Developing an effective data strategy: looking through a golden window pane

When I started in my first IT role way back in the spring of 1981, the UK Number 1 chart single was 'This Ole House' by Shakin' Stevens, an Elvis clone from my home city of Cardiff In Wales. One line in the song has always stuck with me over the years: "He seeks a new tomorrow through a golden window pane".

This lyric seems highly relevant when talking about data in general and BI in particular. Much of the investment in data and platforms made by organisations both large and small has been done with a primary aim of crafting a golden window pane, a window into their own ole house, the outside world, and the future. But for many, this investment has not consistently delivered the insight and clarity they hoped for. Instead they have ended up with a dirty and obscured view, so they frequently clean the pane, only for the grime to reappear time and again.

A key reason for this is that too much BI investment has been made to provide tactical solutions to resolve pressing needs and opportunities. In other words, it tries to develop a window of insight without the context of an overall architectural blueprint of the house as a whole, starting from the foundations up. In the last year I have worked with a number of companies in both the UK and mainland Europe who have done this. They have invested heavily in BI but without exception they have been disappointed with the outcome. Often IT blames the business for poorly defined requirements and shifting needs; the business blames IT for not delivering what it promised. The end result is a proliferation of data warehouses and data lakes that are not trusted (and often not referenced) by their users. Data quality is poor, metadata non-existent, costly BI tools are under-utilised, and expensive Analytics specialists and Data Scientists spend 80% of their time trying to find the data they need and cleaning it up to make it fit for purpose before they can exercise their expertise to leverage any insight.

When these failures become more evident, and failure costs rise, the need to re-architect data and the IT estate from the foundations up becomes the only viable long term solution. But how do you bring this about? The best starting point is to develop and implement a coherent data strategy, which provides a clear and shared vision for what the organisation is trying to achieve by investing in data and the platforms and tools that deliver it, and lays out a plan for delivering it.

Traditionally, the starting point for building a data strategy was to start with the business strategy. The business strategy lays out the organisation’s current and future goals and aspirations, for example in relation to its markets, customers, products & services, revenues and so on. In this scenario, the data strategy specifies how these requirements translate to data needs and insights, asking and answering questions such as: what data do we have today to support our goals? How fit for purpose is this data? What do we need to do to enhance it? What new data do we require today and for the future?

In this model the data strategy is subordinate to the business strategy. Although still relevant in some circumstances, this model is rapidly changing. As many organisations aspire to become more digital and data driven, the data strategy takes on a renewed importance. Here business and data strategies are mutually reinforcing. What the business can achieve in terms of new digital products & services, and ways of doing business, is to a large extent, driven by its data strategy. In these cases, business and data strategies become equal partners, with each informing the other as to what can be realistically achieved, and what cannot.

So what makes an effective and deliverable data strategy? Having helped a variety organisations produce them, these are the five ‘must haves’:

• Clearly defined motivation: Create a clear alignment between business goals and the data strategy. There should be a direct line of sight between every action listed in a data strategy (and remember any strategy should be a plan with actions, timescales and owners) and the business drivers and goals of the organisation. This ensures that as the strategy is executed, the benefits of what it delivers can be directly linked to one or more business goals or objectives. This also ensures it gets the backing of senior business people, critical for success.

• Business leadership: One way to ensure that a data strategy fails is to put IT in charge of it. Despite its best intentions, IT is not suited to lead as only the business can define the motivation and data priorities. In any case, putting IT in charge reinforces the long held misperception that data is IT’s problem. Undoubtedly IT has a key role to play, but in a supporting role to the business. Critical is that business accountability for data is recognised and formalised. In other words, that well-defined data governance is in place and business recognises it is at the vanguard of change.

• Focus on key data: Strategies should be focused on the data that really matters to the business if it is to achieve its goals. Put simply, if the primary business aspiration is to grow revenues by selling more to its current high value customers, making customer data fit for purpose (e.g. knowing who your high value customers are) is self-evidently a top priority. Lack of focus and priority inevitably results in a scattergun approach.

• Baseline your starting point: If it is to meet expectations, a data strategy must be realistic. This can only be guaranteed if you have a clear understanding of the organisation’s current levels of data management maturity. If it’s immature, with ineffective or no data governance, poor data tooling, universal data quality problems and so on, there is no point in aiming for the sky when you first need to lift one foot off the ground. You need to map out the journey and the destination, but know where your starting point and identify achievable milestones along the road.

• KPIs and measures: To demonstrate a data strategy’s progress and business value, you must set measurable targets and implement systematic processes to measure attainment against them. Ideally these should relate directly to the business goals and objectives highlighted above. The importance of measurable targets is often stated and has become a cliché, but is all too often neglected. Too many data strategies are little more than wish lists and statements of vague aspirations, and as a result never deliver anything of real substance. To monitor KPIs and measures BI has a key role to play in helping produce data dashboards and other artefacts.  

• Cultural adoption: Last, and never least, people ultimately determine if change happens or not. Becoming a data driven organisation requires that all people accept and play an active part in making it happen, and receive the training and upskilling required to enable that. Any data strategy must reach out to all, and be understandable to all.
 
In summary, developing an effective data strategy is both a technical and cultural challenge. Too many data strategies focus too heavily on the IT and BI platforms and tools required to deliver it. This is a necessary but not sufficient precondition for data strategy success. Getting people to embrace it is more than half the battle. To find the Golden Window Pane to a better tomorrow, put people at the forefront of strategic change and data strategies. After all, they live in this ‘ole house’ and will embrace a new tomorrow if they feel they have played a part in envisioning and building it.  

Nigel Turner, Principal Information Management Consultant EMEA at Global Data Strategy, will be presenting at the Data Warehousing & BI Summit on March 27 & 28 in Utrecht.