learn-data-governance-with-phoebe / Session 7 of 8
Learn Data Governance with Phoebe · Session 7 of 8

The operating model + maturity

Turn one overloaded DPO into a machine: policies, decision rights, and named roles that share the load - plus a maturity score that tells the board exactly where Himalaya stands and what to invest in.

🔴 Advanced DA · DE · DS · AI DPO track DAMA + DCAM 45 min live + self-study depth
0-3 · Welcome 3-23 · Concepts (live cards) 23-42 · Build-along on Himalaya 42-45 · Q&A
Part 0

A DPO who does everything alone is a single point of failure

You've mapped the data (S3), fixed consent (S4), drilled the breach (S5), and cleaned the lifecycle (S6). But if all of it lives in your head and your inbox, Himalaya's governance dies the day you take leave. Today we build the machine: the policies that set the rules, the decision rights that say who decides, the roles that spread the work across the business, and a maturity score that tells you - and the board - exactly where Himalaya stands and what to invest in next.

Live - presented in session Self-study - read after class ★ Do it now on Himalaya DAMA operating model · DCAM concept
★ What you walk out with today A picture of the governance operating model - council, owners, stewards, DPO office - and who does what; the difference between a policy, a standard, and a decision right; a simple maturity scorecard with Himalaya scored honestly (low on most capabilities); and the honesty flag that keeps you credible: DCAM's exact component names are member-gated, so we teach the scoring concept, not invented names.
Part 1 · covers DAMA governance area

The governance operating model 7 min live

Governance is a team sport played by the whole business, not a solo act by the DPO. The operating model answers three questions: what are the rules (policies and standards), who gets to decide (decision rights), and who holds each seat (the roles). Get these named and the DPO stops being the bottleneck.

Governance council cross-domain decisions + priorities Data owners accountable for a domain Data stewards day-to-day quality + curation DPO office privacy oversight, advice + monitoring across every tier Policies + standards flow DOWN ↓ Issues + exceptions escalate UP ↑ The DPO office advises and monitors across all tiers - it does not own the data. Ownership sits with the business.
🔍 Click to zoom - the tiered governance org: authority flows down, issues escalate up, DPO office oversees
LivePolicies vs standards vs decision rights2 min

Three words people use interchangeably and shouldn't. They stack:

  • Policy - the what and why. A short, board-endorsed statement of intent: "we retain personal data only as long as needed." Rarely changes.
  • Standard - the how. The specific, testable rule that implements a policy: "consumer records are purged 90 days after client offboarding." Changes as systems change.
  • Decision right - the who decides. The named authority to approve an exception, sign off a new dataset, or resolve a cross-domain dispute. This is the heart of governance.

Tools don't govern; decision rights do. A catalog can flag a policy violation, but only a named person with a decision right can say "approved" or "stop."

Real world

At Himalaya, "keep data safe" was a vibe, not a policy - so no one could say who approved the US-vendor transfer. You write a one-line transfer policy, a standard ("no overseas transfer without a signed comparable-protection clause"), and give the decision right to approve exceptions to the governance council - not to whoever set up the pipeline.

LiveThe roles - owner, steward, council, DPO office3 min

Four seats, four different jobs. Confusing them is why governance stalls.

RoleWhat they're on the hook for
Data ownerAccountable for a data domain - a senior business person who answers for its correct, lawful use. Sets direction, approves access.
Data stewardDay-to-day quality and curation - definitions, fixing issues, keeping the domain healthy. Hands-on, closest to the data.
Governance councilCross-domain decisions and priorities - resolves disputes between domains, sets shared policy, allocates investment.
DPO officePrivacy oversight - advises, monitors, and is the regulator/data-subject contact across every domain. Independent; does not own the data.

Key distinction to hold onto: the owner is accountable (answers for it); the steward is responsible (does the work). The DPO advises them both and judges the privacy risk - but never becomes the owner, or independence collapses (Session 1).

Real world

Who owns Himalaya's consumer database? Not you - the DPO can't own the very data whose lawfulness you judge. It should be Deven (CPO), whose modules collect it: he's the accountable owner. A senior data engineer on Raj's team stewards it day to day. The council settles it when marketing and product disagree on how the consumer list is used.

Self-studyStewardship workflows - how issues get routed2 min read

Roles only matter if there's a path an issue actually travels. A stewardship workflow is that path:

  1. Detect - a quality check, a user report, or an access request surfaces a problem (e.g. duplicate consumer addresses).
  2. Route - it goes to the steward for that domain, not into a shared inbox nobody owns.
  3. Resolve or escalate - the steward fixes routine issues; anything needing a decision (an exception, a cross-domain conflict) escalates to the owner, then the council.
  4. Record - the fix and the decision are logged, so the next occurrence has a precedent and the DPO can evidence it.

Without a workflow, every issue lands on the DPO - the single point of failure we opened with. With one, most issues resolve at the steward tier and only genuine decisions reach the top. That's the machine working as designed.

Part 2 · covers DCAM - concept only

Maturity: knowing where you stand 7 min live

You can't ask a board to invest without telling them where you are. A maturity model turns "governance feels weak" into a defensible score per capability, on a ladder everyone can read. Himalaya, honestly scored, sits near the bottom - and that low score is exactly what unlocks the Session 8 roadmap.

6 · Enhanced optimised, continuously improved 5 · Achieved operating and measured 4 · Defined documented and rolling out 3 · Developing being built, uneven 2 · Conceptual planned, not yet real 1 · Not initiated nothing in place Himalaya is here Rung names shown are a generic illustrative ladder - not DCAM's official component names, which are member-gated.
🔍 Click to zoom - a generic maturity ladder with Himalaya scored low (conceptual)
LiveWhat a capability maturity assessment is3 min

A maturity assessment scores each governance capability (say, data quality, metadata, protection, retention) on a ladder from "nothing in place" up to "optimised." The point isn't the number - it's the shared, evidence-based picture of where you are and where the gaps are.

The scoring idea is consistent across models:

  • Score each capability on evidence - is there a documented, operating practice, or just intention? A policy that exists but nobody follows scores low.
  • Score on engagement - is the business actually doing it, or is it a shelf document? Adoption matters as much as existence.
  • Compare current score to a target - the gap between them is your investment case, capability by capability.
Real world

Score Himalaya's retention capability: there's no schedule, ex-client data lingers, "keep everything" is the norm. Evidence = none, engagement = none → it scores at the bottom rung. Do that across six capabilities and you have a one-page picture the board can't argue with - and a ranked list of what to fix first.

LiveHonesty flag - DCAM is the model, but member-gated2 min

Teach this straight. The leading data-management maturity model is DCAM (the Data Capability Assessment Model, from the EDM Council). It's the one large enterprises and regulators most often reference. But its exact component names and the full model are member-gated - you need EDM Council membership to use the official framework.

  • So we teach the maturity / scoring concept - ladder, evidence, engagement, gap-to-target - which is public and transferable.
  • We do not invent DCAM's official component names or pass off a made-up list as "the DCAM components." A DPO who fakes a framework loses trust the moment they meet someone who has the real one.
  • If Himalaya needs a formal DCAM assessment, that's an EDM Council engagement - name it as such to the board rather than improvising.
Say it right "We'll use a DCAM-style maturity assessment - the scoring concept - and if we want the certified version, that's an EDM Council membership engagement." That sentence is honest, useful, and credible. Naming eight official DCAM components you Googled once is none of those.
Self-studyUsing a maturity score to get board investment2 min read

A maturity score is a fundraising tool disguised as an assessment. The move:

  1. Show the picture - current score per capability, most in the red. Boards respond to a heat map far better than to prose.
  2. Anchor to risk - tie the lowest scores to the concrete exposure: the near-miss breach, the unlawful transfer, the missing DNC check. Low maturity = the fine waiting to happen.
  3. Set a target and a cost - "move retention and transfer from rung 1 to rung 4 in 12 months, here's the headcount and tooling." Specific asks get funded; "we need to do better" doesn't.
  4. Re-score to prove ROI - come back next year with the same ladder and show movement. That's how you keep the budget.

Himalaya scores low/conceptual on most capabilities today. That's not a failure to hide - it's the baseline that makes next year's improvement visible, and it becomes the backbone of the Session 8 first-90-days roadmap and board report.

Part 3 · covers DAMA Quality

Data quality as a governed capability 6 min live

Quality isn't a nice-to-have that lives on the data-engineering team - it's a governed capability the operating model owns, and it's what makes the Accuracy Obligation (Session 6) real. If users can't trust the data, no amount of policy will save you.

Accuracy reflects reality Completeness nothing missing Consistency agrees across systems Timeliness current enough to use Validity conforms to the rules Uniqueness no unwanted duplicates Quality signals feed the catalog - so users see a trust score and know whether to rely on a dataset.
🔍 Click to zoom - the six data-quality dimensions a governed program measures
LiveQuality dimensions feed trust2 min

DAMA frames data quality as a set of measurable dimensions: accuracy, completeness, consistency, timeliness, validity, and uniqueness. A governed program doesn't just hope for quality - it measures each dimension, sets thresholds, and surfaces the result where users will see it.

  • Quality signals feed the catalog (Session 3): a dataset carries a visible trust indicator, so an analyst knows before they build on it whether it's reliable.
  • Quality ties directly to the Accuracy Obligation (Session 6): the controls that keep data accurate are your quality capability, now owned and measured, not left to chance.
  • The steward runs the day-to-day quality checks; the owner is accountable for the domain hitting its thresholds; the DPO cares because inaccurate personal data used for decisions is a legal risk, not just an analytics one.
Real world

Himalaya's three-conflicting-addresses problem from Session 6 is a consistency and uniqueness failure. As a governed capability, the steward sets a source-of-truth rule and a dedupe check, the catalog shows a trust score, and the wellness client stops routing a health notice to the wrong address. Quality control and the Accuracy Obligation are the same fix wearing two hats.

Build-along · everyone builds

Design Himalaya's stewardship org and score its maturity ★ 19 min · on Himalaya

Two artifacts that turn you from single-point-of-failure into a program: a stewardship org (who owns and stewards what) and a maturity scorecard (where Himalaya stands). We build both for Himalaya together, then you repeat on your own org for homework.

Your company

Himalaya - one overloaded new DPO (you), no named data owners or stewards, no council, "keep data safe" as a vibe not a policy, and governance capabilities that mostly score at the bottom rung. Today you assign the seats and put a number on the gap.

Assign owners and stewards for the top 3 datasets. Consumer database (owner: Deven/CPO), warehouse pipelines (owner: Raj), marketing contact list (owner: Ilsa/CRO). Name a steward for each. You (DPO) advise all three, own none.

Draft one data policy. Pick retention or access. Write it as three lines: the policy (what/why), one standard (the testable rule), and the decision right (who approves exceptions - the council).

Build a maturity scorecard. List ~6 capabilities (consent, retention, transfer, protection, quality, metadata/catalog). Score Himalaya 1-5 on each, with one line of evidence per score. Most will land low - that's honest.

Identify the biggest gap. Rank the capabilities by risk-weighted gap. For Himalaya the transfer and retention gaps tie to concrete legal exposure - those lead the Session 8 roadmap.

Keep the honesty flag. Label the scorecard "DCAM-style maturity concept" - not "the DCAM assessment." Note that the certified version is an EDM Council membership engagement.

★ Do it now - the operating-model promptYou are helping me, a Data Protection Officer, stand up a governance operating model and score maturity. Here is my situation: [paste the Himalaya summary above, or describe your own org - key datasets, who's around, current state of governance]. Produce two artifacts, using ONLY what I gave you: 1. A RACI-style stewardship table: dataset | data owner (accountable) | steward (responsible) | council role | DPO office role 2. A maturity scorecard: list ~6 governance capabilities, score each 1-5 with one line of evidence, and flag the biggest risk-weighted gap Treat this as a DCAM-STYLE maturity concept, not the official DCAM model & do not invent DCAM's member-gated component names. List anything you'd need to confirm as an open question.
Data tip Score maturity on evidence, not optimism - if there's no artifact you can point to, the capability scores at rung 1, however hard the team is trying. Describe your org's state in shapes, not names and real records, when you draft this with an AI assistant. An honest low score today is what makes next year's improvement provable.
Homework

Try it yourself - this week ◐ 30-45 min total

Source material

What this session covers

This session teaches the working content of the frameworks below. Member-gated frameworks and certification exams stay with their official sources - this page makes you fluent in the body of knowledge, honestly flagged where the official model is gated.

DAMA governance - policies, standards, decision rightsPart 1 · the operating model; full depth in DMBOK2
Roles - owners, stewards, council, DPO officePart 1 · accountable vs responsible; stewardship workflow
Maturity / scoring concept - ladder, evidence, engagementPart 2 · assessment concept + board investment case
Data quality as a governed capabilityPart 3 · six dimensions, feeds catalog trust + Accuracy
DCAM (EDM Council) - official model + component namesPart 2 · MEMBER-GATED - concept taught, official names not reproduced (edmcouncil.org)
Check yourself

Three questions before you go 🎯 ◐ 90 seconds

1 · Your board asks for "the official DCAM component names and full assessment." What's the honest DPO answer?

DCAM's exact component names and full model are member-gated. Teach the maturity/scoring concept honestly; name the official route rather than inventing component names.

2 · Deven owns the consumer database; a senior engineer curates it day to day. Which is right?

The owner is accountable (answers for the domain); the steward is responsible (does the day-to-day work). The DPO advises and oversees but must not own the data - that would break independence.

3 · Marketing wants a policy exception to reuse the consumer list for a new purpose. What decides whether it's allowed?

Governance is decision rights, not tooling. A tool can flag a violation, but only a named authority (here, the governance council) can approve or refuse an exception.

Session 7 cheat sheet · pin this

Policy vs standard vs decision rightPolicy = what/why. Standard = the testable how. Decision right = who decides. Tools don't govern; decision rights do.
Owner vs stewardOwner is accountable (answers for the domain). Steward is responsible (day-to-day quality + curation).
Council + DPO officeCouncil settles cross-domain decisions. DPO office advises + monitors across all tiers, owns no data.
Stewardship workflowDetect → route to steward → resolve or escalate → record. Keeps issues off the DPO's single desk.
Maturity assessmentScore each capability on a ladder by evidence + engagement. Gap to target = the investment case.
DCAM honesty flagDCAM (EDM Council) is member-gated. Teach the scoring concept; never invent official component names.
Data quality dimensionsAccuracy, completeness, consistency, timeliness, validity, uniqueness. Feeds catalog trust + the Accuracy Obligation.
Low score = the roadmapHimalaya scores low/conceptual on most capabilities - that honest baseline becomes the Session 8 plan.