A regulatory-change dashboard can show every project on time while the bank remains exposed. The project may have delivered a policy, but the revised disclosure is absent from a mobile journey. A contract amendment may be complete while a supplier still cannot provide the information required during an incident. Effective implementation therefore follows a requirement into the process where it must work.
Publication, legal adoption and application are separate events. The EBA's 5 November 2026 internal-governance announcement is a recent example of a development requiring detailed applicability assessment. A second example is PRA SS2/21: on 15 November 2026, its page distinguishes the current November 2024 version from the March 2026 version effective in March 2027. Neither example should be reduced to one undifferentiated “new rule” entry.
Build an obligation record that can drive work
For each material change, record the official source, provision, version, status and relevant dates. Then identify the affected legal entity, jurisdiction, regulated activity, product and customer group. Keep interpretation and legal text separate. An internal explanation should be concise enough to drive implementation while retaining a link to the authoritative material.
Write obligations as testable statements. “Improve governance” is not sufficiently specific. “Maintain an approved record identifying the responsible owner for each applicable activity” can be connected to a record, owner and test. Avoid splitting a rule into so many fragments that the relationship between them disappears. A parent requirement with a few operational conditions often works better.
Map the real dependencies
Connect each obligation to existing controls, procedures, technology, contracts, training and reporting. This is a many-to-many relationship. One change can affect several customer journeys, while a shared control can satisfy parts of several obligations. Assign a lead owner and identify the teams whose delivery is necessary for that owner to succeed.
Examine shared services early. A payment platform, document generator or vendor contract can affect several subsidiaries. A group policy does not prove that local implementation is complete. Record local departures and the reason for them, including differences in legal scope and customer treatment. The purpose is to make dependencies visible before the deadline becomes urgent.
Separate readiness from assurance
I recommend four distinct states: interpretation agreed; design approved; implementation complete; effectiveness demonstrated. These are management states, not regulatory terminology. A project may legitimately sit between implementation and assurance while real transactions accumulate, but that uncertainty should be visible.
Define evidence before delivery. A revised customer communication might require approved wording, deployment evidence, accessibility checks and a sample showing that the correct version reaches the correct population. A system rule might require boundary testing, production reconciliation and documented handling of exceptions. This avoids discovering at closure that no one retained the evidence needed to support the readiness claim.
Use a realistic exception process
In a hypothetical change programme, one legacy product cannot adopt a new notification process before an application date. The bank should identify the affected population, legal implications, interim options and decision authority. A project-level risk acceptance cannot waive a legal obligation. Escalation must therefore reach the people who can change the activity, provide a compliant alternative or determine the necessary regulatory engagement.
Time-limited exceptions should show their owner, exposure and expiry. Monitor whether several individually small exceptions depend on the same team or supplier. That concentration can turn a manageable local delay into a group-wide implementation failure.
Retain the history after the deadline
Keep superseded interpretations and implementation decisions in a controlled archive. When a later incident occurs, the bank must be able to reconstruct which version applied and why a particular design was chosen. Link relevant complaints, breaches and audit findings back to the obligation record.
A useful post-implementation review asks whether the change altered actual decisions and outcomes, whether controls survived staff turnover and whether any manual workaround became permanent. Regulatory change is complete when the operating model can sustain compliance and explain its evidence, not merely when the project closes its last task.

