads from source systems, subject to freshness and access controls. It does not approve a loan, establish an account, correct source records or execute a financial transaction.
This separation helps during incident response: teams can identify who requested an action, which system applied it, which evidence establishes the result and who can authorise correction.
Test the map with a failure, not only a successful demonstration
Suppose the core establishes a lending record, but its response to the origination platform is lost. The offer workflow still shows “establishment pending”. Replicated data has not yet reached the ODS.
The map should make three things clear. First, the missing response does not prove that establishment failed. Second, absence from a delayed read store is not proof that the source record does not exist. Third, a repeated command requires a defined duplicate-handling and outcome-recovery contract.
The authoritative owner should support an agreed status or reconciliation path using the original reference. Communication waits for evidence appropriate to the message being sent. A command labelled with a BIAN operation name does not, by itself, establish these behaviours.
Use this exception to review the mapping: does it identify the authoritative record, the owner of uncertainty and the permitted recovery action? If the diagram cannot answer those questions, its implementation links need more work.
Decide deployment after the logical boundaries are clear
A bank might retain several mapped responsibilities in one supported platform. It might expose selected capabilities through adapters or split some services for independent change. Each choice has consequences.
Assess transaction consistency, latency, release independence, vendor support, security separation, capacity and the teams available to operate the result. A separate deployment can improve isolation while adding network calls and recovery dependencies.
The logical map should survive those choices. If a product is replaced, the bank should still be able to explain which responsibilities moved, which contracts changed and how ownership was preserved. That continuity is one practical reason to maintain a reference view separately from the application inventory.
Use overlaps and gaps carefully
Two applications mapped to one area do not automatically represent waste. They may serve different products, customer groups, lifecycle stages or jurisdictions. During migration, temporary overlap can be intentional.
Likewise, one apparently empty area may reflect an out-of-scope responsibility rather than a missing system. Investigate the actual behaviour before turning a heatmap into a procurement decision.
A useful review distinguishes duplication of logic, necessary collaboration, deliberate coexistence and uncertain coverage. Give each a different action: consolidate, clarify the interface, manage the transition or obtain evidence.
Make alignment a traceable claim
Ask a supplier to identify the reference release, scope of alignment, selected artefacts, supported behaviour and extensions. Request a demonstration of the required journey and an exception. Alignment evidence and any formal certification claim should be evaluated on their stated scope, rather than inferred from terminology.
BIAN’s Service Landscape page records rationalisation and renamed, added or removed domains in release 14.0. Mapping records therefore need version control. A newer reference should trigger an impact assessment; it should not silently change a deployed service contract. [2]
Keep local extensions visible and justified. A bank-specific requirement may be necessary. Recording it is more useful than describing every difference as failure to adopt the reference.
A practical first workshop
- Define the outcome: agree one journey, its boundaries and success evidence.
- Describe responsibilities: capture decisions and state changes without starting from product names.
- Inspect the reference: review relevant purposes and operations from an identified BIAN release.
- Map the estate: link applications, interfaces, authoritative records and owners.
- Challenge the map: test a timeout, duplicate request or stale read.
- Record decisions: assign unresolved questions and measurable acceptance conditions.
The initial output can be modest: one scenario, a mapping register, a short decision log and an agreed failure test. Measure whether they reduce disputed ownership, repeated interpretation and time spent finding the right incident owner. The number of coloured domains is a weak measure of improvement.
Our perspective
BIAN provides a valuable external reference for examining banking responsibilities. Its practical value grows when the bank connects that reference to actual behaviour, records and decision authority.
A useful Service Landscape helps people make a decision they previously struggled to make. It explains what a responsibility covers, where it is implemented and who answers when the outcome is uncertain. That is a stronger foundation for modernisation than a diagram whose labels cannot be traced to operating reality.
Continue with Learning
Two applications claim the same banking responsibility. What should the map show? Practise checking coverage, authority and recovery evidence in a fictional lending workshop. The companion includes a decision exercise, a walkthrough and a changed scenario to test your reasoning.
The BIAN in Practice series
Part 1 remains the published foundation article. This is Part 2. The proposed next articles move from mapping into contracts, information ownership, modernisation and governance; they are planned topics, not yet-published posts.
| Part | Focus | Purpose |
|---|---|---|
| 1 | A common language for banking architecture | Published foundation: what BIAN contributes and where implementation responsibility remains. |
| 2 | From Service Landscape to clear ownership | This article: map one journey, test boundaries and connect the reference to the application estate. |
| 3 | From semantic APIs to implementation contracts | Planned: operations, business meaning, identifiers, extensions and versioning. |
| 4 | Information ownership and events | Planned: authoritative records, read models, event meaning and cross-domain consistency. |
| 5 | Applying BIAN during core modernisation | Planned: adapters, coexistence, staged replacement and vendor evidence. |
| 6 | Governing and measuring adoption | Planned: reviews, reference changes, ownership and measurable delivery outcomes. |
Next planned: Part 3 will examine how a semantic API reference becomes an implementation contract, including what operation names and schemas do—and do not—establish.
Sources & further reading
- BIAN — Portal: Service Domains and business-scenario tools
- BIAN — Service Landscape and release history
- BIAN — Official public artefact repository
- BIAN — Customer Offer semantic API, release 14.0.0
Sources checked on 7 November 2026. The worked journey, register and adoption recommendations are original editorial analysis. Illustrative responsibility descriptions are not a complete official BIAN mapping. Reference release 14.0.0 is used for the Customer Offer example; implementation and certification claims require their own evidence.
Put this perspective into practice.
Explore the concepts, make a decision and test what changes when the situation changes.
Topics: API Architecture, Banking Architecture, BIAN in Practice, Inside the Banking Backbone, Perspective