Why A Data Strategy Is Needed And Its Key Ingredients

All organisations start out with high hopes for their data-driven efforts, but some will struggle to convert that enthusiasm into value and tangible results, writes Data Army director, Michael Ogilvie.

There’s no shortage of organisations today looking to convert their data into insights that can help them unlock unrealised value, identify growth opportunities and make strategically smarter business decisions.

Becoming data-driven has been a top-three goal of Australian organisations for a number of years, yet our experience is that around half of all data analytics initiatives will run into difficulties, because they lack clarity and direction.

To minimise such issues, clarity and direction should be laid out in a formal data strategy. A data strategy is imperative to every business embarking on data-driven change. This is especially the case for those looking to delve into more advanced forms of data-driven decision-making, such as the use of artificial intelligence (AI).

The data strategy should provide the vision and goals for strategic data use, describe current and future enablers, risks and other operational parameters – such as legal or regulatory guidance – that need to be factored into all actions undertaken in the data area.

How ineffective strategies are born

We estimate today that half of Australian organisations do not have an effective data strategy in place. While organisations often have a document titled ‘data strategy’, it is often too loose and open to interpretation to be considered effective.

There are a number of reasons why a data strategy may not be effective in giving the organisation appropriate guidance on how to progress with data-driven efforts and programs of work.

Sometimes ineffective data strategies are set by executives without considering for what is executable or achievable by a central data team or by teams in various parts of the organisation. Other times, the preparation of the strategy may be outsourced or put into the hands of a data or divisional team to define and write, with little oversight, involvement from business leaders.

Data strategies can be created at a project and business level but should have alignment with overall business priorities. An effective data strategy provides a high-level view, keeping priorities and efforts aligned and ensuring that every task contributes to the broader business goal, guiding teams through challenges.

When leaders aren’t setting a data strategy or direction themselves, and hand off responsibility to teams at the project or business level to determine the finer details without input, the result is often a document that comprises mostly short-term, immediate action items. This lack of alignment to longer-term business strategy is a mistake, given that research shows that aligning data analytics to business strategy produces more effective outcomes than more tactically-focused data efforts.

Aside from missing the perspective and long-term vision of leadership, data strategies written in a decentralised manner may also suffer the effects of assembly by consensus. With everyone being consulted in the course of preparation, everyone has an opinion, the information-gathering phase increases exponentially, and so does the job of consolidating all of those opinions into a clear path forward.

A democratised approach to writing a data strategy risks producing an incohesive document that is not easily followed or explained. There may be multiple owners for different parts of the strategy, with unclear or undocumented lines of demarcation. Specific data projects that spill from the strategy are hard to implement and fail to hit the heights they were expected to – and everyone is left looking for answers as to where things started to go wrong.

The ingredients of an effective data strategy

An effective data strategy addresses a number of key areas:

1. Vision and Goals: Business leaders know what they want to achieve in becoming data-driven, but this needs to be clearly articulated and documented so that the teams executing the vision know exactly what is intended and what their work needs to deliver. The strategy, above all, should be a referenceable framework for the entire organisation.
2. Data Governance: The adage, ‘garbage in, garbage out’ continues to ring true. To become data-driven, an organisation needs good data at its fingertips. That means setting rules and guidelines that ensure data is clean, structured, collected according to standard definitions and formats, has assigned owners/maintainers, and lineage can be established to ensure its veracity and accuracy.
3. Data Architecture and Integration: When data repositories are disparate, becoming data-driven is difficult. The data strategy should describe the approved systems and methods for obtaining data for different use cases, as well as any central architectural elements – such as data warehouses, data lakes, or data marts – that may improve managed access to data.
4. Analytics and Business Intelligence: In addition to backend data architecture, front-end tooling must also be clearly defined. These are the platforms that will be used internally to take data and turn it into insights or visualisations, or to develop and train AI and machine learning (ML) models.
5. Data Literacy and Culture: People working with data need to understand it contextually in order to use it effectively. That means building data skills among domain and divisional experts to do more with their data; and among executives to drill down into the data delivered to them in dashboards and other visualisation vehicles. The strategy should also lay out cultural expectations – that all decisions above a certain threshold require a basis in data, for example.
6. Technology and Tools: Describing the current state, and processes or justifications needed to experiment with or introduce new tooling into the environment.
7. Talent and Leadership: The strategy should articulate how data engineering and data science staff and skill levels are to be maintained, and delivery expectations of certain staff levels. It should also be clear where responsibility for data use ultimately rests.
8. Risk Appetite: Data can be used for a range of activities, from basic reporting to personalisation and prediction of what a buyer might want. The line on acceptable and unacceptable use needs to be clear, and there should be a documented process for consideration of riskier use cases. The line should not only be apparent when it’s crossed.
9. Legislation: A range of laws govern data use, covering regulated sectors such as finance, health or critical infrastructure operators, or usage limitations around privacy. These need to be front-and-centre to ensure data use cases are allowable, before any work takes place.

The nine ingredients for success should steer the data strategy to successful internal and external outcomes and ensure an enterprise is best placed to navigate any dynamics in economic uncertainty.

This article was originally published on 26 June 2024 in Consultancy.com.au.