Tipsheet: 10 Tips for Building an Effective Data Governance Model

Before relying on analytics for all or part of your strategic decision-making, it is essential to implement suitable processes to ensure that data flows smoothly through all business departments while preserving its quality, accessibility, usability and security. Here are Piano's 10 guidelines for building an effective data governance strategy.

1. Diagnose the data assets within the organization

For data to be fully profitable for an organization, it is necessary to know how to select, collect, store and use it effectively, especially as data is both abundant and easily lost.

You can start to do this by taking inventory of all the data present in the company, identifying its various sources (your management systems, websites, social networks, marketing and advertising campaigns, etc.) and then defining the points of friction where there is a loss of value due to poor data quality. Keep in mind the “5 Vs”:

  • Volume: With the growth in the use of connected objects, the development of geolocation and the rise of analytics in digital marketing, the volume of data to be stored and processed in recent years has exploded. Determine the quantity of information held in your databases to guide your data management method.
  • Variety: Data can be complex and diverse, as well as structured or unstructured (voice, biometric, transactional, web analytics, textual, images, etc.). It can also come from a wide range of information systems. Capture it in different places, centralize it and cross-check it to map all your data in an exhaustive way.
  • Velocity: Because we live in an age of immediacy, personalization and predictive marketing, we need to move increasingly quickly and proactively to meet customer needs. Choose high-performance software with powerful computing capabilities that is flexible and incorporates state-of-the-art machine learning. Audit your infrastructure to choose the most efficient tools, in line with your needs, and build a sound technical base.
  • Veracity: This is one of the major objectives in data processing.  The reliability of data collected and processed can be threatened in many ways: declarative errors (forms), the diversity of collection points, the actions of bots, malicious acts and other bugs, human errors and more. There can also be numerous biases in the analysis. It is therefore important to carry out a diagnosis of the quality and accuracy of all your data.
  • Value: The data you use must be perfectly aligned with your organization’s business and marketing objectives and create value for both the brand and your customers. In an environment with an over-abundance of information, it is about being able to unify all your data, and only the data that is useful to you, and act on it swiftly to generate profit or knowledge.

Piano Tip: With our Data Manager tool, you validate each new property sent to your tag before making it available to your company's employees.

2. Unite the whole enterprise around a common data governance strategy

In addition to carrying out the data diagnosis, it is important that all company departments are involved in the use of data, from general management to the operational and field teams, including all team leaders. All employees should fully understand the challenges and benefits of shared, quality, de-siloed data.

To involve the whole company in this transition, consider these phases:

  • Individual or group interviews with the different company departments to better understand the current data situation, determine specific organizational requirements and take into account any expectations regarding data governance. Use the opportunity to make teams aware of the potential risks of exploiting poor-quality or unsecured data, both for their own business and for the company's overall results.
  • Practical workshops, held with the aim of co-constructing a holistic methodological framework for deploying data governance.
  • Real use cases analyzing, with the support of a range of company employees, a specific or recurring business problem associated with a specific data scope. For example, in the e-commerce sector, this could focus on errors in product packaging dimensions which, in cascade, lead to various logistical difficulties and/or cart abandonment because the customer discovers excessive delivery costs. 

The key is to generate interest and launch the data governance project with employees who are receptive, in demand and invested. It is also possible to unite the teams around the drafting of a common data governance charter that outlines the mission, the main objectives and the roles of all those involved.

Finally, establish and communicate the strategic objectives that are common to the whole organization or more specific to each business unit, and then set out all the organization’s performance indicators so that everyone understands their role in supporting the governance model.

Piano Tip: We recommend that you start by defining the key indicators you want to measure in our suite. This will allow you to quickly show your collaborators the value of an analytics suite. You can then expand your data model to meet an increasing number of needs.

3. Choose a data governance operating model that is both appropriate to your structure and agile

When launching a data governance project, you should avoid falling into the trap of trying to tackle all the technical, organizational and regulatory issues at the same time. Overloading everyone's calendars with a plethora of data governance-related tasks will probably jeopardize your chances of success. Don't underestimate the time needed to obtain the first tangible results. Establish a precise roadmap, validated by the stakeholders, with intermediate milestones to evaluate the efforts and progress made so far.

Following the same principle, bear in mind that several different data governance models exist. Choose the one that is best suited to your environment, your needs, your human and financial resources and your data maturity stage.

Piano Tip: In Piano Analytics, your data model is fully scalable; you can add new properties throughout the life of your analytics project. You don't need to plan everything in advance to start your integration.

4. Select stakeholders and identify all data actors

In the past, there has been a tendency to place data governance solely under the responsibility of IT teams. In today's data-driven organization, governance should be implemented and supported across all business units, as each has a role to play when it comes to data oversight.

The first step is to hire or appoint a chief data officer who is responsible for data governance throughout the company. Their role is to get projects approved and prioritized, manage budgets, procure staff for the program and ensure full documentation. Ideally, the CDO should report directly to the CEO. If your organization is leaner, assign this role to another executive at a comparable level, such as a BI & Data manager.

Then, expand the project team by putting together a multidisciplinary group with the following profiles:

  • Data Owner(s): They oversee the data in a specific area or business department. Data owners are responsible for ensuring processes are followed to guarantee the collection, security and quality of data. They must map the data, control access to it, ensure it is protected and define a repository to contextualize the data. In other words, surrounded by an abundance of data, they must determine how particular data is used in responding to a specific problem. So, the marketing director can be the data owner of customer data, the HR director can be the data owner of internal company data, particularly employee data and the CFO can be the data owner of financial data.
  • Data Steward(s): They are the data coordinators and administrators of your data lake, the centralized repository that allows the storage and analysis of all structured and unstructured data at any scale. They are responsible for organizing and managing all data or a particular data entity, with the aim of standardization and compliance with policies and regulations. They capture data elements; they can correct them by ensuring that there are no duplicates in the lake; they can give certain information the status of "reference data;" and they verify the level of confidence and quality of the databases The data steward also has a role in reporting standards and works in tandem with the data engineer (who designs the data pipelines), the data scientist (who designs and applies algorithms) and the data analyst (who, as the name suggests, analyses the data). In tighter organizations, a data manager may fill this data steward role.
  • A Data Custodian: This is more of an IT role, ensuring the control, preservation, transport and storage of data in the company. They are not in charge of data quality issues, which fall under the responsibility of the data steward. As the database administrator, the data custodian ensures the proper life cycle of the data by authorizing and controlling access to the data, defining technical processes to ensure the integrity of the data and carrying out technical controls to secure, back up and archive the data and the changes made to it. In some companies, the data architect may hold the role of data custodian.

Of course, alongside these primary roles, there are also secondary functions such as data analysts, marketing managers, product owners/managers, content or community managers, traffic managers and UX designers who use, consume and analyze data on a daily basis. Not forgetting the support functions, the data protection officer (DPO) is responsible for information, advice and internal control of personal data governance. They ensure the strict application of data protection regulations (GDPR, CCPA, CNIL guidelines).

You can build an RACI matrix (Responsible, Accountable, Consulted, Informed) to model and formalize the roles and missions of each stakeholder.

Piano Tip: Our suite has reporting interfaces tailored to each user profile. Data query for advanced analysts and data scientists, Explorer for analysts and IT teams and Dashboard for marketing and HR teams.

5. Create steering bodies and remove data silos within the company

Once you have assembled your data governance project team, you can bring them together in a dedicated data governance office, a committee that makes strategic decisions about the implementation within the various business units of the company. They approve data policies and standards and deal with any data management, security and quality issues that arise.

Set up one-off or regular sessions with written or spoken feedback if you want to pass on the information and decisions taken to the more operational teams that use the data.

Ideally, you should choose a horizontal mode of governance by putting data at the center of your activity and your business issues. Based on this principle, you can, for example, accelerate the removal of silos between direct marketing, advertising and customer service, and bring together CRM and media expertise and technologies within organizations, brands and their agencies. To achieve this, start by making your employees aware of the benefits of cooperating and sharing data daily. To facilitate the process, you can also set up cross-functional projects that bring together teams that are not used to working together and give them common goals.

Next, ensure that all the data useful for carrying out projects is consolidated on a data management platform or a hub that guarantees the reliability and interconnection of data. It is essential to make all teams aware of the existence of a centralized data asset and to share a common vision of how to manage data between the various parts of an organization to harmonize practices.

Piano Tip: Piano Analytics allows you to centralize and share your data sets. They are accessible by the entire company through a central search zone. You can then integrate them into business-oriented workspaces or share them with a range of business areas across your organisation through Explorer and its Dashboards.

6. Documenting the project and writing shared resources

To successfully implement a data governance project, establish standard processes and adopt a common language within the organization.

An effective way to carry this out, is to provide your teams with a data map, a comprehensive topography of all the data collected and used by the company in the various information systems. It allows the identification of data assets, their flows, their storage and their processing methods. The aim of this process is to make the data fully accessible and understandable to all employees so that everyone can identify the origin of a piece of data, know how it is calculated and spot any duplication. Data mapping consists of several tools:

  • A business glossary: A unique knowledge base common to all employees, it allows for the precise definition of all the terminologies linked to the data in circulation, with the aim of facilitating the exchange of information between the various participants.
  • The data model: This huge table shapes the structure of the company's data and gives information about its storage.
  • A data flow diagram: This provides guidance on the methods of transforming, standardizing and processing data within the different information systems of the company.

Also, the data mapping includes a section on the format in which the different types of data are made available, as well as their conditions of access and use.

Piano Tip: We advise you to make a significant effort on the naming of all your analytics elements so that your employees feel comfortable with the data they are handling.

7. Ensure the quality of your data to improve your decisions

In a data-driven organization, data steers most of your decisions, like the nature and timing of promotional operations or communication campaigns, the segmentation of audiences and the reliability of targeting, the correction or addition of functionalities on a website or mobile application and more. To carry out all these actions, you must have complete confidence in the quality of the data. And using poor-quality data can have serious consequences for your business, such as:

  • Loss of revenue and business opportunities
  • Decrease in the ROI of your actions
  • Reduction of the quality of decisions
  • Contamination of other data projects (CRM, data lake, CDP, etc.)
  • Loss of internal confidence and credibility with your customers

Data can also be altered by a variety of risk factors during its journey:

  • Unmeasured traffic due to missing or incorrect tags
  • Unmeasured traffic due to partial measurement methods (sampling)
  • Overestimated traffic due to bots
  • Traffic blocked by adblockers
  • Overestimated conversions due to poor source attribution
  • Traffic not excluded despite lack of user consent

To avoid these risks, it is important to be vigilant at all stages of the data life cycle starting with the critical moment of data collection, because this phase is permanent. And each modification or update of the website or tracking inevitably poses a risk to the quality of the collection. Put in place effective methodologies and tools to orchestrate and document this process.

First, make sure that the tags are correctly implemented in your tagging plans. Check them regularly and completely, ideally with automated acceptance tests, as manual operation increases the risk of error, and is also tedious and time-consuming.

We believe that this data control process should be extremely fast and easy to monitor, verify and correct. That's why the Piano Analytics Suite offers a dedicated data quality toolkit. It allows analysts and marketers to check, test and modify tags themselves, without the help of technical teams.

Piano Tip: We advise you to create sites for your internal environments, for example a "development" site that will allow you to test the addition of parameters in your tags before deploying them on your production sites.

8. Ensure the regulatory compliance of the data

With the implementation of the GDPR in 2018 and CCPA in 2020 and the French CNIL’s guidelines in 2021, companies are becoming increasingly aware of the importance of respecting the protection of users' personal data on their various digital platforms.

In the event of non-compliance, you may face sanctions ranging from a simple reminder to a heavy administrative fine, as well as severe restrictions on your data capital. Non-compliance can also damage your brand image and lead to a drop in your consumers' trust.  So, it is important to take steps on your websites and mobile applications to ensure that your visitors' consent is properly collected in a free and informed manner. To do this, you need to choose a supplier that has rigorous data management and full respect for the legal regulations.

Piano Tip: The data model provided by Piano Analytics allows you to specifically measure and track the proportion of your users who have not consented to being tracked online.

9. Democratizing the use of data internally

The democratization of data within the company is essential in a data governance approach. It is a process of making data accessible to as many people as possible by adopting a data culture. It means making all the information and resources necessary to carry out their missions and create value available to employees, and only to data consumers. 

To set up a supportive framework for this data evangelization work you can start by:

  • accurately informing all the business teams about all the data managed within the company, its meaning and its context
  • Specifying the use cases for this data,
  • Indicating where it is located and how to access it,
  • Giving details about the quality and reliability of the data,
  • Designating data referents who are able to support users on a daily basis.

Next, we recommend that you set up a specific support program. For example, you can organize training sessions and internal workshops to guide users in the operational use of the tools and in the use of data on specific issues. Also, to encourage all employees to use the data, the data team can design dashboards dedicated to the management of each activity.

Piano Tip: We believe that the most effective way to democratize data is to convey it through reports sent by email as well as through your company's internal messaging system.

10.  Evaluate the performance of your data governance system on an ongoing basis

Finally, once you have laid the first bricks of the foundation of your data governance strategy, it is essential to measure the satisfaction of internal staff, to evaluate the performance of the measures taken, to progressively iterate and improve the processes according to your maturity curve.

Nothing is set in stone in the deployment of data governance - quite the opposite, actually. Be as agile as possible to ensure that the methods applied are in line with the objectives pursued.

Piano Tip: Piano Analytics allows you to monitor the evolution of key metrics over time, so you can visually analyze the benefits of your sales and marketing actions.