Skip to content
Technology transformation expertise firm · ARDIANE Group Innovate, secure, succeed

Data

Data quality management decides where the dashboard only observes

In brief

Data quality management refers, by working convention, to the forum and rules through which an organisation chooses the data it monitors, assigns each gap to a person who decides and determines where to correct it. Management Square offers here a guide to the decisions this management must make, and clarifies the gap that separates it from a dashboard.

Page contents

A quality dashboard can show the same compliance rate for months without any correction being undertaken: it observes a gap, and nothing obliges anyone to draw a consequence from it. Data quality management begins at the point where that gap receives a decision. By working convention, we use the term for the forum and rules through which an organisation chooses the data it monitors, names the person who arbitrates each gap, sets the pace of its meetings, decides where a correction is made and retires the rules that no longer serve.

These five decisions form the outline of this guide. Measurement is not covered again here: our article on data quality metrics gives the formulas, the one on data quality dimensions the classification grid.

A dashboard measures, data quality management assigns

The dashboard is the output of the checks: rates, lead times, a number of open anomalies. It gives the state of the data at a given date. Management adds what measurement does not contain: a chosen scope, a threshold and a course of action for each rule, a person authorised to decide, a log of the decisions made.

One test helps to place an existing set-up. It starts from a gap displayed the previous month and looks for three written traces: the person who examined it, what was decided, the date on which the decision will be verified. If all three exist, management is in place. If the rate has merely been discussed in a meeting, the organisation has a measurement, and the rest remains to be put in place.

The decision log is the document that distinguishes the two. It records, for each gap examined, the outcome chosen (correct, postpone with a deadline, accept the gap), its reason and the person responsible for carrying it out.

The data monitored first are chosen according to use

A forum only examines a limited number of gaps per meeting. Published on gov.uk on 3 December 2020 and read on 12 November 2026, “The Government Data Quality Framework”, the UK government’s data quality framework, asks in its second principle to “Prioritise efforts on the data which is most critical”. It links this choice to knowledge of users and their needs.

We suggest three criteria, examined in this order. Use first: a data item enters the scope because a regulatory return, invoicing or a management decision depends on it, and that use has an owner. The defect observed next: incidents already reported by users point to rules to be monitored before those one might imagine. Finally, the capacity of the forum.

This last criterion can be calculated. Suppose a monthly meeting of 90 minutes, 30 of which are reserved for overdue decisions, and an examination of 10 minutes per gap: 60 minutes remain, that is 6 gaps per meeting. These figures are assumptions, to be replaced by your own. If more rules cross their threshold during the month, the surplus gaps wait for the next meeting. The scope is therefore sized on the number of gaps the forum can investigate; the number of rules the tool is able to check does not enter into this calculation.

Each gap has an arbiter named before it arises

A gap that waits for the meeting to find out whose it is loses a cycle. Our proposal is to write the arbiter into the rule’s record, next to the threshold. In the split described in our comparison of data owner and data steward, the data owner decides on gaps in their domain, on the basis of a file prepared by the steward: the rule concerned, the population affected, the use exposed, the cause if known, the options and their cost.

The forum is only referred what an owner cannot decide alone. Two cases arise. Either the data is produced in one domain and used in another, and the correction costs one what it brings the other. Or the correction requires a change to an application, and therefore a budget and a place in a workload plan.

This forum does not need to be created for the occasion. The one provided for in our page on data governance to arbitrate between domains can take its place, through a meeting devoted to quality. A response time completes the rule: beyond it, a gap left without a decision goes up one level.

The forum’s pace follows that of the uses

Measuring and deciding do not follow the same calendar. An automated check produces a measurement each time the data passes through; the forum, for its part, meets to arbitrate. Its meetings can be aligned with the cycle of the protected use: before a monthly closing, before a periodic regulatory return, before the campaign that will use the file.

A blocking threshold crossed between two meetings goes directly to the owner. Gaps with no immediate stake are grouped together in a less frequent review.

The agenda has three items: new gaps, decisions whose deadline has passed, rules proposed for retirement. A meeting closed several times in a row without a decision may indicate a pace that is too tight, a scope that is too broad or absent arbiters.

Correcting at source or downstream, a choice the forum dates

The same UK framework recommends, in its fifth principle, to “use root cause analysis to solve data quality issues at source, rather than apply temporary fixes”. The levers available and the criteria for choosing between them are the subject of our guide to improving data quality.

Management acts on another level: it records the choice and gives it a deadline. A downstream correction can be justified when the organisation cannot modify the source or a date is approaching. The forum then records it together with the original defect it masks and the day of its re-examination. The list of these temporary corrections is reread at each meeting, and its length shows what the organisation is postponing.

When the cause has not been established, the forum does not choose between source and downstream: it commissions the diagnosis, according to the examinations presented in our article on the data quality audit. The circuit that then carries the correction, from qualifying the anomaly to closing it, falls under our data quality offering.

Retiring a rule is part of management

A set-up that adds rules without ever retiring any can end up producing alerts that no one opens. We suggest examining the retirement of a rule in four situations:

  • the use that justified it has disappeared, or the source application is being replaced;
  • a check at the point of entry now guarantees what the rule verified downstream;
  • the rule has been compliant for long enough that reviewing it no longer teaches anything;
  • its alerts have not led to any decision over several meetings.

In the last case, an alert with no follow-up may signal a useless rule, or an arbiter who does not respond: the examination comes before retirement.

Removing a rule from the review does not necessarily switch it off. It can leave the agenda and remain checked by the pipeline, with an alert only when the threshold is crossed. The freshness and volume signals presented in our entry on data observability can also continue to cover the table.

The rule register then has three states: monitored in meetings, checked without review, retired. A rule that changes state keeps the date and reason for the change, which makes it possible to reinstate it if the use returns.

FAQ

Frequently asked questions

Another question? Ask us

What is the difference between data quality management and data governance?

Data governance assigns roles and sets lasting rules across all data: definitions, classification, access, arbitration forum. Data quality management uses these roles on a narrower object: the gaps observed against quality rules, and the corrections to be undertaken. The first can exist without the second; data quality management without designated owners, on the other hand, has no one to decide.

Who sits on the data quality management forum?

We suggest bringing together the owners of the data monitored, the stewards who investigate the gaps, the function that runs governance and a representative of the team that operates the source applications. The users of a data item are heard when a gap affects their use. Membership is limited to those who decide or investigate; an information session is held separately.

What does the decision log of a data quality forum record?

For each gap examined, the log records the rule and threshold concerned, the date of the measurement, the outcome chosen, its reason, the person responsible for carrying it out and the verification date. The outcome is one of the following: correct at source, correct downstream on a temporary basis, postpone with a deadline, accept the gap. Retired rules are also listed, with their reason. The log is reread at the start of each meeting, before new gaps.

How often should the data quality management forum meet?

The frequency follows from the use the data serves. We suggest aligning meetings with the cycle of that use, for example just before a closing or a periodic regulatory return, and providing for direct referral to the owner when a blocking threshold is crossed between two meetings. The length of a meeting and the time needed to examine a gap limit the number of gaps dealt with, and therefore the extent of the scope monitored.

A project to bring to completion?

A first conversation is enough to understand your challenge, identify the right people and tell you frankly what we can bring.