Why data governance needs an owner

This article was developed from an original interview by the Australian Institute of Company Directors.

Strong data governance is essential to building organisational resilience and adaptability.

As organisations adopt artificial intelligence (AI), they need to ensure their systems are resilient and responsive to change. They must properly manage data to obtain the highest benefit from their technology investments.

The digital resilience of an organisation depends on having technology, data and people who are flexible and agile enough to respond well to events such as a shift in the geopolitical environment or a new business model. That adaptability starts with the right technology choices and a clear plan for managing data.

Who owns data governance?

Too often, data governance is treated as an afterthought, and many organisations fail to define who is responsible for it. It shouldn’t automatically fall to the chief information or technology officer. In some organisations, a chief data officer will take the lead; while in others, the responsibility may sit with the chief operating officer, reflecting data’s central role across business operations.

The key is for the chief executive officer to acknowledge that someone must own data governance, and to identify and appoint the best person to do that in their organisation.

Embracing AI

Data governance is also fundamental to successful AI adoption. In BDO’s recent global survey of more than 1,000 C-suite executives, respondents identified messy and siloed data as one of the biggest barriers to resilience and AI return on investment (ROI).

Many organisations will use similar generative AI tools and large language models (LLMs) to their competitors. What will set them apart and give them a competitive advantage is the quality of the data they feed into those models.

As organisations look to adopt AI, decision-makers should educate themselves on AI types and applications, and ensure their organisation has a clear AI strategy to deliver value. AI should be viewed as an enabler to enhance existing processes. When deployed securely and tailored to business needs, it can create significant value across the organisation.

The challenge of unstructured data

LLMs rely heavily on unstructured data such as emails, reports and product information. It’s generally estimated that unstructured and semi-structured data make up around 80 per cent of an organisation’s information, yet it is often a forgotten part of the data ecosystem. While vast amounts of data exist across the business, individual users typically access only a fraction of it.

Once again, this is where the data governance owner becomes indispensable. To maximise AI’s potential, organisations must know where their unstructured data resides and how to harness it. Poor data hygiene will undermine AI performance and confidence in its outputs.

If organisations don’t value their unstructured data and just start dumping it into their LLM, they can’t be confident in the answers they get back from their AI. This also raises the issue of cyber security. Protecting AI systems is as important as safeguarding platforms like Salesforce or Oracle.

There’s an emerging need to test AI from a cyber security perspective, to prevent bad actors from poisoning AI models and rendering its output as inaccurate, skewed or biased. Like any technology, AI tools need robust governance and guardrails. With careful and considered planning, AI tools can have a hugely powerful enabling effect on business practices and performance.

How BDO can help

If your organisation is exploring AI or digital transformation, start by asking: Who owns data governance? Establishing clear accountability is the first step toward reducing cyber risk and achieving real business value.

Get in touch with our digital experts, who are aligned with the latest in AI, data governance, and digital transformation, and can help guide your organisation to make strong data governance a priority.