Data Quality Management Plan

When to Use

Write a data quality management plan when more than one person touches the data, or when the people who do will not be the same next year. It turns "we check our data" into a document someone can follow, audit and update.

It sits between two things you may already have. A data management plan covers the whole data lifecycle, including storage, security and sharing. Data quality assurance is the activity of checking and correcting. The quality management plan is the part of the first that decides which checks of the second run, how often, and who answers for them.

How It Works

The plan answers six questions in order. Keep each answer short enough that a new officer reads it in one sitting.

1. What is covered? The indicators, data flows and period the plan applies to, and what it leaves out.
2. Who is responsible? Named roles: a plan owner, the people who enter data, those who verify it and anyone independent.
3. What counts as good? For each priority indicator, the standard on the five dimensions (validity, reliability, timeliness, precision, integrity), with numbers where possible. The five data quality dimensions page defines each.
4. How are errors prevented? Training, a one-page definition at each desk, tools that block impossible entries, written procedures.
5. How are errors found? Routine checks, sample recounts and a yearly independent assessment, each with an owner and a schedule.
6. How are they fixed and escalated? Who corrects, who approves, what is recorded, and which finding goes up a level.

Add a seventh element, the review date, so the plan does not freeze.

Key Components

A plan of two to five pages is enough for most programs. Longer plans are rarely read.

Best Practices

• Set numeric targets. "Timely" means little. "Report received by day 10 of the next month, in every clinic" can be checked.
• Name a role, not only a team. Each duty needs one answerable person.
• Link to the checklist you use. Point to the field checklist so everyone runs the same checks; the DQA checklist is one option.
• Match effort to risk. Check indicators used for major decisions more often than low-stakes ones.
• Include training. Quality standards fail when staff do not know them. See the enumerator training checklist for the field-team side.
• Review after anything changes. A new tool, new partner or serious finding should trigger a review before the annual date.

Common Mistakes

Copying a donor template without fitting it. A plan full of checks nobody can afford will be ignored. Write what your team can actually run.

No numbers. Standards without targets cannot be tested, so nothing is ever found to fall short.

All checks inside the data team. Include at least one check by someone outside the chain, such as the yearly data quality assessment.

No path for correction. Finding errors without a fix routine leaves the same errors in the next report.

Writing it once. A plan that is never reviewed describes last year's system.

Examples

A fictional clinic network, Riverbend Community Clinics, reports six routine indicators each quarter. Its plan is three pages. The roles table names an M&E manager as plan owner, a clinic registrar for entry within three working days, a district officer for quarterly recounts of ten records per indicator, and an external reviewer for a yearly assessment of three clinics. The standards table sets timeliness at "report received by day 10 of the next month" and precision at "98% of rows have age group and sex filled in." A concordance below 90% in a quarter goes to the district lead, who decides on a retraining visit.

The outline above, with guidance text, is in the template; the filled plan is in the worked example.

Compared To