Data governance is a powerful concept that is increasingly gaining acceptance in the corporate world. Once the data discovery, change management and technology implementations have taken place, however, somebody has to do the hard work of keeping data in line with the strategy.
Data stewards are being widely appointed to carry out this task. Yet three years ago, this is not a job title that would have appeared in any corporate directory. So what does it mean to be handed the task of maintaining the daily quality and alignment of specific pieces of data? Is it a job for the future or a job from hell?
Kelly O’Dwyer at 3 has been at the centre of growing her company’s data stewardship culture. “My current role is as a data quality analyst in the department working on the new data warehouse in the business intelligence arena. As part of that team, my function is to encourage people to look at the quality of their data and to maintain it,” she says.
A major data integration project of this sort begins to throw up a lot of questions about the ongoing maintenance of information. O’Dwyer says the data directory created for the BI project was an important trigger. “That set us thinking about stewardship. In any growing business, you find you have issues with data ownership,” she says.
In order to avoid the classic problems of “garbage in, garbage out”, her team developed a plan for how business units would own pieces of data and take care of them. This has led to the creation of 35 data stewards across the organisation. Typically, these are individuals with knowledge of a specific area of the business as well as a technical understanding of data.
“We tend to target people at management level. They are high enough in the company to be effective, but not so senior that they are only focused on the bigger picture,” say O’Dwyer.
This is one of the common aspects of data stewardship – it is only rarely a standalone job. Most stewards take on the ownership of a data item alongside their main role. “Typically, they are people who appreciate the detail, and are close enough to the data to understand the finer points, but also senior enough to own it,” she says.
Despite the focus currently falling on data in the business world, data management in its own right is unlikely to gain top level support. In the context of a significant value-adding project, like business intelligence, it will. O’Dwyer says 3 has won senior-level buy-in for data stewardship because of how the data will eventually get used by lines of business.
Reporting on how well this is being enacted is tapered to the requirements of each level of management. O’Dwyer has specific KPIs around data quality, while senior managers have a dashboard relating to the business case, project progress and the like. “That makes it very transparent. With an organisation like ours that’s grown so quickly, it is easy to get into difficulties about who owns data, whose source system it came from. We have made that easier to identify and more visible to the business community,” she says.
In the past, cleaning and maintaining data has often been done in the context of a project with another purpose, such as an IT-driven migration from one platform to another. Data has not been seen as a strategic asset in its own right requiring specific attention and investment.
At 3, the business intelligence project has demonstrated that there is a value to adopting data governance and recruiting stewards. Elsewhere, business transformation programmes may be the driver of change.
PPL is the not-for-profit music licencing organisation which receives revenues from licence holders, like radio stations, and distributes them to rights holders. With 39,500 performers and 3,400 record company members, this involves matching over 30 million items of data each year.
The organisation has recently migrated from bespoke applications written in-house to a DataFlux solution to carry out this matching of airplay information to recorded music data. “The key application is to map back to the sources of recording data any income and payments that should be made,” says Frank Jaschinski, director of IT at PPL. “This is an important project because it may improve data quality in the whole industry in the long term.”
Currently, the organisation has to capture, standardise and deduplicate recording data as it is received from multiple sources. With the new solution, it will become easier to identify where information originated and then address any issues it may have.
The range of revenue sources is accelerating with the growth of Internet radio and digital downloads, for example, even though just four music labels continue to dominate output. Small record labels (and licence holders) are less likely to have sophisticated data management skills in-house.
“We first glimpsed that when we started the UK Download Chart which generates large volumes of data on a very long tail of artists,” says Jaschinski. Data quality is critical to ensure artists get paid. “Our business is all focused on giving money back to artists, big and small.”
PPL’s processes may be capable of being propagated outwards into those data generators pushing data stewardship upstream. The organisation is currently working with Deloitte to look at how data should be organised and whether data stewards are needed. There is also a global initiative working on standardisation of data definitions for recordings.
Distributing knowledge about data governance procedures and data stewardship is also one of the bonuses from a data integration project carried out at Brent Council. This has involved bringing together 1.5 million records held on local residents across multiple systems operated by each of the council’s service providers into a central data hub working with Initiate Systems.
“A lot of data was being captured four or five times and we had one service ringing another to find information,” says Tony Ellis, chief information officer at Brent Council. As an example, when the council lets a property, the property trust has to prove the identity of the tenant, which involved calling electoral services. Changes in data it was capturing would not be passed across to other council systems.
The solution to this was creating a Citizen Hub which would capture and maintain data once while distributing it to service providers. Central to this was the creation of dedicated data stewards in each department who own specific items of data. They have a joint responsibility for auditing the quality of data on their systems, selecting and verifying what is passed into the hub, while championing data in their specialist area.
“Their ‘day job’ is very focused on what their department is doing. We wanted to open that up and look at sources of problems with information and change the way we were doing things to deliver a better quality of service. They were very happy to do it because they could see it would help their day job,” says Ellis.
From this internal focus, Brent Council has created Data Connect, a source of best practice, business case models and information exchange for other councils undertaking similar work. “Other people are struggling in this area so we have got together to share our knowledge and provide sensible information on data matters,” he says.
One of the key learnings from the council’s own project is that, “people don’t understand the value of data”. That is why data stewards are a critical interface to explain the benefits of doing the right thing and ensure conformity to data standards. “IT is the easy bit. The key is to ensure you have the right people,” says Ellis.
Service improvements are one of the reasons why organisations start to open up their data management processes to look for issues. Another may be a drive to cut costs by stripping out duplicated effort. In either case, the data steward is closely aligned to the original business case and how close the original figures are to being achieved.
But what if the organisation’s entire business is data itself? Transactional data co-operative Abacus has to assure its members that their data is protected and used in the appropriate way, as well as ensuring it meets Abacus’s own standards for the Alliance Database.
That job falls to data quality manager David Westland. “We are working with multiple clients with different systems and levels of quality. The people who are directly involved with their databases range from one-person companies using Excel as well as doing everything else up to enterprise-level data warehouses,” he says.
Abacus has three teams covering incoming data, data order fulfilment and database mangement. Each team owns the data involved in their part of the process and manage its quality. For Westland, much of the work involves helping data users to understand quality issues.
“When they supply us with their home file and we supply them back with prospect data, if they find 1 per cent duplication with their customer base, that is within the bounds of reason, because of time lags in the merge-purge. If it goes up to 2 or 4 per cent, that is an issue that needs investigation,” he says.
One reason may be that the client did not provide its full customer data set, so the duplication is unavoidable. Identifying the cause is one thing, explaining it to a client quite another. Equally, it is vital that members understand the impact of poor data quality on everybody. “They need to be aware that it is not just their own data that is affected.
When they load it onto the Alliance Database, it will affect all of them. My role involves educating them,” he says.
Communication skills are therefore important for a data steward to do their job successfully. Data governance is always a team activity that requires contact with multiple functions across the organisation. Being able to persuade stakeholders to change their way of working may be essential to a project’s success.
At Experian QAS, data governance has been identified as a key strategic programme, alongside data security, compliance and business processes. A team has been convened to identify the issues and challenges across the business, with Esther Chambers working as its strategic project manager.
“It is really more of a programme than a project,” she says. “We do have milestones in the early stages, but it is open-ended.” One of the things that has emerged is that, “there is a greater level of understanding one level down from the data governance heading, at the level of data quality. Buying into data governance and the extra elements it involves is a stage further on.”
Chambers believes that data stewards will become necessary because of the critical value of data within the business. Their role will be to assure the correct usage and management of data as it passes in and out of the organisation. Eventually, QAS hopes to reach level one in its maturity model of contact data management so that it can then propagate that knowledge and experience among its clients.
What is striking about data stewards is that their role has very wide boundaries. While the specific pieces of data they own may sit in a single cell, it may be sourced, used or reported on right across a business and even beyond. Owning the truth is not for everyone, but it should give some a real buzz.