Why should data be treated as an asset?
Data is often promoted as an institutional asset. Such statements do not pass even cursory examination when comparing the rigour, accountability and governance of other assets such as finance, staff and estates.
These assets are well understood by most in the institution and accountable to a responsible officer. Each has defined and persistent resources to maintain them, professional qualifications to underpin them and roles/responsibilities to ensure they are deployed in support of the organisational objectives – both operationally and strategically.
The data ‘asset’ clearly has few of these attributes. Data Governance is the framework to transform data from operationally silo’d to organisationally aligned. Data has no value if it does not visibly support what the institution is trying to achieve. The collection of more data or driving up the quality of non-aligned data is counterproductive and expensive, and yet this often represents the activities undertaken at institutions today.
When HEDIIP conducted its review of sectors’ data capability (full review here), a key finding was that weak governance was hampering institutions ability to harness the power of data.
What value will implementing data governance bring?
As with any asset, realising the value will take time, effort and focus. By doing so, the asset value will increase visibly in at least four areas:
1.Cost Reduction / Cost avoidance
The cost of managing traditional assets is well understood. A new building for example will demand careful budgeting both for construction and maintenance. That building will require support from trained, dedicated individuals whose costs are again known and justified. Data is not like this, it is managed in individual departments and the scope of the cost is unclear. The IT group are often considered the main ‘cost’ but this ignores the daily data management within organisational hierarchy, and the cost of reformatting data for different uses.
A data governance framework both reduces the cost of data management through the development and adoption of shared definitions, an understanding of what the data is used for by the whole institution, the minimum level of quality to meet those needs and the right source of the data to start with. From other industries, the savings here should be significant before considering the intangible benefits of well managed data.
Cost avoidance is realised through the understanding of the impact of change. There are many horror stories of wide scale changes (such as moving from faculties to schools) where data is rarely considered. This leads to extreme and sub-optimal outcomes that both attract additional unbudgeted cost and myriad dis-benefits.
2. Student experience
While data is not going to transform the student experience on its own, it will visibly support both the accuracy and presentation to personalise it. A number of institutions already ‘wrap the data around the student’ to provide that personalised experience from timetabling to funding and everything in between.
There has been a focus on the ‘student app’ which, will laudable, has often been built on the unstable foundations of poor quality data. A data governance framework understands what the data is needed for, when it is supplied and how it is to be presented. The bar for data when considering students is extremely high with the commercial applications in daily use. The aspiration for the data governance outputs is to be at least as good. This is not a separate, expensive to maintain data stream- it is part of a well managed data landscape.
3.Harmonisation and rationalisation
These were two of the key outputs of the HEDIIP programme and form the cornerstones of HESA’s Data Futures programme. The benefits of defining and adopting shared definitions are legion; increases in effective use of data, comparability both internally and for shared data returned to the funders, regulators and other bodies, simpler transformation into public metrics such as league tables, performance indicators and – latterly – the TEF.
Rationalisation follows harmonisation. The inventory of data collectors shows 93 organisations separately asking for data, most of which is already provided through the existing HESA return. Aside from harmonisation, the main issue is timing which will be resolved by the in-year collection data platform coming on line in 2019/20.
Data Governance at its heart is about delivering fair access to data for all. Key to this is a business glossary defining the terms and roles (such as student, course, assessment, etc.) in a manner they can be universally adopted. This is not a trivial process, but one that reaps huge rewards in terms of the efficient and effective management of data – both for internal operational and strategic purposes, but also to minimise the cost of participating in the sector returns process.
4.Transparency of process
Centrally managed data has a bad reputation in many institutions. The data from – for example – a planning function is often not considered as ‘good’ as the data an individual or department may prefer having developed it for their own purposes.
Clearly this is not an ideal state of affairs; it brings with it the arguments of whose data is right not a discussion of what the data means, it creates ‘cottage industries’ with the associated cost of lost opportunity, it devalues the value of centralised data and makes the use of that data for multiple purposes (e.g. external returns and internal decision making) extremely difficult.
Data Governance starts with a policy. That policy reflects the organisational needs. It enacts that policy through robust and visible processes on how to collect, store, use and protect the data. It is not the IT department or to planning unit – rather it represents the institutions in a similar manner to audit answering to the stated objectives not individual needs.
How do I get started with data governance?
This is a far wider question than can be answered here, but the key facets of a successful adoption of data governance as business as usual are:
– It is not a project. Data Governance must be a permanent and persistence capability. It will be embedded with existing processes, people and technology. It may need some kind of project structure to get it started, but it must not be considered something that is somehow done at a point in time and then ignored. This is the single biggest failure of data governance initiatives.
– It is not a technology solution. While technical tooling may be important, it is subservient to the requirements of the framework. Starting with a tool inevitably leads to disappointing outcomes.
– It is a people led change. A change in the cultural approach to data is required. In a world with data governance, things will be different. These changes should not be under estimated. A key component of a successful data governance initiative is a wide ranging communications plan to both show what success will look like and how each individual will contribute to that success.
– it requires staffing and visible senior level support. Embedding data governance for a medium to large scale institution will take up to two years. It will need permanent staff (for at least that time period) and senior support to both provide direction and the mandate to making things different. Data Governance is a hard slog at times because it’s dealing with difficult issues. Tenacity and a clear vision from the top of the organisation are vital to move it forward.
– It cannot be implemented overnight. Every data governance initiative is different. This is absolutely right as every institution is different. A framework is built using best practice in support of how an individual organisation works. The building blocks are the same, how they are placed will be specific to the need and priorities. Frameworks however should not be front and centre. Successful data governance does not launch a framework, it uses the framework to support activities aligned to organisational goals. This will take time and need to be carefully planned.
– It does not need to be perfect. There is no need to have a perfect data governance framework to get started. In fact, it’s counterproductive. The framework will respond to adoption and any rigid implementation does not have that flexibility. A clarity of vision, a defined set of initial activities, visible new roles and responsibilities and a organisation wide communication plan are the building blocks for success. None of these need to be perfect.
Data Governance is not easy. If it was, we would already be doing it. Successful adoption will deliver the benefits discussed in this paper. This is proven and repeatable from initiatives in other sectors. The rate and scope of change facing the HE sector makes this the perfect time to align the large, but hidden, costs of managing data with the objectives of the organisation.