Originally posted on The Drum.
Many of the world’s billion-dollar businesses say they’re scaling agentic AI. But new research suggests many aren’t getting very far. Sean Muzzy, Acxiom’s president, explores how data and identity problems are preventing agentic innovation – and what it will take to resolve them.
If you only read the headlines, you’d be forgiven for thinking agentic AI is sweeping through many of our planet’s largest organizations. A hefty 40% of enterprises with an annual revenue above $1bn say they’re already scaling agents, according to McKinsey & Company.
But there’s a huge difference, and a huge amount of untapped ROI, between replicating a successful agentic pilot once and strategically implementing agentic decision-making throughout an enterprise.
New research from Forbes Insights and Acxiom underscores this nuance. The study, based on interviews with 300 business leaders, reveals efforts to scale agentic AI remain in their infancy for the majority (68%) of large US organizations. Just 5% describe their state as fully mature.
The research also sheds light on why organizations aren’t making swifter progress. Let’s dig into the issues, using agentic AI-assisted marketing as our case in point.
A connected view (that’s too disconnected)
Agentic reasoning, like human reasoning, is only as strong as the data behind it. When an agent only has a partial view of a customer, for instance, because customer data is fragmented across systems, it’s less likely to arrive at the right decision. And when agents don’t consistently arrive at the right decision, agentic AI rollouts stall.
Business leaders get this. Overwhelmingly. Ninety-nine percent believe a connected customer view is important for enabling AI-driven marketing. But their data and identity capabilities aren’t delivering that view or making it actionable.
Just 19% of business leaders say their identity framework fully supports agentic and AI-driven marketing at scale. Only 18% say they’re using data to influence automated marketing decisions effectively.
Simply put, the foundations marketing teams require to make agentic AI an everyday operational reality typically aren’t there. The same story often plays out in sales, finance and across the enterprise, meaning isolated productivity gains fail to translate into highly agile, agentic AI-enabled operations.
Understanding (and dismantling) the barriers
Without these data and identity foundations, multiple barriers stand in the way of real-time, AI-driven decisions. The three most frequently cited by business leaders are governance and compliance constraints, data quality issues and a lack of systems integration.
These barriers confound marketing’s attempts to prove and extend agentic AI’s value. Nothing prevents a connected customer view like siloed systems and untrusted data. And nothing puts the brakes on agentic scaling like concerns that, without the tools in place to respect individual preferences, communications may cause compliance issues.
Business leaders understand they need to dismantle these barriers before they can realize AI’s potential. When asked which single area they would prioritize to unlock growth, they put improvements in data infrastructure, data governance, identity and activation well ahead of scaling AI and automation.
Shedding new light on old data problems
In this way, businesses’ efforts to scale agentic AI are set to drag longstanding issues with data-driven decision-making into the light.
An overwhelming 92% of business leaders say data-driven decision-making can break down at their organization. The majority say silos still limit data’s impact, and that it’s hard to share data across platforms and teams. For many organizations, the result is a long delay between data-driven insight and data-driven action. A mere 10% say they’re able to act on data insights in real time. The majority (62%) say moving from insight to action takes them days or weeks.
As soon as you begin to scale agentic AI, these problems with data-driven decision-making become impossible to ignore. It’s time for every organization to revisit its data and identity strategy.
The data and identity strategy for AI decision-making
Most organizations’ data modernization initiatives are failing to meet the needs of agentic AI. The majority of business leaders (54%) admit that their modernization efforts aren’t improving their ability to support AI-driven and agentic decision-making. For another 19%, the improvements have been minor.
So, how can organizations strengthen their data and identity strategy, dismantle entrenched barriers to data-driven decision-making, and lay the foundations for scaling agentic AI?
The first step should be a step back; it’s important to get off the treadmill of ongoing maintenance and optimization, look at your data infrastructure, and consider what you actually need. Then start by creating a small, cross-functional team with clear goals. One immediate objective should be identifying where agentic decision-making is most desirable and valuable. After all, some decisions do require human oversight and don’t need to happen in real time.
The organizations that map their data and identity infrastructure, understand their future needs, and develop a prioritized plan of action will be the ones that not only scale agentic decision-making successfully but also maximize its business impact.