Andrew Sawers
Back to work

Turning data governance from entry into investigation

I changed the primary interaction model from maintaining metadata records to detecting, investigating and resolving data-quality problems—helping teams resolve anomalies 25% faster.

Case study summary

Impact
  • 25% faster anomaly resolution
  • One flow from signal to resolution
  • Advanced capabilities became everyday work
System scope
  • Metadata lineage
  • Data-quality signals
  • Guided remediation
  • Contextual adoption
Company
Solidatus
Ownership
  • Product framing
  • Workflow architecture
  • Interaction design
  • Adoption strategy
Signals
Data GovernanceSystems ThinkingDecision SupportMetadataEnterprise SaaS

The decision I owned

Reframe the product

Stop optimising data entry. Make investigation and resolution the default path.

Users were treating a governance platform as a place to enter and clean metadata. Fixing isolated editing problems would have improved efficiency while reinforcing the wrong product behaviour. I shifted the centre of gravity towards finding, understanding and resolving data-quality issues.

The system before

View full-size imageSolidatus regulatory reporting data map
The existing centre of gravity. The lineage map connected metadata entities, quality signals and remediation actions, but the interaction model still encouraged users to maintain records rather than investigate problems.

What others were seeing

  • Slow inline editing.
  • Difficult merge workflows.
  • Low adoption of advanced capabilities.

What I recognised

These were symptoms of a structural problem. Investigation and remediation existed in the platform, but the experience made record maintenance feel like the primary job.

Options and trade-offs

Structural options considered for Solidatus
Path Immediate benefit Strategic limitation
Optimise entry and editing Faster maintenance work. Makes the existing data-entry mental model more efficient.
Automate remediation Maximum speed for routine issues. Introduces trust and governance risks where human judgement is required.
Guided investigation and remediation Connects insight to action while preserving control. Chosen: changes the behaviour of the product without removing expert oversight.

The system I established

I replaced scattered node editing with a clear investigation sequence. Each stage exposes the evidence and control needed to make the next decision.

  1. DetectSurface a quality signal.
  2. PrioritiseSeparate urgent issues from background noise.
  3. InvestigateInspect the affected entities and relationships.
  4. Understand provenanceTrace where the issue entered the system.
  5. ReviewConfirm the proposed change and its consequences.
  6. ResolveApply the action with human oversight.

Adoption was part of the architecture

Contextual guidance was not added as a layer of tooltips after the redesign. I used onboarding to teach the new operating model at the point where users moved from familiar maintenance tasks into investigation and remediation.

What changed

Solidatus outcomes
Behaviour shift Observed impact
Workflow paradigm Users moved from locating issues to resolving them in one flow.
Resolution speed Anomalies were resolved 25% faster.
Feature utilisation Advanced investigation capabilities moved into everyday work.