AI value proving elusive for many Australian firms

Australian organisations report widespread difficulty in proving the business value of AI investment, according to an APAC survey by analyst firm Ecosystm that focused on data readiness and AI adoption.

The research, commissioned by Snowflake, drew on responses from more than 700 business and IT leaders across the Asia Pacific and Japan region. It examined how organisations apply agentic and generative AI, where they evaluate use cases, and what blocks progress from pilot projects to broader deployment.

In Australia, 81 per cent of respondents said they had difficulty demonstrating the business value or return on investment from AI. The figure ranked second among surveyed markets, behind Korea at 83 per cent. In New Zealand, 70 per cent reported the same challenge, which the research cited as the lowest among surveyed countries.

Customer use cases

The research found Australian organisations most often evaluated AI for customer-facing interactions. Some 67 per cent said they evaluated use cases for interacting with customers across channels. Another 53 per cent cited improving chatbot responses. A further 65 per cent cited generating marketing content.

New Zealand respondents also prioritised customer interactions across channels, cited by 68 per cent. Some 60 per cent reported evaluating AI for generating marketing content. Another 56 per cent cited improving search and summarisation of data and reports.

The research described a gap between early experimentation and consistent delivery of business results at scale. It linked that gap more to data and technology foundations than to issues with AI models.

"While there is widespread AI enthusiasm across the region - particularly in Australia and New Zealand - business leaders are now expecting to see the technology deliver real business value," said Theo Hourmouzis, Senior Vice President, Australia, New Zealand and ASEAN, Snowflake.

"To unlock this value, organisations must deeply integrate AI into their business strategy, rather than pursuing it as merely an experiment. This requires starting with clear, measurable use cases tied to real business needs rather than deploying AI for AI's sake."

Data roadblocks

Across Australia and New Zealand, respondents cited data accessibility as the top data challenge. Some 56 per cent of respondents reported this issue. Data quality followed at 52 per cent. Data security and data observability both came in at 49 per cent.

The research also found low levels of AI integration into overall business strategy. Only 19 per cent of Australian companies said they had fully integrated AI into their business strategy. In New Zealand, 24 per cent reported full integration.

The report described fragmented data and underprepared technology foundations as a common constraint on AI adoption. It also pointed to the difficulty of working with unstructured data at scale across organisations.

Across all surveyed countries, 38 per cent of organisations said they had invested in technologies for analysing unstructured data, according to the research.

Partners and platforms

The research also pointed to a strong focus on external support for AI programmes. It found 85 per cent of Australian organisations engaged, or planned to engage, technology partners for strategic, technology and data needs linked to AI projects. In New Zealand, 76 per cent reported the same approach.

Hourmouzis linked AI outcomes to broader implementation and governance work across organisations and their suppliers.

"AI isn't just plug and play; partners will be vital to helping bridge capability gaps, accelerate deployments, establish governance frameworks, enabling organisations to stay ahead of the next wave of disruption," said Hourmouzis.

The research also set out a set of practices for measuring value over time. It argued that pilot projects could demonstrate feasibility but could still miss full costs and benefits when organisations scale AI into operations. It pointed to the need to track infrastructure and model maintenance, as well as governance and compliance considerations.

The research also cited the impact of fragmented tools across the AI lifecycle. It argued that disconnected systems across data preparation, model development, deployment and monitoring can obscure performance and costs. It described integrated measurement of technical and business metrics as a way organisations could approach visibility and accountability in AI programmes.

The report said organisations in Australia and New Zealand showed signs of shifting towards more strategic approaches to AI programmes, with partner engagement and investment decisions linked to data foundations remaining a key focus for the next phase of deployment.