Data & Analytics
Data Governance & Quality
Ownership, definitions, and quality rules so numbers mean the same thing everywhere.
What this covers
Governance answers simple questions that are hard in practice: who owns this data, what does this field mean, how good is it, and who may see it. HQT helps agencies establish lightweight governance that fits their size rather than an enterprise bureaucracy.
Typical activities
- Data ownership and stewardship roles by domain
- Business glossary with agreed definitions for key measures
- Data quality rules, monitoring, and remediation workflows
- Classification of data by sensitivity and applicable regulation
- Retention alignment with records schedules
- Master data management for citizens, parcels, assets, and vendors where duplication causes problems
What you receive
A governance charter, glossary, classification scheme, quality dashboards, and operating procedures.
Why it matters
Reports that disagree with each other erode trust with councils, boards, and the public. Governance is how an agency gets to one set of numbers.
Signals that this work is due now
Governance is easy to defer because nothing fails outright. The cost shows up in smaller ways. Different departments report different figures for the same program and the same period, and nobody can explain the gap without a meeting. One analyst is the only person who knows which table is authoritative, so every request queues behind that person. Program staff keep private spreadsheets because they do not trust the system of record. A grant report or a council packet gets rebuilt because a definition changed and nobody recorded when or why. A records request stalls while staff work out which system holds the responsive data. When several of these are present, the constraint is ownership and definitions, and more tooling will not move it.
What working governance looks like, and how it fails
Scope is the first decision. A glossary that covers the measures appearing in published reports, budget documents, and funding submissions can be kept current; one that attempts every column in every database cannot. Each domain needs a steward with authority over the process that creates the data, usually a program supervisor rather than a technologist who only moves it. Quality rules should produce a short exception list routed to someone who can correct the source record, not a score nobody acts on. Classification earns its keep when it maps to record types in the retention schedule and to the access controls already set in the source systems.
The failure modes are recognizable. A governance council meets on a cadence and issues no decisions. A catalog crawls every object in the warehouse and defines none of them. Policy documents use vocabulary the staff doing the work never say out loud. The effort runs as an IT project, and the departments that create the data never attend.
Where governance meets the systems already in place
Most authoritative data in an agency sits in vendor systems: the ERP or financial suite, permitting and licensing, computer aided dispatch and records management, utility billing, the student information system, case management. Governance does not replace those systems. It decides which one is authoritative for each concept and records that decision where analysts can find it.
Two constraints come up repeatedly. A hosted system may not expose its underlying schema, so quality rules often have to run against an export, an API, or the warehouse rather than in place, and the existing contract terms for extracts, API access, and termination decide what is possible there. Separately, the shared reference layer in local government is usually the parcel or address file maintained by GIS, which makes that team a steward of data almost every other domain joins to, whether or not they were invited into the conversation.
What to write into the solicitation
Ownership and handoff are where this work most often fails to stick, so be specific about both.
- Require the glossary, classification scheme, and quality rules in open, editable formats the agency keeps, and prohibit definitions or rules that exist only inside a vendor tenant.
- Require quality rules delivered as runnable SQL or code in an agency repository, each with a written statement of the business condition it tests and the source fields it reads.
- Require steward roles mapped to positions and classifications rather than to individual names, so assignments survive turnover.
- Require every glossary term to trace to at least one field in a source system and at least one published report or funding submission, so the glossary stays tied to something in use.
- Set acceptance on rules running in the agency's own environment after knowledge transfer, rather than on a document review.
- Require classification and retention work to reconcile with the retention schedule the agency has adopted and with the confidentiality rules that follow the records, since one dataset can be subject to more than one, and name the records officer or counsel who approves that alignment.
- Require a baseline captured before the work starts, such as exception counts on the target fields and the rework cycles on a named recurring report, so the result can be compared with something.
Working with HQT on this
HQT scopes this kind of work as a fixed-price or time-and-materials engagement with a written statement of work, named staff, and a defined acceptance process. We are candid about what we can take on and prefer a well-run project to an oversized one. Contact us to discuss a requirement, a draft scope, or an upcoming solicitation.
Frequently asked questions
What is the difference between data governance and data security?
Security controls who can reach a system and what they can do inside it. Governance decides what the data means, who is accountable for it, how good it has to be, and which system is authoritative for each concept. The two meet at classification, because sensitivity labels drive both access enforcement and retention decisions. An agency can have sound security controls and still publish two conflicting numbers for the same program.
Do we need to buy a data catalog tool before starting data governance?
No. A catalog is worth buying once there are definitions, owners, and rules worth recording in one, and it does little before that. Most agencies can carry the first domain in a spreadsheet or a page on an existing intranet, then choose tooling once they know how much they will actually maintain. Buying the tool first tends to produce an inventory of technical metadata that no program staff member opens.
Who should own data governance, the IT department or the program departments?
Program departments should own definitions, quality standards, and stewardship for their domains, because they control the forms and intake processes that create the data. IT owns the pipelines, monitoring, access enforcement, and the environment where quality rules run. A small coordinating function maintains the glossary and escalates disputes that stewards cannot settle. Governance placed entirely inside IT tends to stall, because IT cannot change how a field gets filled in at the counter.
Where should a city or county start with data governance?
Start with one domain where disagreement is already costing something visible, such as the measures in a council report, a grant submission, or a public dashboard. Define those measures, name a steward, write quality rules for the fields that feed them, and work the exceptions until the list is short. That produces a working pattern other departments can copy and a reason for them to take part. Widening scope before the first domain is finished is a common way this work stalls.
Will data governance help with public records requests and grant reporting?
It helps indirectly. Knowing which system is authoritative for a record type, and keeping classification and retention aligned with the schedule the agency has adopted, shortens both the search and the review when a request arrives. For grant reporting, written definitions and a record of when they changed let staff explain why a figure moved between periods, which is a common question from auditors and program officers. Neither benefit appears without the ownership decisions behind it.