Risk Assessment & Mitigation Flashcards

A marketing analytics team shares segment-level revenue dashboards with regional managers who have access to data across all regions. What governance risk does this create?

Answer: Managers may view revenue data for regions outside their authorized scope, violating data access controls

Providing cross-regional visibility without row-level security violates the principle of least privilege and exposes sensitive business data beyond authorized access.

Which type of risk is introduced when a marketing team uses a vendor-provided 'black box' ML model for lead scoring without access to its feature logic?

Answer: Model explainability risk, limiting the team's ability to audit, validate, or identify discriminatory patterns

Black box models prevent internal audit of scoring logic, making it impossible to detect errors, bias, or regulatory compliance issues.

A company runs an incrementality test by splitting customers into holdout and exposed groups using ZIP codes. What risk does this geographic split introduce?

Answer: Geographic confounding where ZIP-level socioeconomic differences skew incrementality estimates

ZIP codes vary in demographics and purchasing power, so geographic splits can confound treatment effects with pre-existing regional differences.

An analyst notices that website conversion data in the analytics platform is 25% lower than the CRM's recorded sales. Which risk management step should be taken FIRST?

Answer: Conduct a data reconciliation audit to identify the source and scope of the discrepancy before making decisions

Reconciliation must precede any correction to identify whether the gap stems from tracking gaps, attribution windows, or data definition differences.

Which marketing analytics risk is BEST mitigated by implementing automated data freshness alerts on a reporting pipeline?

Answer: Stale data risk, where decisions are made on outdated metrics due to a silent pipeline failure

Freshness alerts detect when data stops updating, preventing teams from unknowingly acting on data that reflects past rather than current conditions.

A digital marketing team uses last-click attribution and reallocates 60% of budget to paid search. After six months, overall revenue remains flat. What risk most likely explains this outcome?

Answer: Attribution model risk: last-click over-credited paid search by ignoring upper-funnel channels that drove awareness

Last-click attribution systematically over-credits paid search, leading to budget cuts in awareness channels that were actually generating demand.

To reduce the risk of personally identifiable information (PII) being exposed in A/B test logs, which practice should be implemented?

Answer: Tokenizing user identifiers before storing test assignment records in the experimentation platform

Tokenization replaces direct identifiers with pseudonymous tokens in test logs, preventing PII exposure while preserving the ability to analyze user-level behavior.