In Q1 2024, a $44 billion-asset regional bank lost its VP of Data Governance to a competitor. Within 30 days, the bank received an examination notice. The VP had been the sole owner of 23 critical data domain certifications, the only person who understood the lineage mapping for the bank's stress testing data, and the keeper of the informal escalation paths that actually made the governance program function.
The exam did not go well. None of the 23 certifications could be validated. The stress testing lineage documentation could not be explained to examiners. The informal escalation paths did not exist in any system. The bank received three MRAs directly traceable to the departure of a single person.
This is hero-dependency risk at exam time, and it is the highest-beta governance risk in Tier 2 banking right now.
Why Exams Expose Hero-Dependency
Exams are designed to test whether governance works as a system, not whether it works because of a person. Examiners specifically look for single points of failure. When they ask "who owns this data domain?" and the answer is a name instead of a role backed by documented authority, they note the dependency. When they ask "how do you validate this lineage?" and the only person who can explain it just left, they note the gap.
The exam does not care why the VP left. It cares whether the governance structure can function without them. In most Tier 2 banks, it cannot. The governance lives in people, not systems.
The Three Failure Modes
Hero-dependency at exam time shows up in three ways:
Ownership vacuums. When the person who owned a domain leaves, the domain becomes unowned. Not temporarily reassigned. Unowned. Because the ownership was defined by the person, not by a system that could have reassigned it to a backup owner or escalated it for coverage.
Knowledge concentration. The examiner asks a question about data lineage or quality thresholds. The remaining team knows the answer exists somewhere but cannot find it because it lived in the departed person's files, emails, and memory. The governance documentation does not contain it, or contains it in a form that cannot be navigated by anyone else.
Process collapse. The informal workflows that actually made governance function disappear. The standing meeting that only happened because the VP pushed for it. The escalation that only worked because the VP had the relationships. The certification review that only happened because the VP chased it. Without the person, the process stops, and the exam finds the gap.
The Uncomfortable Calculation
For every senior person in your governance function, ask: if they left tomorrow, how many certification validations would fail? How many data domains would become unowned? How many exam questions would go unanswered?
The number is your hero-dependency index. If it is above zero for any individual, you have a known, measurable gap that an examiner will find. The question is not whether. It is when.
What Resilience Actually Requires
Resilience is not about having backup people. It is about having backup systems.
Role-based ownership, not person-based. Data domain ownership should be attached to a role with documented authority, not to a named individual. When the person in the role changes, the role's certifications automatically transfer to the successor or flag for recertification. The system handles the transition. The governance does not break.
System-based knowledge, not person-based. Lineage, quality thresholds, and certification evidence should live in a system that any authorized user can query. Not in personal drives, email threads, or institutional memory. If the knowledge is not in the system, it does not exist for governance purposes.
Workflow-driven process, not relationship-driven. Governance processes like certification reviews, quality escalations, and attestation cycles should be driven by automated workflows, not by individual initiative. When the person who normally "pushes" a process is gone, the workflow should still fire. The notification should still arrive. The escalation should still trigger.
The CoComply Model
CoComply is built to eliminate hero-dependency. Ownership is role-based and system-managed. When a role changes, the certifications automatically transfer or flag. Knowledge lives in the certification system, not in people's files. Processes are workflow-driven, not relationship-driven. The governance survives the departure because it was never dependent on the person in the first place.
A Question for Your CRO
If your most critical data governance person resigned this afternoon, which certifications would still be valid next week? If the answer is not "all of them," you have a known regulatory risk that you are choosing not to fix. Examiners do not accept "we lost a key person" as an explanation. They expect systems, not heroes.