Retail has never been more excited about AI. From automating supply chains to creating dynamic product recommendations, AI is dramatically changing how brands engage customers, manage operations and drive revenue.
But beneath the promise of AI is the reality that most retailers aren’t fully ready to make the most of it. The problem isn’t a lack of ways to apply it to their business; it’s a shortage of usable, clean data. According to Amperity’s 2025 State of AI in Retail report, 90% of retail leaders say their organization is already using AI in some way, and 97% plan to increase investment over the next 12 months. However, only 11% report they’re actually prepared to scale those efforts across the enterprise.
This is a classic case of vision outpacing execution. Retailers know what they want to do with AI, but fragmented data keeps getting in the way.
Where AI Goes Wrong in Retail
As generative AI rapidly becomes a core part of retail, it powers automated customer replies, creates dynamic copy and delivers in-the-moment personalization. However, when these tools are trained on fragmented or incomplete data, they fail in high-visibility ways. Common missteps include sending a discount offer to a customer who just purchased at full price or mistaking a one-time visitor for a high-value repeat guest. These kinds of mistakes frustrate customers and can prevent organizations from building long-term trust and loyalty.
Consumer expectations around AI are also rising. Shoppers assume that brands using AI should know who they are and what they want. When that promise falls flat, the damage can be significant. This is why data quality and identity resolution must come before experimentation. The best AI experiences are built on trust, which starts with accurate, unified data.
However, having the right data is only part of the equation. Retailers must also focus on bridging the gap between strategy and operational execution.
The Ambition-Execution Gap
Even as AI use cases expand—from personalized marketing to inventory forecasting and customer support automation—many retailers lack the operational infrastructure to fully support them. The study shows that retailers have ideas, but just don’t have the infrastructure to deliver on them.
Among the most commonly cited barriers, 58% of retailers say data fragmentation is the biggest blocker to accessing AI’s potential. And, 33% of retailers report that lack of integration is holding them back from realizing the full value of AI.
But underneath these infrastructure challenges lies a deeper issue: most of today’s AI struggles can be traced back to a single point of failure: fragmented customer profiles.
Should You Trust AI to Build Your Dream Car?
Before electronic computers became widespread, NASA’s most advanced technology wasn’t a machine. It was people. Teams of human “computers”–many of them women–performed intricate calculations using nothing but pencils, paper, and chalkboards… Continue reading
The Identity Problem
Retailers interact with customers across dozens of touchpoints such as mobile apps, e-commerce sites, in-store systems, loyalty programs and third-party delivery platforms. Each one generates useful data, but they often remain siloed, preventing a cohesive customer view.
That means AI tools trained on this data are working with incomplete or inconsistent information. For instance, a customer might appear as a first-time visitor in one channel and a VIP in another. Offers may reach the wrong audience, product suggestions may feel generic or conversation histories aren’t unified to provide meaningful context. These fragmented systems erode trust in both the customer experience and the AI systems themselves.
The retailers making the biggest strides with AI aren’t necessarily using more advanced or sophisticated models. They’re solving the identity problem first by using AI to clean, reconcile and unify their customer data, creating a single, actionable profile that the rest of their technology stack can effectively use.
Why Identity Unlocks Better Results
Unified customer data makes the entire business smarter. In the survey, retailers who had strong identity resolution capabilities were nearly two times more likely to use AI across multiple business units. Identity resolution (especially when powered by machine learning) doesn’t require perfect data inputs. It can reconcile discrepancies, connect anonymous sessions to known users and infer relationships between transactions, devices and behaviors. This level of intelligence makes everything easier, from churn prediction to real-time offer generation to adaptive loyalty programs.
Where Retailers Can Start
Closing the gap between AI ambition and execution doesn’t require scrapping everything. But it does mean approaching customer data differently. Forward-thinking retailers are focusing on:
- Modernizing data infrastructure to enable unified, real-time access to customer information.
- Investing in identity resolution to connect cross-channel interactions into actionable profiles.
- Deploying AI tools in targeted use cases that drive revenue or reduce operational costs.
- Training teams to not only use AI tools, but to understand how to validate and apply their outputs effectively.
By treating data as a foundational capability, retailers can build AI initiatives that scale effectively across the organization.
The Future of AI in Retail Starts Now
The challenge in retail isn’t imagining what AI can do—it’s executing on that vision. And execution starts with data.
Unified, accurate, real-time customer data is the foundation that turns AI experiments into measurable business results. In a market where customers expect seamless, relevant experiences, it may be the single most important differentiator retailers can build.
Because at the end of the day, AI doesn’t fail for lack of strategy. It fails for lack of data.
