As AI-driven marketing accelerates, brands grapple with the costly consequences of flawed customer data. Mark Venables explores how identity resolution challenges undermine personalisation, compliance, and revenue, and how AI offers a path forward.
Amid the rapid acceleration of AI-driven marketing, businesses still struggle with a fundamental issue: knowing who their customers are. Brands operate across industries with profoundly flawed customer data, from retail to travel. Mistaken identities – mismerged profiles, duplicate records, and splintered customer histories – are more than an IT nuisance. They erode customer experience, waste marketing budgets, and put businesses at legal risk.
“What we see across brands is pretty shocking,” Matthew Biboud-Lubeck, AVP, UK/EMEA at Amperity, says. “Amperity uses AI to combine people-based data to mimic human perception, addressing issues that traditional rule-based methods struggle with. Across the 400 brands we work with, we have assessed the current state of data quality and helped them improve it. On average, about 20 per cent of large brands’ customer data is misidentified. This could mean customer profiles are splintered across the organisation and not connected, or they could be incorrectly merged, linking two different people as if they were the same.”
The impact of these errors is far-reaching. Customers receive irrelevant marketing offers, loyalty schemes fail to recognise high-value shoppers, and businesses struggle to execute meaningful personalisation.”The reasons for these errors vary by industry,” Biboud-Lubeck adds. “In retail, for example, customers often provide multiple email addresses to take advantage of first-time purchase discounts, which can fragment their identities in a brand’s system. Beyond that, data is naturally messy. People use multiple email addresses, change names when they marry, and move to new addresses. Customers may have a relationship with a brand for decades, and their identity shifts over time. The ability to bring all that data together and consistently understand who a customer is becomes crucial.”
The limits of virtual avatars
While AI improves identity resolution, some argue the next step is a persistent, virtual customer avatar, a single, unified profile that follows users across digital and physical interactions. The implications, however, are complex. “That raises significant privacy concerns,” Biboud-Lubeck says. “The old advertising world relied on tagging users when they browsed a website, with little transparency about how their data was being used.
“That is the model we should be moving away from. Instead, customers should opt in with clear consent for a comfortable experience. The idea of a virtual avatar storing all personal data is something I would challenge—both in terms of its feasibility and how much customers would want it. Many would find it unsettling rather than useful.”
Data challenges by region
While identity resolution issues exist globally, the regulatory landscape changes the stakes. Europe’s GDPR framework has made identity management a legal obligation rather than an operational preference. “Some things are different, but some are the same,” Biboud-Lubeck explains. “The most significant difference is that Europe has the world’s most stringent data privacy regulations. GDPR has set the standard for consumer protection, and the penalties for getting identity management wrong are much higher. This makes it a legal obligation rather than a nice-to-have.
“However, how brands have tried to solve these problems is fairly universal. Historically, brands have relied on rule-based approaches to match customer identities, but these methods have serious limitations. Take my name, for example – Matthew Biboud-Lubeck. If I get married and change my surname, a rule requiring last names to match across records would never link my old and new profiles. This highlights the flaws in rule-based identity resolution, which we see consistently across regions.”
AI’s role in solving identity resolution
Solving customer identity issues at scale requires AI-driven approaches. The complexity of the problem means rule-based systems fail, and businesses must adopt advanced machine-learning techniques. “There are two core challenges in data unification,” Joyce Gordon, Head of Generative AI at Amperity, says. “The first is ensuring accuracy, stitching profiles together correctly. The second is handling the sheer computational complexity of the process. Comparing tens of millions of records pairwise is an enormous challenge.
“We use several techniques. The first is semantic tagging and standardisation, which helps us identify common elements across different data sources, such as name, phone number, or address, before matching records. Then, we reduce the computational complexity by identifying the best possible matches so we do not have to compare every record against every other.
“We run over 45 similarity models to assess matches, accounting for names, nicknames, and email format variations. We then apply ordinal regression to assign a confidence score to each match. This is important because identity resolution is an unsupervised problem; there is no single correct answer. By making the process transparent, we allow brands to decide how conservatively or liberally they want to match records based on different use cases, such as marketing campaigns versus financial reporting.”
The risks of poor identity resolution
Beyond poor customer experience, brands face financial and compliance risks. GDPR regulations demand businesses accurately process customer data, and failure to do so invites legal scrutiny. “The biggest risk is compliance,” Biboud-Lubeck warns. “If a customer opts out or requests to be forgotten under GDPR, but the brand’s systems are disconnected and identities are split, that request might only be actioned in parts of the business. If audited, the brand would be found non-compliant. If there were a lawsuit, they would be non-compliant.”
Beyond compliance, revenue loss is a primary concern. “The financial impact is significant,” Biboud-Lubeck continues. “One major retailer we worked with estimated they were losing around £47 million in potential personalisation-driven revenue due to poor identity resolution. Brands that get this right will lead the market. Those that do not will fall behind.”
Creating one version of the truth
A fragmented customer view impacts every part of the business, from marketing to logistics. Brands must avoid fixing identity issues at a department level and embrace an enterprise-wide strategy. “Brands should stop solving this issue for individual use cases,” Biboud-Lubeck says. “If they only fix identity resolution for e-commerce personalisation, they perpetuate the problem by reinforcing data silos. The solution must be enterprise-wide, making a unified data foundation available across teams. That means supply chain teams understand top customer purchasing behaviour, corporate analytics teams access comprehensive customer insights, and omnichannel marketing teams can engage consistently across platforms.”
Unlocking audience attribution in fast casual dining
A Fortune 500 fast casual dining brand faced a significant challenge: understanding and engaging its entire customer base, not just its loyalty program members. With over 100 million in-store transactions annually, most customers paid with credit cards without providing personally identifiable information (PII). This lack of data meant the brand couldn’t personalise experiences for many of its customers, optimise marketing spend, or accurately attribute paid media investments. In a competitive landscape where personalisation drives consumer choice, this data gap disadvantaged the brand.
The brand partnered with Amperity to address these challenges, leveraging its advanced customer data platform (CDP) to build a comprehensive, always-on Customer 360 database. Amperity’s identity resolution and audience expansion technologies played a critical role in bridging the gap between fragmented transaction data and actionable insights.
Amperity’s patented identity resolution layer unified online and offline first-party data, stitching together disparate records from store transactions, loyalty programs, Wi-Fi registrations, and location data. By probabilistically matching records across these systems, Amperity created a persistent identity graph that provided a holistic view of each customer. This process helped the brand deduplicate multiple loyalty memberships and accurately link customers to their transaction histories.
To unlock insights from non-loyalty customers, Amperity deployed its audience expansion technology, which intelligently analysed sparse transactional data, such as credit card purchases, to identify potential customer identities. By incorporating behavioural data, purchase patterns, and predictive analytics, the brand uncovered 30-50 per cent of previously unknown in-store customers. This dramatically expanded the marketable audience, enabling the brand to engage these customers through digital and direct mail marketing.
The results were transformative. The brand increased its marketable audience by 144 per cent and built a unified Customer 360 database that integrates billions of historical records with over 30 million new records daily. With a clearer picture of its entire customer base, the brand optimised media spending by suppressing existing loyal customers from paid campaigns and reallocating the budget toward new customer acquisition. Additionally, it improved marketing effectiveness by matching in-store transactions with digital impressions, enhancing attribution and ROI tracking.
Beyond marketing gains, integrating Amperity’s platform supported compliance with data privacy regulations like the California Consumer Privacy Act (CCPA), allowing the brand to maintain consumer trust. By leveraging data-driven insights, the company transformed its customer engagement strategy, ensuring that every customer—loyalty member or not—receives a personalised and seamless experience.
The evolving role of AI in customer data
As AI advances, businesses must prioritise data accuracy to benefit from AI-driven personalisation. Identity resolution is the foundation upon which the future of customer engagement is built. “When generative AI emerged, many brands saw it as a way to personalise at scale,” Gordon concludes. “The marginal cost of content creation dropped to near zero. However, this is only useful if you have a robust data foundation. Without it, you end up with even more irrelevant messages and emails. You cannot take advantage of AI-driven personalisation without first solving the underlying data issue.”
The message is clear: Businesses that continue to rely on outdated identity resolution methods will fall behind. Those that use AI to create a unified, accurate customer identity will unlock greater personalisation, stronger customer loyalty, and higher revenue. Those who hesitate risk being left in the past, unable to effectively engage customers in an increasingly AI-driven world.
