Why retailers can’t agree on a fraud strategy

By Matt DeLauro, President, GTM SEON

Retailers have spent years investing in artificial intelligence (AI) to fight fraud, yet many still struggle to articulate a coherent fraud strategy. While machine learning now anchors most fraud prevention programs, priorities remain fragmented across teams, tools and incentives. The result is a familiar contradiction: more intelligence in the stack, less clarity in decision-making.

That gap between intelligence and choice is widening. Nearly 68% of retailers expect to deploy agentic AI within the next 24 months, even as fraud budgets and headcount continue to grow. AI has become the baseline, but simplification has not delivered the alignment companies need, and with costs rising faster than confidence in outcomes, leaders are left with powerful technology and no shared definition of success.

How competing incentives undermine fraud strategy
Fraud decisions can quickly become a negotiation rather than a strategy when organizations lack a shared reference point for acceptable risk. When teams are evaluated on conflicting goals, such as optimizing loss reduction, protecting conversion rates, guarding profit margins or focusing on eliminating friction, each metric makes sense in isolation. However, our recent AI Reality Check: 2026 Fraud & AML Leaders Report — based on a global survey of over 1,000 fraud and anti‑money laundering (AML) leaders — shows that only 47% of retailers operate fully integrated workflows. That misalignment creates an operational tax in the form of slower investigations, higher false positives and duplicated effort. The organization pays for it twice: once in technology spend, and again in manual work.

This tension becomes most visible during high-pressure flash sales. Growth leaders push for looser thresholds to protect revenue, while fraud teams respond with manual overrides. Rules change mid-campaign, friction increases for legitimate customers and no one can clearly explain which tradeoff the business chose.

Why tool sprawl keeps retail fraud disconnected
Many retailers now manage fraud through a growing collection of tools rather than a cohesive system. More than 80% of merchants struggle to use data and technology effectively in their AI tools, often responding by adding new products instead of fixing their core architecture, turning theoretical flexibility into practical complexity.

Because only a minority of organizations run fully integrated workflows, many retailers rely on separate tools for checkout, accounts, loyalty and buy now, pay later (BNPL). Each system evaluates risk through a narrow lens and generates conflicting signals that teams must reconcile manually. Consequently, decision speed slows, operational loads increase and consistency becomes harder to maintain at scale.

How data silos fragment defense systems
AI now sits at the core of most fraud stacks, but governance has lagged behind adoption. Our latest report shows that 98% of organizations use AI in fraud and AML, and 95% express confidence in its effectiveness. Yet, explainability and human accountability remain top concerns. When executives ask why a legitimate customer was declined, many fraud leaders still struggle to provide a clear business outcome.

That challenge is compounded by fragmented customer data. Creating a single, trusted view of identity and transaction activity remains one of the hardest problems in modern defense. According to our research, 80% of leaders say achieving unified visibility is challenging, with more than 40% rating it as extremely difficult. Shared dashboards can create the illusion of alignment, but when decision logic remains siloed by channel, outcomes stay inconsistent.

What it looks like when fraud strategy aligns
Agreement on fraud strategy no longer means choosing the right tools or deploying more AI. The real shift leaders need to make is architectural. Retailers must move from disconnected tools to a command center mindset, transforming reactive alerts into systems that consistently translate risk signals into defensible decisions.

For retailers, alignment becomes tangible when a shared risk appetite is understood across all departments, rather than renegotiated during every volume surge. Identity and transaction intelligence must flow through a unified backbone rather than being reprocessed by each tool. Ultimately, AI agents should operate as governed copilots, not black boxes that increase headcount because no one trusts their decisions.

Turning alignment into action
Retail leaders do not need another abstract call for transformation. They need a practical way to reset how decisions get made. A strong starting point is a cross-functional fraud posture workshop that brings fraud, payments, growth, finance and customer experience around a single scorecard. By viewing loss, revenue and customer impact together, teams can agree on explicit trade-offs.

From there, focus should shift to architecture rather than feature checklists. Tooling needs to rationalize around a single identity spine, treating integration quality and time-to-value as strategic levers. Retailers that lead, invest in AI literacy, data integration and cross-departmental collaboration so that people, processes and models finally move in the same direction.

Favorite

The post Why retailers can’t agree on a fraud strategy appeared first on Digital Commerce 360.

from Digital Commerce 360 https://ift.tt/Y2dZt61