While enterprises are ramping up investment in AI and automation, challenges around trust and data quality continue to slow adoption, according to new research by Alteryx.
AI adoption is growing fast, according to Andy MacMillan, CEO of Alteryx:
“Our research shows that compared to a year ago, two-thirds of business and IT leaders are using AI more in their roles. We’re also seeing AI move closer to individual departments.”
Around 48 percent of leaders plan to increase their spending on AI infrastructure and tools. AI platforms are expected to make up a larger portion of data stacks over the next few years, rising from 33 percent in 2024 to 51 percent in three years.
But there is a growing disconnect between AI ambition and real-world impact, the research found. Despite heavy investment, most enterprises are yet to move beyond pilot programs, as they are held back by low trust in AI outputs, poor data quality and legacy technology that can’t support scale.
The result is that fewer than one in four AI pilots successfully move into full production.
Trust in AI systems and outputs remains one of the major barriers to adoption. Nearly half of respondents to Alteryx’s survey said they trust AI to automate repetitive tasks, draft content and monitor systems, but there is a notable gap in confidence for high-impact uses. Only 28 percent trust AI to support decision-making, and just 27 percent trust it to facilitate forecasting or planning.
To help overcome concerns about outputs, data quality is key. Around 49 percent of leaders cited high-quality, accessible and well-governed data as the top factor for agentic AI to achieve its full potential. As MacMillan pointed out:
“The most advanced organizations are doubling down on improving data quality and integrating AI across their operations.”
The findings highlight a core problem behind many stalled AI projects, Alteryx noted. Trust erodes when systems are rolled out without the business rules and context needed to deliver reliable, explainable answers.
In many cases, companies are placing GenAI straight on top of unrefined data. The outcome is familiar, fabricated responses, uneven results, and answers that shift from one prompt to the next, making it hard for leaders to rely on AI in day-to-day decision-making.
AI Growth Depends on Data Orchestration and Trust
To move forward, organizations need to strengthen the basics that can make AI dependable at scale. That means well-governed data, clear performance measures and workflows that pair GenAI models with predictable, rules-based logic, while giving business teams the flexibility to update those systems as priorities change. Some leaders recognize that, with 28 percent of respondents planning to prioritize data governance improvements.
That will become more important as ownership for AI workflows shifts, with business and IT leaders expecting responsibility to move away from centralized teams, MacMillan noted:
“Over the next three years, leaders expect responsibility for AI workflows to shift to specific lines of business, rising from 22% today to 33% by 2028.”
Alteryx’s report aligns with recent research from KPMG, Deloitte and PwC pointing to the importance of good data practices and orchestration for AI to deliver real business value.
Many organizations still struggle with fragmented data and disconnected systems that make it hard to scale AI and deliver measurable business results, especially in customer experience strategies. KPMG found that 66 percent of B2B CX leaders cite “data access, quality and management” as their biggest obstacle.
That reflects the idea that simply plugging generative models into raw data isn’t enough. Without robust data integration, governance and coordinated workflows, AI outputs can be inconsistent and untrustworthy, just as stalled CX projects reveal.
Strengthening the technical and organizational foundations, based on governed data, clear metrics and combined AI-plus-rules workflows, supports reliable decision making and the kind of connected customer experiences the research shows are necessary for measurable performance improvements.
