Shoprite Group • Data Governance Operating Model

0. Quick Start The 60-second version

One sentence: the business owns its data through appointed Data Sponsors (executives) and Data SMEs (operational experts); the Data Governance Office provides standards, tooling and coordination; forums make the decisions; everything is measured.

The five things everyone should know

Every domain has a Data Sponsor. An executive accountable for that data's quality, definitions and access. Their Data SMEs handle the day-to-day.
Found a data issue? Log it — DQLabs if it came from profiling, the SharePoint Form for anything else. Your domain's Data SME picks it up. Never just work around bad data.
Access to personal information is layered, not forbidden. POPIA requires safeguards, not prohibition. Access needs a legitimate business need and Sponsor approval.
Retention has rules. Default is 7 years unless the Retention Schedule says otherwise. Pharmacy and Money Market data follow their own statutes. See the Data Retention Standards app for any "how long" question.
Privacy incident? Any suspected unauthorised access to data — digital or physical — goes to privacy@shoprite.co.za immediately.

Where decisions get made

Steering Committee strategy, funding, appointments — quarterly. Domain Councils definitions, CDEs, quality rules for their domain. DDRB design QA for data models. AI Governance Committee high-risk AI use cases. DGO runs it all day-to-day and escalates risk monthly to the MRF, quarterly to the ARC.

1. Why the DGOM Exists Purpose & drivers

The DGOM exists to fix specific business problems: fragmented ownership, inconsistent definitions, data quality issues, uncontrolled data sharing and increasing regulatory risk. It translates our policies (Data Governance Framework, Privacy, Quality, Retention, Document Management) into structures people can actually use.

What it delivers

Clear, business-owned accountability for data across all key domains
Higher, more transparent data quality for Critical Data Elements
Safer, faster, more controlled access to data — aligned to privacy and security
Better reuse and understanding through standardised definitions, metadata and lineage
A governance capability that scales with growth, digital transformation and AI

Strategic alignment

Strategic driver	DGOM impact
A truly customer-first culture	Trusted customer data drives better decisions, personalised experiences and compliant use of data (POPIA).
Develop future-fit channels	Consistent, high-quality data across channels supports seamless digital and omnichannel experiences.
Enable precision retailing	Reliable, fit-for-purpose data powers analytics, AI and supply chain optimisation.
Leverage platform advantage	Data treated as a strategic asset improves reuse, reduces cost, increases efficiency.

The AI-era argument: competitors can copy our AI models. They cannot copy our trusted data. Governance is what makes the Group's AI lead defensible — that is the DGOM's value proposition in one line.

2. Operating Model & Forums Who decides what

The model is federated: accountability sits with business domains, coordination and standards come from the centre. Structures stand up as Data Sponsors and Data SMEs are onboarded (Framework §7) — see the Roadmap for sequencing.

The federated model (Framework §7): strategy at the top, execution in the domains, enablement underneath.

Forums and decision rights

Forum	Frequency	Key decision rights
Data Governance Steering Committee CAO chairs; Data Sponsors; Group Risk; GISO; IT Governance	Quarterly	Strategic alignment · capital for enterprise tooling · formalise Sponsor/SME appointments · budget & resourcing · resolve escalated deadlocks · prioritise data projects
Data Governance Councils (per domain)	Quarterly / as needed	Approve domain definitions & CDEs · DQ rules & remediation · domain access rules within Group policy · oversee privacy & retention adherence · escalate to Steerco
Data Focus Areas (Data SMEs)	Continuous	Execute the DQ lifecycle · propose rule amendments · recommend remediation · build the Glossary & Catalogue
Data Design Review Board	As needed	Approve/reject data & integration designs · verify metadata, lineage, quality and retention are embedded · flag design risks · aligns with the ARB
AI Governance Committee	As needed	Review/approve high-risk AI use cases · define ethical guidelines · enforce bias & model-risk mitigations · escalate material risks

Risk reporting lines (Framework §8): the DGO reports data risks monthly to the Management Risk Forum and quarterly to the Audit & Risk Committee and the Steering Committee. The Board exercises its data accountability through the ARC. No parallel structures — governance rides the Group's existing risk machinery.

3. Data Domains & Ownership Interactive explorer

Every domain has a Data Sponsor — an executive accountable for its quality, definitions, access and lifecycle — supported by Data SMEs. Tap a domain for details. RED EDGE = heightened regulatory obligations. TEAL EDGE = added in V1.1.

Live register: named individual appointments are maintained in the DGO's appointment register — this explorer shows the accountable business unit per domain. Framework §6 notes the domain list "may not be exhaustive"; Healthcare & Pharmacy and Financial Services were added under that provision to reflect the Group's heaviest regulatory exposure.

4. Roles & Accountability Who does what

Terminology: Shoprite's canonical terms are Data Sponsor and Data SME (Framework §9). If you see "Data Owner" or "Data Steward" in vendor tools or industry material — same roles, one-to-one.

Accountability flows down, escalation flows up. SMEs are existing employees with domain knowledge — not new hires (Framework §9.1).

Supporting roles

Role	What they do
Data Governance Office	Maintains the framework, policies, tooling, forums and KPIs; escalates risk (monthly MRF, quarterly ARC); supports every domain's rollout.
Enterprise / Solution Architect	Designs data and integration architecture aligned to governance, EA and security standards.
Security & Compliance	Sets and oversees security, privacy and regulatory controls; enforcement of access; breach handling.
Analytics / BI / AI Lead	Consumes governed data; flags gaps; supports ethical, compliant AI use.
HR / People	Embeds governance roles into job profiles, performance objectives and training.

5. Core Processes Issues · Access · Quality · Glossary

5.1 Data issue lifecycle

Issues come from profiling (logged in DQLabs) or day-to-day business use (logged via the SharePoint Form). Both consolidate into the DGO's central register — the single system of record. Fix at source, not downstream.

Seven steps. The Data Sponsor is accountable throughout; the Data SME coordinates; the Custodian fixes.

5.2 Accessing personal information — the layered model

POPIA does not prohibit access to personal information; it requires safeguards. We apply layers cumulatively — like an onion around the data:

The layered safeguard model from the Approach to Accessing Personal Information (Shoprite Analytics).

Who approves: Data Sponsors approve access to their domains — particularly restricted or sensitive data — with Security & Compliance providing risk oversight. External sharing requires contracts and secure channels. Classification lives in Erwin; enforcement (entitlements, masking, approvals) is executed by Security/IAM.

5.3 The eight data quality dimensions

Per the Data Quality Policy §4 — the basis of every DQ rule, threshold and scorecard:

AccuracyCompletenessConsistencyValidity IntegrityUniquenessTimelinessTraceability

Sponsors and SMEs set thresholds per Critical Data Element; detailed rules live in the Data Quality Standards, not here — one source of truth.

5.4 Glossary governance lifecycle

PROPOSE any SME or business user, via the catalogue workflow → REVIEW domain SME checks accuracy & duplication → APPROVE Sponsor / Domain Council (Group-wide terms: Steerco) → PUBLISH with owner, definition, classification, effective date → REVIEW ANNUALLY or on process/system/regulatory change.

6. Retention & S/4HANA Lifecycle & migration

One rule to remember: data cannot be stored indefinitely. Default retention is 7 years unless the Data Retention Schedule – Legal Standards says otherwise. For any "how long do we keep X" question, use the Data Retention Standards app — the interactive companion to this one.

Lifecycle in brief

Retention periods come from the Legal Schedule, operationalised by the Retention & Archiving Implementation Standard. This model references them — it never restates them.
Sponsors approve domain schedules and sign off destruction/anonymisation events. SMEs run lifecycle reviews. Custodians implement archiving, deletion and masking. Compliance monitors and archives destruction certificates.
Exceptions: Healthcare & Pharmacy follows the Pharmacy Act (not overridden by the Companies Act); Financial Services carries a parallel FICA clock (5 years from relationship end). These domains never take the default without a schedule check.
Every disposal event generates an immutable audit trail: date, ranges removed, mechanism, approval reference. Corporate evidence for regulators.

How the DGOM plugs into S/4HANA

Mechanism	What it means
DMOR retention rulings	Every migration object carries a signed-off ruling — period, trigger, residence, tier plan, disposal — delivered just-in-time per wave. Nothing migrates unconfirmed.
Design checkpoints	Solution designs pass the Data Design Review Board; S/4 design sessions reference catalogue definitions, so the build stands on governed terms.
Ownership recruitment	DMOR objects need a named Sponsor and SME before migration — the migration mandate is how the stewardship network gets built.
PI & POPIA	Profiling flags PI into the POPIA inventory; PI-bearing objects get event-based destruction triggers engineered into the S/4 build, not retrofitted.
No shadow copies	Snowflake and lake copies inherit retention as catalogue metadata. No downstream copy outlives its archived source.

7. KPIs & Maturity How we measure

Targets are phased: FY27 targets track the wave-based rollout so the scorecard shows progress, not perpetual shortfall. End-state targets apply from FY28.

Domains with named Sponsors & SMEs

FY27Wave 1 & 2 by Q2 · priority domains by Q4

End-state100%

CadenceQuarterly

CDEs with defined quality rules

FY27≥95% of Wave 1 & 2 CDEs

End-state≥95%

CadenceQuarterly

Data quality score — key domains

FY27Baseline all Wave 1 & 2 systems; improving trend

End-stateDomain thresholds

CadenceMonthly

Governed datasets with business metadata

FY27≥90% of Wave 1 & 2 datasets

End-state≥90%

CadenceQuarterly

Critical flows with recorded lineage

FY27Wave 1 flows after pipeline fix (Q1)

End-state≥90%

CadenceQuarterly

Restricted access turnaround

FY27Baseline Q2; SLA agreed Q3

End-state≤ agreed SLA

CadenceMonthly

Open high-severity issues

FY27Decreasing from Q2 baseline

End-stateDecreasing trend

CadenceMonthly

SME participation rate

FY27≥85% of appointed SMEs

End-state≥85%

CadenceQuarterly

Use cases on certified data

FY27First certified datasets Q3

End-stateIncreasing trend

CadenceQuarterly

S/4: DMOR objects with signed-off rulings

FY27100% Wave 1 · zero DG-attributable delays

End-state100% all waves

CadencePer wave

Maturity

Assessed annually against DAMA-DMBOK. Baseline established 2024. Committed target: DAMA Level 3 (Defined) across all key operational domains by end FY28. Each annual assessment reports progress against both baseline and target.

8. Implementation Roadmap FY27 wave plan

The Framework is explicit: structures stand up as and when Sponsors and SMEs are onboarded. This is the transition path — commitment, not just vision.

FY27, quarter by quarter. Wave 1 = S/4-critical systems; Wave 2 = PI-heavy systems; Wave 3 (FY28) = the remainder.

Resourcing reality: the central DG function is deliberately small — a manager and two analysts. The model scales through federation, not headcount: the DGO builds standards and tooling; appointed Sponsors and SMEs execute in the business (SMEs are existing employees, not new hires); Custodians implement; external partners carry S/4 template load. Every forum, onboarding and KPI phase respects this constraint.

How data governance works at Shoprite — in practice

0. Quick Start The 60-second version

The five things everyone should know

Where decisions get made

1. Why the DGOM Exists Purpose & drivers

What it delivers

Strategic alignment

2. Operating Model & Forums Who decides what

Forums and decision rights

3. Data Domains & Ownership Interactive explorer

4. Roles & Accountability Who does what

Supporting roles

5. Core Processes Issues · Access · Quality · Glossary

5.1 Data issue lifecycle

5.2 Accessing personal information — the layered model

5.3 The eight data quality dimensions

5.4 Glossary governance lifecycle

6. Retention & S/4HANA Lifecycle & migration

Lifecycle in brief

How the DGOM plugs into S/4HANA

7. KPIs & Maturity How we measure

Maturity

8. Implementation Roadmap FY27 wave plan

FAQs Plain answers