# EnterpriseGovernance — Teaching Dataset

> Synthetic enterprise data-governance snapshot. 12 business units × 8 governance capabilities × 24 months of operational-signal trajectory + an annual self-assessment that disagrees with the signal.

**Wholly synthetic.** Any resemblance to a real company's data is coincidental.

## Files

- [`business_units.csv`](business_units.csv) — 12 BUs (4 without dedicated steward)
- [`capabilities.csv`](capabilities.csv) — 8 governance capabilities
- [`assets.csv`](assets.csv) — ~6,300 data assets
- [`capability_scores.csv`](capability_scores.csv) — operational-signal trajectory
- [`self_assessment.csv`](self_assessment.csv) — annual self-rated capability (biased)
- [`SCHEMA.md`](SCHEMA.md) — full schema

## What it represents

A fictional enterprise with 12 business units running through a data-governance maturity program. The Chief Data Officer needs to allocate next year's governance investment across the BUs. Two competing data inputs are available: an operational-signal view (catalog coverage, lineage completeness, certification status, classification coverage) and an annual self-assessment survey filled out by the BU data steward (or the IT director substituting for the BU that doesn't have one).

The two inputs **disagree systematically**: stewarded BUs over-claim on the self-assessment (justifying continued investment); non-stewarded BUs honestly under-report (surfacing the gap). The dashboard's job is to triangulate.

## The dataset's reason for existing

The bias is the point. A dashboard that uses self-assessment as ground truth produces the wrong investment allocation — exactly the failure pattern the [EnterpriseGovernance starter BRD](../../appendix-e/enterprisegovernance-starter.md) describes in its Business Context.

The chapter exercises that use this dataset are about teaching how to surface the bias visually, not how to hide it.

## Deliberate imperfections

- **Self-assessment bias** — Stewarded BUs self-rate +0.3 to +1.0 above their operational level. Non-stewarded BUs self-rate honestly or slightly below.
- **Asset certification skew** — Untagged-asset rate is 77% in non-stewarded BUs vs 42% in stewarded BUs.
- **Catalog presence skew** — Non-stewarded BUs have ~42% catalog presence vs 78% for stewarded.

## Quick start

```bash
cat SCHEMA.md
head capability_scores.csv
head self_assessment.csv

# Load into Power BI:
# 1. Get Data → Text/CSV → select all five CSVs
# 2. Build relationships:
#    assets.bu_id → business_units.bu_id
#    capability_scores.bu_id → business_units.bu_id
#    capability_scores.capability_id → capabilities.capability_id
#    self_assessment.bu_id → business_units.bu_id
#    self_assessment.capability_id → capabilities.capability_id
```

## Reference BRD

The [EnterpriseGovernance starter BRD](../../appendix-e/enterprisegovernance-starter.md) is the BRD for this dataset. It is the highest cross-stakeholder coordination cost of the four starters and the most interesting BRD-authoring exercise as a result.

## License and reuse

Released for teaching use alongside *The Defensible Decision*. Use freely.