Sequencing a Data Management Program Under a Visible-Win Mandate
Northwind Logistics has 1,200 employees and six business units. Data management is largely decentralized: each unit runs its own reporting, and a central team of four analysts fields requests. Three initiatives compete for a $1.2M budget: a carrier master data management program, a cloud data lake, and enterprise data quality tooling. The executive sponsor wants a visible win this year. Meanwhile, 14 Microsoft Access databases maintained by business analysts feed a regulatory filing; two have no named owner and one has not been backed up since March. A recent audit flagged the filing as high risk. Business units resist central standards, saying they slow delivery.
1. Given the $1.2M budget and the sponsor's demand for a visible win, recommend a sequence for the three initiatives. Justify what you defer.
Read the worked answer
Sequence by risk reduction and credibility, not by technology appeal.
- First 90 days: the regulatory filing path. Inventory the 14 Access databases, assign owners, back up, document transformations, and put controls around the filing. This is bounded, high-risk, and produces a visible result the sponsor can cite to the board. It also creates the data-ownership precedent.
- Next: operating model and standards with the central team, not a tool purchase. Publish a small set of standards for ownership, regulatory definitions, and change control, and establish domain data owners.
- Carrier MDM: build a business case with quantified benefits such as duplicate carrier records, payment errors, and onboarding time before committing the largest share of budget.
- Defer the cloud data lake until there is a funded use case and an accountable owner; otherwise it becomes a cost center and a new shadow environment.
A defensible allocation is roughly 20% filing remediation, 20% operating model, 50% MDM if the case holds, and 10% contingency.
2. What operating model fits Northwind, and how should the four-person central team spend its time?
Read the worked answer
Neither pure central nor pure federated fits. Central delivery with four people cannot serve six units; pure federation has already produced 14 unowned Access databases. Use a hub-and-spoke model: business units own their data as domain owners accountable for quality and definitions; the central team is a small center of excellence that sets standards, runs the governance forum, operates shared platforms, and coaches domain stewards.
The four central roles should be a data governance lead, a data architect and standards owner, a platform and operations lead, and a data quality analyst. Do not let the central team become a ticket queue for report requests; route those to domains. Explicitly fund domain steward time, or federation will be nominal.
3. How would you measure whether the program is delivering value in 12 months?
Read the worked answer
Use operational and financial outcomes with baselines captured before work starts: regulatory filing error and rework count, and time to produce the filing; number of unowned critical data stores, targeting zero for regulatory scope; carrier onboarding cycle time and duplicate rate; report request lead time; and adoption of standards, measured as the percentage of in-scope systems with named owners and documented lineage.
Avoid composite maturity scores as the primary measure; they move without behavior change. Review quarterly with the sponsor and kill or re-scope initiatives that miss leading indicators two quarters running.