Data Quality program

4 key questions you should ask yourself before implementing a Data Quality program

An email sent to a client with information about another customer, or a client who has been taxed more than they should have been, or a decision taken based on an incorrect report. Each of these situations carry a high cost for the organisation in which they occur, either reputational or operational. They are all caused by the incorrect treatment of data which could be avoided with an effective Data Quality program.

At Harwell Management we build a governance structure around data quality which can be initiated by answering the following key questions.

1. What are the data quality issues you currently face?

The first action to perform when implementing a Data Quality program is to build a Data Quality inventory. This inventory should include details of all existing data quality issues in the organisation or in a specific department.

The people best positioned to help you build this inventory of issues are the ones closest to the data, such as people who input or update data or the data users. Listen to them and ask them about corrections they frequently do, complaints they receive or incidents they report. This information will help you create the first list of data quality issues.

2. How are you going to prioritize the data quality issues logged?

The prioritisation of the issues logged should be linked to the organisation objectives. Does your organisation want to improve client satisfaction? Then give a higher priority to those data quality issues which make clients unhappy. Does your organisation want to be more efficient? Give a higher priority to the data quality issues which result in increased workload.

The prioritisation exercise should be done and validated by all stakeholders. With this you will ensure that everybody understands and knows what data quality issues will be analysed in the first place. Once all data quality issues are logged and prioritised you can begin analysing them and the most important, solving them!

3. Which steps are you going to follow to analyse and solve a data quality issue?

Harwell Management has a five-step approach to solving data quality issues. By following these steps, you will appropriately identify, measure and control known data quality issues:

Step 1: Definition

Define and document the data elements that need to be analysed to find the root cause of the data quality issue and determine the data dimensions which will be used to measure the quality of the data (e.g. completeness, accuracy, uniqueness, timeliness, etc.).

Step 2: Measurement

Measure the current quality of the data elements defined based on the data dimensions selected (e.g. 85% of values of a data element are completed).

Step 3: Quality Threshold and Benefits

Engage with your key stakeholders, and with them define the minimum data quality threshold expected along with the achievable benefit delivering this data quality improvement. (e.g. 90% of completeness, instead of the current 85%, will increase revenues by 2%).

Step 4: Solutions and Cost

Find and recommend a solution (or several solutions) to improve data quality from the current level to the threshold defined in step three (e.g. cost of going from 85% of completeness to 90%). You should also determine the cost of implementing the solution or solutions proposed.

In tallying costs, indirect costs such as reputation or penalties such as the ones proposed by the GDPR should be considered.

Step 5: Remediation and Control

If the benefit of the solution proposed is higher than the cost of implementing it, you should perform a remediation action. After this a control must be established to inform you to act when the quality of the data is lower than the required threshold.

4. What is your communication plan?

One of the crucial actions to achieve the implementation of a Data Quality program is communication (a communication strategy should be part of a wider Data Governance framework within the company). Before starting a Data Quality program, it is important that all stakeholders are informed about the benefits of good data quality and about the costs of poor data quality, as this will increase the chances to get their full support.

It is also essential that stakeholders are informed along the journey about the progress of the program, the data quality issues prioritised, the solutions implemented, and the improvements brought to the company. Again, this should allow them to see the benefits of the program and the necessity to keep it.

The implementation of a Data Quality program should be based on the answers to these four questions. This program allows you to have a complete and prioritised overview of the data quality issues your company or department faces, and it gives you immediate added value by providing a methodology to remediate them. Consequently, you will be less likely to send emails to clients with other clients’ information, less clients will be wrongly taxed, and less decisions will be based on incorrect reports!


Risk Appetite

Risk Appetite

APPRÉHENDER LE RISK APPETITEComment concilier Risque & Rentabilité tout en répondant aux exigences réglementaires ?Qu’est-ce que le Risk Appetite ? Le Risk Appetite se compose du niveau de risque et des types de risques qu’une banque est prête à assumer pour...



L’ÉCLAIRAGE HARWELL MANAGEMENTOrganisation du travail, veille réglementaire & IALe monde du travail n’a jamais autant subi de changements qu’à l’heure actuelle, tant dans les métiers que dans les aspirations profondes des Français. Nous vivons une véritable...



Le nudge management Pour aller plus loin dans l’accompagnement au changementLes promesses du Nudge  Le coup de pouce, l’incitation douce, en d’autres termes le Nudge Management, est revenu sur le devant de la scène depuis quelques mois, à la faveur, ou plutôt à cause,...



BIRD et IReF Au-delà de la convergence Data, les prémices d’une transformation profonde du Métier Comptable ?Projets BIRD et IReF – Les Essentiels Contexte Face à la demande croissante du régulateur en termes de reportings réglementaires et une exigence toujours...