July 16, 2021
- Posted by: admin
Learn all you ever need to know about data stewardship and ownership
This article explains the data stewardship concept, the benefits of having a data stewardship program in place, how to get started, and challenges to watch out for.
- The different types of data steward
- What a data steward should do
- The difference between data stewardship and data ownership, and;
- How to set up data stewards for success
Read on to learn more…
What is Data Stewardship?
Data stewards play a pivotal role in any data governance initiative. We need them to help us wrap our arms around the vast quantity of data that exists in any enterprise today. Data stewards ensure our data is documented, high quality, and fit for use by all enterprise stakeholders. As a result, picking the right person is key.
Let’s take a quick look at the basics before we begin:
What is stewardship and where does data come in?
Let’s start with the definition of stewardship. The Oxford English Dictionary has this to say:
Stewardship is: the act of taking care of or managing something, for example, property, an organization, money or valuable objectsOxford English Dictionary
They go on to give an example usage: “The organization certainly prospered under his stewardship.”
So far, so simple. The first set of examples we see are “taking care of property, an organisation… or valuable objects“. This leads us to realise that we need to treat data as an enterprise asset.
Merriam Webster includes this on stewardship:
The conducting, supervising or managing of something. Especially: the careful and responsible management of something entrusted to one’s care
So again, we see that the role of a steward is in the careful management of something valuable. In the case of data stewardship, that valuable thing is the enterprises’ data assets.
What are data stewards?
Data stewards are the glue that binds our data governance program together. Appoint key people into these roles – they must ensure data in their care is fit for purpose. They are a specialist role in the organisation. To succeed they need a deep understanding of the data and how it is used. Most importantly, they are responsible for the upkeep of this data on behalf of the rest of the business.
This means that data stewards exist to take care of data on behalf of everyone. Because of this, we appoint people to this role with the right skills to perform it. Firstly, the data steward needs to have a deep understanding of the data they maintain. Secondly, they must have the ability to see beyond their silo or role and ensure data is fit for everyone to benefit from it. Lastly, they must roll up their sleeves and get to work to put rules and processes in place that make the data in their stewardship as good as it can be.
Why you need a data steward
Take a look at the data in your organisation today. Ask yourself, is this data fit for purpose? Does it meet the needs of all our staff, systems and processes that run this enterprise? Or is it of low quality, leading us to waste time trying to understand basic facts about our business, such as “how many customers do we have?” or “are the invoices we raise accurate?”.
Unless you have someone playing the role of data steward, it is likely that your data needs to be improved. Whilst we may periodically improve data in the absence of formal data stewardship (e.g. asking the sales team to update customer contact details), unless this is an ongoing process the data degrades. Dun & Bradstreet state that data degrades at up to 2% per month. This is especially true of customer data, where people change addresses, jobs, telephone numbers and email addresses on a frequent basis.
The result of this is chaos and cost in any large enterprise. Executives waste time trying to understand business data. You waste money sending products or service teams to the wrong address. Confusion reigns when executives have different definitions of the same important business terms and facts.
3 reasons to appoint a data steward
The first reason to appoint a data steward is to ensure we have someone in the organisation that is responsible for keeping data fit-for-purpose. Unless there’s a formal role allocated (even if only part-time) you won’t get consistent improvements in data. By allocating responsibility for data to a specific person, we ensure there is a go-to resource that will keep data in good shape.
The second reason we want to appoint a data steward is to document the knowledge and experience that our people have about our data. We call this “metadata management”. A simple example of this is a description that tells us what data means. For example, a list of numbers “9750 3525, 8586 8688, 3427 8898” etc is relatively meaningless. This could be about anything. If we describe the metadata about this data (e.g. a Column Header that says “Phone_number”), we can now understand that this is a list of telephone numbers.
Context is king
Obviously, this is a basic example. There are numerous negative consequences if we fail to provide meaning and context to data that is used broadly across our business. Data stewards help us avoid this confusion.
The third reason we appoint data stewards is to collaborate with other resources. The objective is to understand how their data is used throughout the business. Your data steward must be an expert about the data in their domain. At the same time, they need to work cross-functionally to ensure data is fit for the purposes of all your employees, systems and processes. As they go about this work, they will document quality expectations for the data that will help define rules for data creators to follow.
How important is data stewardship
For any data-driven business this role is essential. You can think of the data steward as the quarterback of the data governance world. This means they’re holding together the definition, documentation, improvement and use of the data under their stewardship.
Without the right people in the right roles, you can’t expect to win. Data is increasingly seen as a strategic enterprise asset. It comprises up to 87% of an organisation’s value, so we have to deploy resources to protect it.
The context of governance roles
You can think of this in the same way you might think of your finance function. The finance team handles the cash and monetary assets for your business. This means they put rules down around the use of this cash – for example, expenses policies. You have people responsible for designing and implementing these checks and balances in order to maximise the return on your enterprise assets.
Our data stewards will provide the same level of assurance over data assets. This means:
- Documenting acceptable use and privacy policies for data, which is essential under regulations like GDPR
- Increasing trust in data, by defining its true meaning, and ensuring that data quality levels are acceptable
- Documenting the provenance of your data. Where does data come from and how is it used across your business?
- Preventing misunderstanding, which directly leads to improved business performance
- Enabling the organisation to make better decisions, more quickly
As you can see, without a business person deliberately taking this role, the data in the organisation is unlikely to be fit-for-purpose.
What a data steward does
The data steward is responsible for the upkeep and definition of data and metadata used in their domain. Specifically, this means:
- Documenting the meaning of data
- Stipulating data quality expectations
- Setting rules on how the data can be used/stored/archived/deleted, and;
- Ensuring these definitions and rules are both clear and enforced
Documenting the meaning of data, by enriching the metadata, is one critical area of focus for a data steward. The bulk of this work includes defining and documenting business terms and capturing these in a business data glossary. The result of this work benefits the rest of the organisation who rely on the data steward to make data easy to understand, find and use.
Data quality responsibilities
Creating high-quality data is one clear reason to appoint data stewards in the first place, so it’s no surprise this is part of the job scope. To do this effectively we need to define the data quality requirements of the business, then implement these as rules and standards. Data quality checks deliver ongoing monitoring of the data, which is also a responsibility of the data steward. When data quality issues are uncovered, the data steward can help to drill back to the root cause. They can communicate with data creators to ensure everyone is aware of what good data looks like, and their role in creating it.
Data privacy laws are a major concern for many enterprises. Data stewards help classify our data. This helps mitigate the risk of being fined.
Data stewards should also help classify data for internal use. This will ensure only authorised personnel can see or use the enterprises’ data assets. Data sharing agreements internally and externally are essential. To create them we must first classify what purpose(s) the data can be used for.
The responsibilities also include translating regulations into data policies. These policies are used to guide other staff on how to create good data and use it responsibly and ethically.
Types of data steward
All data stewards are going to perform the same types of tasks and activities, so why have we included a section that classes stewards into different roles? Put simply, because your data governance framework must be unique to your organisation. There are many ways to structure the framework. The type of data steward you need depends on the goal of your data governance program as well as the way your firm is set up and run today.
Here are four different options to consider, with the pros and cons so you can decide the right approach for your business.
Data Object stewards
A data object steward focuses on one critical business object (or entity). A simple example of this is a customer data steward. The data object is “Customer”, and the popularity of this type of data steward is that one business person is responsible for any data related to one data object.
Because a data object like “Customer” is broad, this type of steward must be able to operate across functions and silos in your business. Their work will impact the definition of a customer, the data attributes you capture about your customers, and they will be central to any effort to standardise the Master Data about this entity (e.g. a Single Customer View project).
Benefits of a data object steward
The benefit of this type of stewardship is that you have one person that’s responsible for data about your critical business entities. You may have a customer data steward, an employee data steward, a product data steward etc. By centralising responsibility to one individual, you highlight the importance of this business entity. As a result, this data steward is required to have deep knowledge and strong stakeholder support to do their role.
Drawbacks of a data object steward
On the other hand, challenges for this type of steward arise in decentralised businesses. The ability for one individual to document one business entity is much harder when the organisation has multiple different operating units, with different levels of autonomy. Using our example of a customer data steward, there could be different types of customers depending on the operating unit. For example, a bank can serve Retail customers as well as Commercial customers.
Business data steward
This role is perhaps mistitled, as all data steward roles should be from the business. What it means is that you deploy a single person for the data within one business unit or function. Of course, this leads to the need for a more rounded understanding of the data within that function. As a result, this type of steward must be knowledgeable about a range of data domains.
Benefits of a business data steward
In contrast to the data object steward, the business data steward role works well when there’s decentralisation. In our example of a bank, having business data stewards would lead to the creation of a Marketing data steward for the Retail bank department and a separate Marketing data steward for the Commercial bank department. This makes sense because the type of marketing channel, spend, and information for an individual would be significantly different than for a company.
The narrower scope of the business data steward also makes it easier to measure the impact of this role. For example, our marketing data steward can easily demonstrate their value to the business unit by reducing the cost of bad contact data (e.g. physical mail offers being sent to the wrong address).
Drawbacks of a business data steward
This role becomes more challenging when data must be shared between several business units.
For example, you might have a customer called Mr Robert Walsh in your retail business. The marketing data steward has done a great job cleaning duplicates from the system, harmonising Bob Walsh, and Robert J. Walsh into a single record.
However, the Head Office may want a report to understand their entire business relationship with Robert. Perhaps he also owns a company that does business with the Commercial Bank. If the marketing data steward in the Commercial Bank has taken the approach to use “Bob Walsh” as the standard, it may not be possible to link the two entities together, and reporting at HQ level suffers as a result.
Process data steward
Business process data stewards are responsible for the data flowing across critical business processes. Examples include Customer Onboarding data stewards, Billing data stewards, Employee Onboarding data stewards etc. This type of steward works well when you are improving business processes, and can help create detailed data lineage documentation from source to target.
Benefits of a process data steward
If you are documenting or reengineering your business processes, the data stewardship function can be absorbed into the cost of this program. For example, you might want to improve your customer onboarding process, and have a team set up to document and improve customer journeys. This team would need to establish their data requirements for engaging the customer anyway, so it is an ideal opportunity to document data touch points and handoffs between all systems that interact with customers (or customer support staff).
Drawbacks of a process data steward
For this to work, you must have good data governance oversight. The governance oversight is there to ensure the work of multiple process data stewards are aligned. As an example, our Customer Onboarding data steward may define customer data in a way that minimises friction when creating a new customer, perhaps by eliminating fields on signup for the customer address. If another business process needs this data (e.g. Billing), there’s a conflict between processes. Having a data governance council that can intervene and establish good end-to-end governance is essential for success.
System data steward
The fourth and final type of data steward is the system data steward. Sometimes this role is also called a Technical data steward or an IT project data steward. This is a common form of stewardship before a formal data governance program has been deployed. For example, you may have a CRM data steward or a data warehouse data steward.
Benefits of a system data steward
We most commonly see this type of stewardship in companies that score less than 2.5 on our data maturity assessment. This is the Initial or Repeatable level, where data governance has often been established by the I.T. team alone. By making someone responsible for data within a system, the I.T. team can quickly enforce better rules around the data within that system. It’s the first step toward engaging the business, as the system data steward will need to work with business users to do their job effectively.
Drawbacks of a system data steward
Long term, an I.T. role is not the right choice for data stewardship. You must involve the business and engage them in these roles for optimal outcomes. Furthermore, the system data steward will lack a detailed understanding of the business uses of the data. As a result, this can be considered a stopgap type of data stewardship until a formal process is put in place.
Who should be a data steward?
First, the good news. Anyone in your firm can play a data stewardship role successfully. In fact, we want all our staff across the enterprise to take the stewardship of data seriously, and to understand the benefits of getting this right. As a result you can consider the role of data steward more as a relationship between the person and the data in their care, and less of a formal job title. It’s more important to get the activities right than the naming.
So how do we prioritise staff for this role? Who should be considered to support us? And how do we support them?
Attributes of a data steward
Firstly, we need someone that has some degree of technical understanding. Whilst this is a business role, we are still dealing with data. The person must have at least a basic knowledge of data, data modelling, and data architecture. Business Analysis skills will be very beneficial. This is because the data steward will often be the bridge between business and I.T.
Next, we need the person to have business acumen. If you can find someone that knows your business processes inside-out, you’re on to a winner. Typically you’ll be able to identify the business data steward easily. Ask yourself “who is the go-to authority on this data subject area today?”. There’s almost always one person that stands out. If this person is already answering questions on this data element today, they’re almost certainly the most knowledgeable resource to appoint to the job.
Last, our data steward needs to have some personal qualities. Many firms fall down by appointing someone that loves being the central point of focus for questions on this data. We can’t have that going forward. To succeed we need to document the knowledge in a way that helps us democratise it throughout the firm. As a result, the following characteristics are desirable:
- Carries respect
- If the data steward is already the authority on this data in your firm, you’re onto a good start
- Builds consensus
- Just because they are the expert does not mean other opinions can be ignored
- Manages up and down
- They must be able to get action from their peers and escalate when more help is required
Data Steward Responsibilities
Adding to the above, data stewards need to be able to:
- Be the expert about the data they are responsible for
- Understand what the data is used for and how it is created
- Create definitions and document knowledge about the data
- Take responsibility for the quality of data in their domain
- Guide policies on how this data should be created and used
- Bridge out to the data management team to ensure the data governance rules are applied properly
Data Stewardship vs Data Governance
This might seem like a strange end to Chapter One, but in my experience there is a lot of confusion about these topics. I’ll start by providing Cognopia’s definition of Data Governance:
Now you might be thinking “why have they added yet another definition of what data governance means?”. Please bear with me. This definition includes the following 6 requirements:
- That you run data governance as a business program
- You create a set of guiding principles to direct staff on how to manage data
- These principles are codified into policies that dictate what data management practices you enact
- Roles, responsibilities and decision rights are allocated amongst staff to ensure policies are enforced and working
- That business functions are established to manage these new roles, and;
- That metrics are documented to measure the success of this initiative
Where do stewards fit in?
Great question. As you can see from this first chapter, a data steward is a role that looks after data. This means they are one of many roles defined in step 4 above. Based on the first step, you can see they are a business role first and foremost.
The data stewards will help translate the principles you create in step 2 into policies you deploy in step 3. They will be part of the new business functions you set up in step 5. And lastly, their performance will be measured by the metrics you agree upon in step 6.
Data governance is the broader program of work. It entails placing rules and obligations on data activities across the firm. You need executive engagement to help direct this work and to ensure it aligns with business priorities. You’ll also need someone with seniority to break deadlock decisions, and fund the program. Data stewards are the all important foot soldiers that turn the overarching vision and mission of the data governance program into reality.
Chapter 2: How does Data Stewardship Work?
In this chapter, we’ll expand upon the descriptions of chapter one. The goal is to paint the picture of the day-to-day tasks of a data steward. By the end of the chapter, you’ll know what your data stewards should do, and where they sit in your organisation. Let’s crack on…
Data Stewardship tasks and activities
First and foremost, data stewards are there to document the most important data within their area of expertise. One key challenge we see in many data governance programs comes from teams that are not focused. When you clean and organise data without a plan, you run into trouble. Specifically you end up “boiling the ocean” and not delivering business value.
As a result, one of the first tasks our data stewards can participate in is to identify the most critical data in their domain. This is not purely a task for the data steward (see “Data Stewardship vs Data Ownership” for more details). However, the data steward will play a very large role in this task.
Now we know what data we need to focus on first, let’s look at the other tasks our data steward must complete.
Metadata Management and Data Classification
This is one of the core functions of a data steward. As noted above, metadata is simply “data about data”. Unless we have a data steward documenting this information, meaning and context can get lost. Data quality will suffer. Let’s take a deeper look at how our data steward can help.
Technically, identifying Critical Data Elements was the first step of this process, but it’s so important I wanted to highlight it up front. As you go through the process of identifying your critical data, you can enrich it further by documenting other key areas:
Every business has its own language. You may not realise, but I bet you use a dozen acronyms on a daily basis. When you first started working, these were like a foreign language. But over time you got to know them, and now you’re speaking gibberish whilst understanding it like a native. To ensure the rest of your enterprise can understand your team, the data steward needs to translate. One of the major tasks to eliminate ambiguity in your business is to capture this language. The data steward will contribute it to your business glossary or data glossary, which will act like an enterprise-wide Rosetta Stone.
Data lineage sounds complex, but it’s relatively simple. Imagine your data going on a journey, from the moment it is captured to the moment it is consumed in a report. This is your data lineage. Like a precious painting, you want to know the provenance of your data so you can trust its authenticity and accuracy. Your data stewards task is to trace this route. Much like a data Indiana Jones (maybe with less swashbuckling), the data steward must dig through your processes and systems to map out where your data came from.
This work provides a map of how data is used, who uses it, what their quality needs are, and who should be responsible and accountable for its quality along the journey. Critical work indeed.
Lastly, we need the data steward to document any important business rules about the use of this data. Are we allowed to share it? For what purpose can it be used? When must it be destroyed or archived? All of this needs to be written down so the data management team can do their job.
READ MORE: Metadata Management – NOTE THIS NEEDS TO BE LINKED ONCE THE CONTENT CLUSTER IS UP!
Data Stewardship and Data Quality
Data must be of high quality if we want to gain business benefit from it. Put simply, data quality improvements will deliver the biggest ROI of your data stewardship programme. Our data stewards lead the line in documenting quality expectations and ensuring data in their care is fit-for-purpose. But what does that mean? And how do they achieve it?
What is data quality anyway?
Beauty is in the eye of the beholder. Or so the saying goes. The same is true of data quality, and quality in general. If we think of something common – say a dinner out – then the quality of that dinner depends on the context. If I’m a hungry business traveller rushing for a plane, “high quality” is anything I can grab quickly and eat before takeoff. I don’t need a filet mignon and fine-dining service. If I’m a man taking his wife out for their 10 year wedding anniversary, a takeaway sandwich from the local bakery won’t meet my wife’s quality expectations. She will want a nice environment, probably some candles, definitely some champagne, and the expectation is of a top-class dinner prepared with the best ingredients.
Where data fits in
Data quality is no different. If I’m a sales guy trying to close a deal, I might only need to know the name, email and phone number of my prospect. Bam! Deal closed! I get paid my commission and the data was “fit for purpose”. Unfortunately, the data about that customer is going to be used elsewhere in the business. The Finance team will need to set up the customer account. They might have needed credit scores before I closed that business to protect the firm from bad debts. They definitely need the invoice address and contact details of the person that will pay any invoices. Without this information, our newly-minted customer record is “bad quality”.
So how do they achieve this? What tools do they use?
Data quality processes
As you can see above, the first step was getting our business to tell us what they need to do their job. This may involve observing the business user as they go about their task. Sometimes our staff are so used to dealing with bad data quality they forget how much time and effort they waste battling it. 37% of companies we speak to say their data is poor quality, but they have no processes in place to fix it.
Data profiling is a process that documents the “as-is” state of data. You can do this manually, or you can use one of many tools on the market. In our 2021 research, we discovered that 63% of our respondents relied on in-house scripts for managing data quality. Your I.T. team can help, if they know what to look for.
You must profile your data with your business colleagues. Whilst your data steward should be able to spot data errors alone, many hands make light work. Profiling will highlight common data issues:
- Missing values
- Incorrect values
- Multiple/inconsistent/incorrect formats
- Duplicate records
As you identify each issue, document it and define the data quality rule to match. For example, if there are many missing “first name” values, and we always aim to call our customers by their first names, we need a data quality rule that stipulates “first name cannot be null”.
Data quality rules
Armed with the rules above, you need to apply them. Again, your I.T. team can either write some scripts and put some reports together, or you can pick a proper data quality tool. The important part of this is that the rules you created in step one are constantly monitoring data quality.
Whenever you discover a rule that fails, the data steward steps in. Typically the immediate fix is to clean the data where you identified the problem. For example, you might spot missing first names in your customer records. You trigger a process that alerts the sales team, who then add those first names manually. Good job! However, doing it this way is time-consuming and wasteful. A better approach is to fix the root cause of the issue.
Root cause analysis
Root cause analysis is a process that seeks to identify why your data quality problem arose in the first place. By digging back to the root cause, you can apply a policy or rule at source and prevent data from ever going bad. Winner!
To do this, you can use techniques such as:
5 whys in action
In our example of the customer record, we know the first name field is missing. Let’s ask why:
- The first name field is missing because the sales person did not capture it; why?
- The CRM allowed the sales person to save the record with no first name; why?
- There are no data validations on capture for the first name field; why?
- Because we have yet to write a policy mandating this rule is enforced
In this case we only need 4 whys. Our data steward writes a policy that tells everyone a Customer record is only valid if there is a valid first name. Their colleagues in data management pick up this policy, and change the CRM to reject new customer records without first names. From now on, the problem of missing first names has been overcome.
READ MORE: Data Quality Management – NOTE THIS NEEDS TO BE LINKED ONCE THE CONTENT CLUSTER IS UP!
The Data Stewardship team
So far we have focused on the role of the individual data steward. However, our data stewards are not operating in isolation. Given the breadth, depth and volume of enterprise data they need to be part of a team.
There’s no need for this team to be huge. The data steward just needs support to get their work done. The data steward leads this team, hence they need good interpersonal and leadership skills.
Subject matter expertise
The data stewards themselves should hold significant knowledge about their data. At the same time, they cannot be expected to know everything. Bring together experts that can collaborate to document and describe the data. This additional expertise could fall under a few categories:
- Data subject matter expert
- Bring in specialist knowledge to describe the meaning of critical data elements
- Augments the knowledge of the data steward
- Business process subject matter expert
- Deep knowledge of the business processes that create and consume data
- Useful for mapping data lineage
- Their knowledge is used by the data management team to improve data handling processes
- Regulatory expertise
- If you are unsure of new regulations and their implications on data management, bring in legal experts to guide your data classification
- You may also be a highly regulated industry, in which case bring in resources that engage the regulators for their input and to keep them informed of your progress
- Data quality expert
- You may need technical resources to apply data profiling techniques or use the data quality technology
- Bring in specialist skills to help root cause analysis
- Project management
- Coordinating the tasks involved in improving data is hard, get help from experts in your firm to keep the project on track and the stakeholders informed
Bringing the right roles in at the right time will ensure your data stewards succeed.
Data Stewardship tools
This is often the first place data governance teams look to make improvements. At Cognopia we would recommend that you start off manually, using Excel and other free tools, then select and deploy technology to automate once you have the People and Process side nailed down. At the same time, we are practical and pragmatic, and no guide to data stewardship would be complete without listing technology.
Let’s look at some common technology to support the data steward.
Data Quality/Profiling tools
We covered what data profiling is in this section. Many tools exist, and your I.T. team can run queries using SQL or other scripts. These are probably the only tools we would recommend at an early stage. Profiling technology is fast and can help bring data problems to light quickly. Use free versions to demonstrate data issues and get senior management buy-in.
Once you have a working data quality process in place, you’ll need to monitor it. At this stage, adopting a data quality tool to implement your quality rules and report on any issues will help.
Data documentation tools
In this section we talked about the need to document metadata. Begin using a simple business glossary on excel. This will get the basics right, and will show you what you want or need to automate. Once you’re comfortable with this process you may want to shop for something more robust.
There are lots of tools to help document metadata. They range from reasonably cheap to very expensive. Key features to look out for:
- Basic – Data Dictionary/Glossary type tool:
- Captures data definitions and business terminology
- Documents formats and data types
- Lists responsible data stewards/data governance roles
- Medium – Metadata repository tools (as above, plus):
- Automated ingestion of metadata
- Visually represent data lineage
- Some tools can automatically build lineage
- Includes workflow to route data queries and issues to the right people
- Advanced – Data catalog tools (as above, plus):
- Allows the formation of curated data sets
- Encourages additional notation on data uses by data curators
- Workflow processes included for data access and sharing requests
- Often includes peer review of data sets
- Advanced search/data discovery features
This is only a basic overview of the features that exist. Make sure you purchase based on a sound set of business requirements. Do not go shopping to buy tools just because you can.
Many of the standard tools have reporting built-in. At the same time, data stewards may want to consolidate all their reports onto a single technology. Consider creating reports for:
- Data quality status
- Open issues logs
- Data glossary progress (how much progress your team is making in documenting critical data)
- Dashboards and scorecards
Data Stewardship challenges
You can’t make an omelette without breaking eggs. In the same way, your data stewardship program will also run into roadblocks. Here’s a few of the common ones and how to overcome them.
Lack of authority
Your data steward needs to be respected. A later section breaks out the difference between data stewardship and data ownership, but even with data owners in place the stewards need to command respect. When you appoint someone with limited authority, you run the risk of failure. Data stewards must be able to lead the data stewardship team. Without respect and authority they will fail in this task.
The culture of a company dictates how it behaves. Unless you are very senior, the chances of changing this culture are limited. As a result you have to design your data stewardship program to work within the current culture. Some challenging cultural battles include:
- Data hoarding – when teams refuse to collaborate for the greater good of the enterprise
- Flat organisational structures – leaving no overall decision maker to break deadlocks
- “Just do it this way” – if the data steward is not trusted to achieve their goal, you have the wrong person in the role
In our 2021 Data Maturity Report, we discovered that 33% of firms have no budget for data governance. Another 37% rely on existing I.T project budgets for ad-hoc data governance. No program of work that manages enterprise assets can thrive without funding. Imagine telling the CFO she can’t have budget for a team of people to track expenses. You’d end up in chaos. Managing data without funding is similarly suicidal.
The wrong people in the wrong roles
When data governance is kicked off in I.T. there is a tendency to get the wrong people in the wrong roles. Technical data stewards are better than no data stewards. On the other hand, they are not as good as having a business person responsible for this work.
Data governance work is seldom seen as exciting. This creates a further challenge. The kind of resources that will make a success out of data stewardship are likely in demand. You need to make the case for getting their time and help to improve your data. Otherwise you’ll end up with a team full of resources that can be spared (i.e. no one wants them on their team). In other words, people that aren’t much use. In this situation you are unlikely to succeed, so rather than accept this team you should go back to the drawing board and make a better business case for change.
Chapter 3: How to implement Data Stewardship
So now you know what stewards are supposed to do, how do you actually implement this function in your business? Let’s start with the most important thing, getting stakeholders engaged. When we measured the data maturity of firms last year, the difference between those with the most engaged vs the least engaged stakeholders was a whopping 1.16 points (out of a possible 5). Get this right and the rest follows.
Securing executive sponsorship
In many firms these days, the executive accountable for data is called the Chief Data Officer, or “CDO”. If you are fortunate enough to have a CDO, then you ought to have data stewardship in hand already. If not, read our guide on the CDO’s first 100 days.
If you are yet to appoint a CDO, you’ll need someone with clout and money to help get this off the ground. Let’s look at how:
Make a business case
Unsurprisingly, we found that 64% of firms that lack a data governance budget also lack a data governance business case. If you want to make a change, you need to spell out WHY.
Your business case must be aligned to your data strategy (see below). The data strategy must support your business strategy. This may sound complex, but it’s easier than you think.
Data strategy, please
A data strategy should align the use of data in your firm with the business strategy. Broadly speaking businesses will be looking to do one or more of three things:
- Increase revenues
- Decrease costs
- Reduce risks
You can find these listed out in your Annual Report (if you’re publicly listed). If not you should set up meetings with senior executives to understand the company strategy.
Let’s take decreasing costs as an example. If your firm is in cost-cutting mode, look for areas where data is hurting the bottom line today. Specifically, find data quality problems that you can fix. Bad data quality will cost your business money, customers, time and increase risk. Consequently, you may discover that there are errors in your invoicing data (missing contact details). Perhaps you discover that your invoices are being paid late because of this.
Where are the costs?
- Cost of staff time
- Accounting staff search for information that is not available
- Customer service staff are asked to call to chase for payment – more time cost
- Cost of late revenue
- Your business has a “cost of capital” – every day an invoice remains unpaid is like loaning another business money without charging interest
- Cost of customer distrust
- Harder to measure, but your customers may start to ask “if they can’t get my invoices right, should I really do business with this firm?”
- Look for changes to Net Promoter Score or CSAT metrics, and align these to lost revenue and dissatisfied customers
Align your data strategy against a business strategy. Spell out how much data is contributing to the business strategy today. Then pull it all together to make the business case for investing money in improving your data tomorrow.
Bring it all together
The business case lays out why you are going to govern data. It should demonstrate how the data stewardship program is going to add value to the business. Once you have engaged a senior stakeholder with these draft numbers, you need a roadmap to follow.
Data Stewardship roadmap
You have identified major business challenges the data strategy will help to solve. You’ve put a dollar figure on the improvements and made a business case. Now is the time to bring it together and link the outcome with the actions to get there.
First, create an overarching principle that will guide behaviour in the absence of fixed rules. For example: “we will treat enterprise data as an enterprise asset”. Guiding principles set the tone of your program. This principle tells staff that data is important to your organisation. Other principles can be added to expand upon the ambition. The purpose of this is to set the overall direction for your data roadmap.
Coordinated tasks and actions
Next, you translate the overarching principle into a sequence of coordinated tasks and actions. Document the gap between where you are and where you need to be.
- What organisational change is required?
- Who will need to play what role?
- Do you need new technology to help?
- Where will budget be spent, and on what?
By mapping these items out you can build a coherent plan of action. As a result, your roadmap will take shape. You will know what to do, who to involve, what tools they’ll use and how much it will cost.
The Data Stewardship framework
It should be clear by now that there will be more than one data steward in your organisation. The data stewardship framework describes the roles and responsibilities you will deploy in your firm. It must be bespoke, and whilst there are some common themes, there is no “one size fits all” framework to deploy.
We have dedicated an entire chapter of this guide to the difference between data ownership and data stewardship. A data owner also plays a role as a data steward. For this section, all you need to understand is that the data owner is accountable for the data whereas the data steward is responsible for the data. Data stewards are there to do the work, data owners carry the can if the data is not fit for purpose.
One decision you need to make is to choose the right type of data steward for your business. Let’s assume you’ve chosen to deploy data subject stewards. Perhaps you’re keen on improving your customer experience, so the data subject would be “customer”. Your data owner is likely to be the Head of Sales or Marketing, or perhaps a Chief Customer Officer. They will be supported by one or more customer data stewards, depending on the size of your firm.
Data Stewardship Councils
There need to be working groups that come together to get the job done. These are part of your data governance function. Some companies call these data stewardship councils or data stewardship committees. Fundamentally there are usually 3 different layers required:
- Working group level
- In our example, the customer data steward will convene weekly. This will bring together I.T. and business subject matter experts to progress the work of governing customer data.
- Most of the work is done at this level, with any issues that this team can’t resolve pushed up to the next level.
- Cross-steward council
- Our customer data steward won’t be the only steward. We may also have a team of data stewards in the finance function, for example. They don’t work in a vacuum, as some of the decisions made about customer data will impact the finance team too.
- Cross-steward councils allow for these teams to coordinate their work and learn/share best practices with one another
- Executive council
- This brings the executive stakeholders into the conversation, and as a result it means they meet on a monthly basis.
- Any issues that can’t be resolved by the other 2 councils will be decided here.
- This council is also responsible for providing overall direction to the programme, and funding any new projects or requests for resources
Ensure senior stakeholders engage
The executive council could be an existing steering committee, or a new group solely focused on data. Our research shows that 54% of firms report that senior stakeholders are not at all engaged, or only some are engaged in data governance. This leads to substantially lower data maturity, underfunded initiatives, and poorer performance overall.
Data stewardship artefacts
Data stewards have a few tools in their arsenal to help with their work. Beyond the data steward tools, our team will need some standard documents to get moving. Start on paper (or on Excel/Word) and automate once you get the process going. Here’s a list to begin with:
- New data element request
- Used by end-users that want the business glossary updated with new definitions or additional fields
- Data quality issue form
- Provide a mechanism by which data quality issues can be reported
- Exception request
- Used to request an exception from using already defined data standards
- Data Governance Council meeting notes/log
- Create a standard template so your meetings are well run and any issues are documented
- Ways of working document
- Describe how meetings will be chaired and run, and who and how you respond to issues raised in the forms listed above
You must track KPIs that document your progress. In Cognopia’s 2021 Data Maturity Report we discovered that 77% of firms do not track change related metrics and KPIs, or do so in a poor fashion. We are data professionals, yet we do not measure our own impact on the firm we work for!
KPIs need to be useful to the business. Your data strategy will dictate the value your work will bring to the business. Measure progress against it:
- How much time/cost did your data quality program deliver?
- How broadly is your business glossary being used?
- Define the growth in numbers of business terms captured through time
- How much time and effort is the glossary saving your teams?
- What is that worth to your business?
- Have you managed to rationalise databases or reports?
- If so, how many?
- How much did this save your business?
- How widely have you classified private data?
- What is that worth in risk-reduction to your organisation?
The KPIs should be linked to commercial value as far as possible. This will allow you to justify the ongoing investment in data governance, and prioritise your roadmap activities over time.
Chapter 4: Data Stewardship vs Data Ownership
So by now it should be clear that data stewardship is a hands-on role that is essential to improve and maintain your business data. At the same time, it’s clear that you need senior stakeholders involved in data governance to succeed. These are your data owners. In short, a data owner can be considered the most senior data steward in their domain.
Who are data owners?
So if the data owner is the most senior data steward, who plays the role? Let’s take a look at an example that might clear this up.
Getting finance data right has a direct impact on the bottom line. Making it easier to send accurate, timely invoices and chase payments has a large operational impact on many businesses.
Your CFO will be delighted if you remove “Excel Hell” from the month-end processes. In many large firms you have armies of accountants, all running manual journal postings to the general ledger. Closing the books on time becomes a major headache.
Where does data ownership come in?
Who has the problem in the scenario described above? Certainly the CFO, but they are probably too distant from the day-to-day challenge in a large firm. In this situation, the CFO would make a better executive sponsor (they may still be the right person in a smaller firm).
Who has the budget and authority to fix this data? Again, in an enterprise setting it may fall on the Controller or Head of Accounting to play the data owner role. In a smaller firm it may be the Finance Director or CFO.
In a major enterprise, these roles will lack enough “hands on” understanding of the day-to-day data challenges. As a result they will delegate authority to data stewards. Consequently the stewards will be responsible for carrying out the role of governing this data.
What does data ownership mean?
Let’s look at the basics. What does the dictionary say about accountability:
the quality or state of being accountable especially : an obligation or willingness to accept responsibility or to account for one’s actionsMerriam-Webster
This may seem overly basic, but we have seen major issues when this is not spelt out. Data owners initially say they’re OK with the role, but then they go AWOL when the real work needs to be done. The fix to this is to change KPIs to enforce accountability. If your data owner fails to improve the data in their domain, they face consequences.
This may seem like a bad deal. However, we are all used to this in any given company. If I am given a work laptop to use, I’m responsible and accountable for keeping it safe and free from viruses. The same must be true of the data assets we create or use.
Why do we need data ownership?
In smaller firms, data ownership takes on both accountability and responsibility for data. There is no need to split the roles. However the larger the organisation, the less likely it is that one person can govern data in their domain.
Management structures and decision rights are normal for managing other elements of our business. We can leverage these power structures to manage data too. By assigning accountability to those with power we can leverage their ability to mobilise teams and resources. This ensures that we have the right support, both financial and from manpower, to improve data across the business.
Data ownership benefits
Clearly the power and influence the data owner brings helps “get things done” in improving data. There are other key benefits the role brings:
- Closer to the organisational strategy
- Useful when aligning the data improvement work to the business goals
- Data owners are more likely to know where the business is going than the “rank and file” staff
- Helps prioritise data improvements to areas with business value
- Understands the pain of bad data
- They will see the financial cost of bad data in their domain
- The data owner can direct improvement efforts to areas with the greatest dollar benefit
- Helps when you need to buy technology
- At some point you will need to invest in technology to automate your processes
- Data owners can help fund this, you need to capture their requirements well
If you lack data owners that are at the right level, you’ll miss out on these benefits.
Data ownership challenges
Data ownership is not all plain sailing. The challenge of appointing someone senior into this role is, by the very nature of their seniority, they have less time to invest than others in the firm. To overcome this you need to support the role with data stewards that are empowered to make run-of-the-mill decisions.
When working with other data owners there can be conflicts. For example, a data owner is unlikely to have the depth of operational knowledge to know the full details of data produced upstream of their domain. Again, the fix is to support the role with data stewards and data custodians that can fill in these knowledge gaps and handle these issues independently.
In some firms, the culture is very “top-down”. Management doesn’t want to “get their hands dirty” by working with data. If you have management that is disengaged then you need to focus on building the business case and securing executive sponsorship more than most.
When a firm lacks a data-driven culture, it can also be hard to persuade the right people to play the right role. You’ll see disinterest in data issues and, as a result, you’ll get lower engagement. Focus on the dollar value of your work to highlight the need for senior sponsorship.
Chapter 5: Who else helps the Data Steward?
From this article we may have given the impression that your data steward is an internal superhero, battling all data issues on their own. Nothing could be further from the truth. In reality, the data steward is more like Iron Man or Captain America in the Avengers. They assemble a team and motivate the other superheroes to battle your bad data. Let’s take a look at who else is on the team:
Chief Data Officer and Data Stewardship
Not all firms have a CDO. They are increasingly commonplace as organisations look to maximise the value they get from data. The Chief Data Officer is the most senior data role in your firm. They act as the executive oversight of data governance and drive data strategy in the C-suite.
CDO’s act to oversee the data governance and data management program in the firm. As a result, they are ultimately dependent on and responsible for the success of our data stewards. The CDO will also own the data strategy, hence the work of our data stewards should align closely with the objectives of the Chief Data Officer.
The CDO is like Nick Fury in our Avengers analogy, pulling the team together to battle the bad data.
Data Steward vs Data Custodian
Sometimes these terms are used interchangeably. At Cognopia we make the distinction that a data custodian is more of an I.T. role, whereas the data steward is more of a business role. In practice, this means the data custodian would usually be able to do the work to manage the data, whereas the data steward should direct the work and determine what needs to be done. consider this example:
A customer data steward might determine that we must have First Name, Last Name and email address fields for a customer record to be valid. Furthermore, the email address must be in the correct format, and active (there are tools that can check this automatically). They document these rules and monitor compliance against them.
In order to implement the rule, our customer data steward calls on a data custodian. The data custodian is able to change the validation rules in the CRM system to prevent staff from saving a record that lacks a first name, for example.
Data custodians are responsible for the safe custody, transport, storage of the data and implementation of business rules. Simply put, data stewards are responsible for what is stored in a data field, while data custodians are responsible for the technical environment and database structurewikipedia
Data Stewardship vs Data Management
Data stewardship is a role that sits in the data governance team. The data management team are there to execute the rules passed over by the data governance team. Data custodians belong in the data management team. They are equally important to the success of our project.
Other data management roles that are important to the success of the data stewardship program include data modellers, data architects, and data quality professionals.
The data architects are the technical leads that design how systems interoperate. They deal with the plumbing – how data moves from one system to another. They also help decide where data is stored – picking data warehouses, data lakes or other choices and other technical design elements that make our lives easier.
Changes to the data architecture may be required to support the goals of our data stewards. The data steward needs to outline business requirements and work with the data architect to make these a reality. The data steward will document what the data environment needs to be capable of, the data architect will translate that into the right technical solution for your firm.
Data modellers are able to translate the “real world” relationships and entities your business has into tables and columns in your systems. They design the database structures needed to store the data about your business transactions. The data steward needs to provide the details about which relationships your business has, which entities you engage with, and help the data modeller understand the business reality they are modelling.
Here’s a quick summary of who’s who:
How Human Resources can help Data Stewards
One area we see a huge opportunity for improvement in is how data stewards work with their HR team to get results. In our 2021 survey, we found that 19% of respondents thought HR was “not relevant” to the success of their data governance program. They are missing an opportunity. The HR team should help:
- Identify and promote the right people into data governance and stewardship roles
- Allocate training budgets to develop critical data skills
- Adjust and update KPIs to incentivise better data behaviours, and;
- Track and monitor any resistance to changing these data behaviour
Hopefully, this guide has helped you understand the need for data stewardship. If you are interested in adopting data stewardship in your firm, we hope it has provided enough information. Should you want to discuss any topic raised here, please get in touch using the form below: