Half of AI projects falter as enterprises struggle with data readiness

New research from Fivetran has found that close to half of enterprises are experiencing delayed, underperforming, or failed artificial intelligence projects due to insufficient data readiness.

The global survey, executed by Redpoint Content for Fivetran, uncovered that despite substantial investment and ambitious strategies centred around AI and data centralisation, poor data readiness is hampering project outcomes and causing financial losses for organisations.

Among the findings, 42 percent of enterprises reported that more than half of their AI projects have been delayed, underperformed, or failed as a direct result of data readiness issues. The research attributes these failures to challenges such as data that is not fully centralised, a lack of governance, and the absence of real-time availability for feeding AI models.

The study highlighted that while 57 percent of surveyed organisations believe their data centralisation strategy is highly effective, nearly the same proportion noted that over half their AI initiatives are not delivering on expectations. The primary obstacles cited include integration bottlenecks and the heavy maintenance burdens associated with managing data pipelines.

The effects of poor data readiness extend beyond operational setbacks. The report noted that 68 percent of organisations with less than half their data centralised have lost revenue as a consequence of unsuccessful or delayed AI projects. Additionally, 59 percent of enterprises identified regulatory compliance as their foremost challenge in managing data for AI deployment.

Business outcomes are being affected as well, with 38 percent of enterprises reporting increased operational costs stemming from failed AI projects. The research also found that reduced customer satisfaction and retention was the most common consequence when AI initiatives did not succeed.

The Fivetran survey emphasised the need for enterprises to modernise data infrastructure, particularly by adopting automated data integration tools. Automation and integration were cited as essential factors that can help reduce the complexity of data pipelines and allow engineering teams to focus on advancing business value through AI.

According to the report, 65 percent of organisations plan to prioritise investment in data integration tools as a means to enable successful AI initiatives. Nearly three-quarters of respondents manage or expect to manage more than 500 data sources, which underscores the growing need for scalable and automated solutions.

Despite efforts to centralise data, many enterprises are still finding it difficult to progress beyond initial AI pilot projects. The survey found that 67 percent of highly centralised enterprises continue to allocate over 80 percent of their data engineering resources to maintaining existing data pipelines. This leaves minimal capacity for AI innovation and development.

Lack of real-time data access was identified by 41 percent of organisations as a barrier to AI models' ability to provide timely insights. Additionally, 29 percent of respondents saw data silos as the main obstacle to AI project success.

The survey stated, "Until these challenges are addressed, organisations will continue to struggle with AI performance and fail to unlock the full value of their investments."

When examining industry sectors, the report found that healthcare and retail are at the forefront of AI readiness because of stronger strategies for automation and data integration. In contrast, finance and manufacturing sectors are less prepared due to limitations presented by legacy systems and integration issues.

There are notable regional differences regarding AI readiness. Asia-Pacific achieved the highest AI readiness score at 8.8 out of 10, followed by the US at 8.2. The UK recorded the lowest score at 6.0, attributed to less developed integration strategies and fragmented data infrastructure.

The survey canvassed the opinions of 401 data leaders and professionals from several regions, including the US, UK, Europe, the Middle East, Africa, and Asia-Pacific. Respondents represented enterprises in a range of industries such as technology, finance, healthcare, retail, and manufacturing, and included organisations ranging in size from 500 to more than 5,000 employees.