97% of retailers say they’re all-in on AI. Only 11% are ready to scale it. Here’s what that disconnect really means.
The headlines will tell you that AI is transforming retail. Chatbots. Product recommendations. Personalized emails that “feel human.” Internally, sales and support teams are already riding the wave – nearly half of retailers say they’re using AI weekly, if not daily.
But underneath the optimism, a different picture is emerging: one of unprepared systems, disconnected data, and organizations far less ready than they think.
That’s the clearest takeaway from Amperity’s just-released 2025 State of AI in Retail report, which surveyed 1,000 professionals across marketing, IT, analytics, and executive leadership. The findings are sharp. Nearly every company surveyed plans to maintain or increase its AI investment over the next year – but only 11% say they’re prepared to scale AI across the enterprise.
This isn’t an AI problem. It’s a data problem. And the gap between ambition and execution is widening.
The AI Gold Rush Meets a Data Bottleneck
Let’s start with the good news: AI is no longer on the fringe. According to the report:
- 45% of retailers use AI daily or several times per week
- 97% will maintain or increase AI investment in the next 12 months
- 65% believe AI will improve customer lifetime value
- 63% believe it will strengthen customer loyalty
Clearly, leaders see AI not as a side project but as a strategic lever.
But now the reality: fewer than half are using AI for customer-facing use cases. Only 23% have it in production for identity resolution or marketing data prep. Just 21% are “very confident” in their ability to understand and act on customer behavior.
In other words: lots of experimentation, very little operational trust.
Why the Customer-Facing Layer Is Still a No-Fly Zone
Internally, AI has found traction – think automation, support queries, or internal analytics. These are low-risk, high-ROI zones. But customer-facing deployments, where the brand is exposed, remain rare.
According to the data, only 43% of retailers are using AI to shape real-time customer experiences. The reluctance isn’t ideological – it’s infrastructural. Fragmented customer data. Limited identity resolution. Inconsistent personalization logic across channels.
The key obstacle? Many retailers simply don’t have a clean, unified view of the customer. Without that, putting AI in front of customers is like putting a jet engine on a bicycle frame: the power is there, but the structure can’t handle it.
CDPs: The Real Line Between AI Dabblers and Doers
One of the most important findings in Amperity’s report isn’t about AI at all. It’s about infrastructure.
Retailers with Customer Data Platforms (CDPs) are operating at a fundamentally different level:
- 60% of CDP-equipped retailers use AI daily or multiple times per week (vs. 29% without)
- 35% use AI in production to prepare data for marketing or analytics (vs. 9% without)
- 22% report full AI adoption across multiple business units (vs. 10%)
Put plainly: CDPs don’t just support AI – they enable it. Without unified data, AI can’t personalize, predict, or act with confidence.
Retailers who treat AI like a plug-and-play capability will hit a wall. Those who build the infrastructure to support it – beginning with identity resolution and data unification – will scale faster and outperform.
The False Comfort of AI Enthusiasm
There’s a dangerous assumption hiding inside the retail industry’s enthusiasm: that the mere adoption of AI tools implies readiness.
It doesn’t.
As the report shows, AI usage is common but meaningful AI outcomes are not. A 45% usage rate might sound impressive until you see that just 11% of companies feel prepared to deploy AI at scale. It’s like buying a 3D printer and assuming you now run a manufacturing plant.
The challenge isn’t access to tools. It’s the ability to integrate, govern, and operationalize them – especially across departments where data ownership and goals still diverge.
Identity Resolution: The Underfunded Keystone
If AI is the engine and data is the fuel, then identity resolution is the ignition switch.
And it’s being ignored.
Only 20% of retailers are prioritizing identity resolution as an AI use case even though it underpins nearly every meaningful personalization effort. Without accurate, cross-channel identity resolution, your predictive models are guessing at best, hallucinating at worst.
It’s also telling that only 3 in 10 marketers can manage customer data without leaning heavily on IT. In 2025, that’s a structural liability. AI doesn’t just need clean data – it needs it accessible, adaptable, and owned by the people closest to the customer.
What’s Really at Stake
Retailers are under pressure from every side: thin margins, tariff uncertainty, shifting consumer behavior. The promise of AI is not abstract – it’s operational. Better targeting. Faster decision-making. Adaptive offers. Reduced CAC. Increased LTV.
But here’s the real headline:
The brands that win in AI won’t be the ones with the flashiest chatbot.
They’ll be the ones that make sense of their own customer data first.
This means solving the hard stuff: breaking down data silos, building real-time profiles, and investing in foundational systems that support trustable automation.
Retailers don’t need more AI pilots. They need more data strategy.
Four Questions to Ask Right Now
For executives trying to separate signal from noise in their own AI journey, start here:
- Do we have a unified, accessible view of the customer across channels?
If not, AI will only amplify your fragmentation. - Can marketing and analytics teams operate without leaning on IT for segmentation, data prep, or experimentation?
If not, your AI tooling may stay stuck at proof-of-concept. - Are we prioritizing identity resolution as a strategic capability, not just a technical one?
If not, expect personalization blind spots that undercut your best campaigns. - Are we clear on where AI adds customer value – not just internal efficiency?
If not, you risk spending big on tools that never touch the people who matter.
Final Thought
AI is not a feature. It’s a force multiplier. But only for companies whose data foundation is strong enough to support it.
Amperity’s report makes one thing clear: there’s no lack of intent in retail. But there is a serious lack of readiness. Closing that gap will take more than vendor demos and pilot programs. It will require structural investment in how customer data is collected, resolved, activated, and owned.
The winners won’t be the first to use AI.
They’ll be the first to trust it with their customers.
