← ISO 20022: integration and operational fundamentals
08 / 10 · 60 MIN

Workshop: partial outcomes and operational decisions

Reconcile responses by reference, separate unknown outcomes, and prepare reprocessing decisions with evidence and authority.

Local vocabulary and evidence

The lesson model uses four invented names: RECEIVED, PENDING, ACCEPTED, and REJECTED. They are not presented as official pacs.002 codes or universal settlement states. RECEIVED means only technical receipt under this contract. UNKNOWN is the local conclusion when an expected reference has no observation; it is not a received event. Starting an incident meeting with these definitions prevents one word meaning functional acceptance to business and queue arrival to infrastructure. In a real project, obtain meaning, group or transaction scope, conditions, and references from the applicable documentation.

Start from the expected set

The exercise expects T01,T02,T03. Observed events leave T01 accepted, T02 rejected, and T03 unknown. Two of three expected references have observed outcomes; only one is accepted. Showing coverage 2/3 and acceptance 1/3 preserves the business denominator. An acceptance rate among known outcomes would answer a different question and would need labelling. Do not replace unknowns with success, failure, or another reference’s last value. A response for T99 remains separate because it does not match the agreed set. Investigate association before assigning that outcome to any pending request.

Replay and contradiction

Each fictional event has an identity, reference, sequence, and state. Receiving e2 again with identical data creates no additional acceptance. A second e2 version carrying a different state is a conflict rejected by the program. Different events with the same current sequence and conflicting states also need investigation. The model retains the highest sequence per reference under its explicit exercise rule; a late lower sequence does not regress the projection. This implementation is deliberately small: it models no complete transition, correction, archival, or concurrency protocol. Production requires additional design and testing.

Decide under timeout

A timeout is a communication observation, not proof that all effects failed. With T01 accepted, T02 rejected, and T03 unknown, resending everything can repeat an action already performed. Changing every identity may also destroy correlation useful for replay detection. APS should gather outcomes, consult the reprocessing contract, separate items with known effects, and escalate uncertainty. The PM coordinates owners, schedule impact, and decision authority. No rule in this lab authorizes real banking resends. A temporary hold can have an owner, an update deadline, and an explicit release criterion.

Workshop in English

Use the forty-five-minute guide below with three roles: operations, functional owner, and PM. During the first ten minutes, classify events and state unknowns. In the next ten, add a replay and a conflict, predicting outcomes before execution. Reserve fifteen minutes to propose reprocessing options and prepare an English decision summary. In the final ten, another participant tries to reproduce the diagnosis using only the delivered package. Record reference, evidence, pending decision, owner, and next update. A solo learner can perform roles sequentially without claiming validation by another team.

Acceptance with explicit limits

The package should contain code, profile version, synthetic inputs, expected outcomes, observed outcomes, and execution instructions. A green image is insufficient to repeat the rehearsal. Deriving expectations from rules before running a new version avoids automatically approving a regression. Passing thirty-four checks confirms these local cases; it establishes neither real-schema validation, ISO conformance, settlement, nor autonomy of a shift that did not participate. For production handover, separate technical conformance criteria from operational demonstration. The final summary should state what was observed, what remains unknown, and which evidence the next decision still needs.

DR original workshop / Oficina original DR: 45 minutes
Synthetic only. No banking connections, official status codes or real resend authority.
Apenas dados sintéticos. Sem ligação bancária, códigos oficiais ou autorização de reenvio real.

0-10 min: Expected / Esperado = T01,T02,T03
 e1,T01,1,RECEIVED
 e2,T01,2,ACCEPTED
 e3,T02,1,REJECTED
State the three outcomes, coverage 2/3 and observed acceptance 1/3.
Indicar os três resultados, cobertura 2/3 e aceitação observada 1/3.

10-20 min: Predict, then run / Prever e depois executar
Repeat e2 unchanged: no additional accepted reference.
Repetir e2 sem alteração: nenhuma referência aceite adicional.
Add e2,T01,2,REJECTED: conflicting event identity.
Acrescentar e2,T01,2,REJECTED: conflito de identidade do evento.
Add e4,T01,1,RECEIVED: current highest sequence remains 2.
Acrescentar e4,T01,1,RECEIVED: mantém-se a sequência atual mais alta, 2.

20-35 min: Prepare an English decision note / Preparar resumo em inglês
Observed: T01 accepted, T02 rejected, T03 unknown under DR-REC-1.
Impact: one outcome is unresolved; a full resend could duplicate an effect.
Decision needed: authorized reconciliation and contract-specific reprocessing.
Owner: assign a fictional functional owner and APS investigator.
Next update: choose a specific time and the evidence needed by then.
Do not describe ACCEPTED as financial settlement.
Não descrever ACCEPTED como liquidação financeira.

35-45 min: Handover / Passagem à operação
A second participant reproduces the case and explains escalation.
Outro participante reproduz o caso e explica a escalada.
Solo practice: record self-review only, not independent acceptance.
Prática individual: registar apenas revisão própria, sem aceitação independente.
Deliver: profile, code hash, fixtures, expected/actual outcomes, unresolved questions.
Entregar: perfil, hash do código, fixtures, esperado/observado e questões pendentes.
IN PRACTICE

Case: three physical responses include a replay and still leave T03 without a state. Counting deliveries does not establish per-reference completeness.

Common pitfalls

Timeout treated as rejection; absence treated as acceptance; group state applied to each item without a rule; replay counted as a new operation.

Related topics: Messages, headers, and states · Layered validation · Mapping and data integrity

Take this idea with you

Start from expected references, keep unknowns visible, and justify each recovery decision using the applicable contract and evidence.

Create account

Reference: ISO 20022 Message Definitions · BigSavant ISO20022 fundamentals professional assessment2026.10

ISO 20022 is a trademark of the International Organization for Standardization (ISO). bigsavant.com is an independent preparation platform and is not affiliated with, associated with, sponsored, authorised or endorsed by ISO. Content and questions are original, are not official exam questions, and completing our tests does not award or guarantee any certification. Names are used only to identify the subject. All other trademarks belong to their respective owners.