Why data needs a named person
Data that belongs to everyone tends to be looked after by no one. A customer list is updated by sales, support and finance, each with a slightly different idea of what a correct entry looks like. Errors pile up because each team assumes someone else will fix them.
Data stewardship is the practice of giving specific people the day-to-day job of looking after a defined area of data: keeping it accurate, understood and used appropriately. It is a working role rather than a senior title.
Owners and stewards are different
The words are sometimes used loosely, so it is worth separating them. Practices vary, and organizations define the terms in their own way, but a common split is:
| Data owner | Data steward | |
|---|---|---|
| Focus | Accountability and decisions | Day-to-day care and quality |
| Typical level | Senior manager or executive | Subject-matter specialist or team member |
| Typical questions | Who may use this data, and for what? How much risk is acceptable? | Is this record correct? What does this field mean? Who fixes this error? |
| Output | Policies, approvals, priorities | Definitions, corrections, issue tracking |
The owner answers for the data, and the steward does much of the practical work on the owner’s behalf. A single owner may have several stewards, and in a small organization one person may hold both roles for a time. The same split between accountability and daily care appears in the wider discussion of system ownership roles, where a system owner decides and others operate.
Typical steward responsibilities
Duties vary with the organization, but commonly include:
- Maintaining definitions. Recording what each important field means, which values are allowed and how the measure is calculated, so that people stop using the same word for different things.
- Monitoring quality. Running or reviewing regular checks for duplicates, gaps and out-of-date entries, and following up on what they find.
- Resolving issues. Acting as the contact for questions and error reports, deciding the fix or passing the problem to whoever can make it.
- Supporting access decisions. Advising the owner on who needs access and checking that requests make sense, in line with agreed policy.
- Documenting lineage. Noting where data comes from and where it flows, so changes in one system do not break reports elsewhere unnoticed.
- Helping with change. Reviewing how a system change, a merger or a migration will affect the data in their area.
Other roles in the picture
Stewardship sits among several roles that are easy to confuse:
- Data custodian or administrator, usually technical staff who run the databases and platforms, manage backups and apply security controls. They look after the container, while the steward looks after the meaning and quality of the content.
- Data governance lead or council, which sets the framework, policies and priorities within which stewards work.
- Data consumers, the analysts and managers who use the data and are often the first to notice problems.
A short written description of each role, with a few examples of decisions it can make, prevents most overlap.
Choosing stewards
Good stewards are typically people who already know the data well and care when it is wrong, not necessarily the most senior or the most technical. Finance staff often steward financial data, and a supply planner may steward product and supplier information. Points to consider:
- Pick people close to the work. They understand the context behind a strange value.
- Make time explicit. If stewardship is added on top of a full workload with no recognition, it tends to be the first thing dropped. Even a defined few hours a month can help.
- Give them authority to act. A steward who can only report problems and not get them fixed will lose interest.
- Name a backup. Absence or departure should not leave an area unowned.
Keeping it light
Stewardship can slide into committees and paperwork. A few guards help. Start with a small number of high-value data areas, such as customer, product and supplier information, rather than every table in the organization. Keep a simple register listing each area, its owner and its steward. Hold short regular check-ins, and track a handful of quality measures rather than dozens.
Centralized and local stewardship
In a centralized model, a core team may define standards and steward the shared data sets. In a more decentralized or federated setup, local stewards in each business unit look after their own data to common rules, with a central group coordinating. Neither is automatically better. Central stewards bring consistency but can be remote from the detail; local stewards know the detail but may drift apart in how they apply the rules. Many organizations use a mix, with central definitions and local day-to-day care.
Common pitfalls
- Naming stewards in a document but never telling them.
- Giving the role to someone with no authority to change processes.
- Expecting stewards to fix everything personally instead of improving the process that creates the errors.
- Treating stewardship as a one-off project, when data keeps changing and the work is ongoing.
This article offers general information. Any roles that carry legal or regulatory duties, such as those relating to personal data, should be defined with the help of qualified professionals.