There is a version of personalization that everyone finds irritating: the ad that follows you around the internet for three weeks after you already bought the thing, the email that addresses you by first name and then proceeds to be entirely irrelevant, the recommendation engine that confidently suggests you buy a second sofa. These failures are not incidental. They are the visible edge of a much larger tension at the centre of modern marketing, between the technical capability to personalise at scale and the human judgment required to do it well.

Hyper-personalization, the use of real-time behavioural data, machine learning, and predictive analytics to deliver genuinely individualised experiences, is not a new idea. What is new is the infrastructure to make it work across millions of interactions simultaneously, and the raised consumer expectations that have made mediocre personalisation actively damaging to a brand.

What the performance data shows

The commercial case for personalisation is well established. McKinsey estimates that organisations in the top quartile of personalisation generate substantially greater revenue from personalisation activities, up to roughly 40% more than average performers. Businesses using advanced personalisation see revenue lifts between 10% and 30%, and 78% of consumers say they are more likely to repurchase from a brand that provides personalised experiences. (McKinsey, 2021 / Salesforce, 2023)

The Amazon and Netflix examples from the original text hold up because they are genuinely instructive. Amazon's recommendation engine accounts for approximately 35% of total revenue, driven by systems that model not just what a user has bought but what purchasing patterns across millions of similar users predict they will want next. Netflix's personalisation extends to thumbnail selection: the same title can present differently to different users based on what visual language that user has previously engaged with. These are not marketing tricks. They are product decisions with direct revenue consequences.

What makes them worth studying is less the technology and more the underlying logic: both companies treat personalisation as a product feature rather than a communications layer. The experience is personalised, not just the message about the experience.

The paradox brands have not resolved

Here is where the conversation gets complicated. An Adobe study found that 44% of consumers feel frustrated when brands fail to deliver personalised experiences, while 70% are uneasy about how their data is collected and used. Both numbers are real. They describe the same consumer, which is the actual problem (California Management Review, 2025).

People want to be understood. They do not want to be surveilled. The line between those two experiences is not always obvious from the outside, which is why personalisation so easily tips into something that feels intrusive rather than helpful. Roughly 44% of consumers cite concern over how AI uses their data, and algorithmic bias, where AI systems unconsciously reflect the biases present in their training data, has emerged as an ongoing regulatory and ethical concern. One marketing analyst cited in research on this topic noted: their AI was recommending premium products exclusively to users in high-income postal codes, a pattern that had gone undetected until a bias audit surfaced it.

These are not edge cases. They are structural risks in any system that learns from historical data and then uses those patterns to shape future behaviour.

The trust architecture

According to Salesforce, 92% of consumers are more likely to trust brands that clearly explain how their data is used. That figure is useful precisely because it reframes the problem. The question is not whether to personalise. It is whether the brand has built the trust infrastructure that makes personalisation feel like a service rather than an extraction.

This is where zero-party data, information that consumers actively and willingly provide in exchange for a better experience, is becoming strategically significant. Unlike third-party data scraped from external sources, or first-party data collected passively, zero-party data is given with explicit intent. A consumer who fills out a preference quiz, saves items to a wishlist, or configures their experience is not being tracked. They are participating. That distinction matters enormously for how the resulting personalisation is perceived, and increasingly for regulatory compliance as GDPR, CCPA, and equivalent frameworks continue to tighten.

Where most brands actually are

The gap between what hyper-personalisation can do and what most brands have implemented is significant. 59% of marketing leaders now use AI to improve personalisation, but using AI for personalisation and doing personalisation well are different things. The failure modes are consistent: fragmented customer data across platforms that prevents coherent personalisation, over-reliance on behavioural signals without contextual understanding, and personalisation that optimises for short-term conversion rather than long-term relationship quality.

The brands that have genuinely closed this gap share a characteristic: they treat personalisation as an organisational capability rather than a technology purchase. The tools matter less than the data governance, the cross-functional alignment between product, marketing, and engineering, and the ongoing human judgment applied to what the systems surface. AI can identify that a customer is likely to churn. It takes human understanding to know whether the right response is a discount, a conversation, or simply leaving them alone.

Hyper-personalisation is not the future of customer experience. It is already the present. What remains unevenly distributed is the discipline to do it in a way that builds trust rather than quietly eroding it.

References

• McKinsey & Company. (2021–2022). The Value of Getting Personalization Right. McKinsey. Link
• California Management Review / Berkeley Haas. (2025). Balancing Personalized Marketing and Data Privacy in the Era of AI. CMR. Link
• Salesforce. (2023). State of the Connected Customer. Salesforce. Link
• Adobe. (2023). Digital Trends Report. Adobe. Link
• Gomez-Uribe, C. A., & Hunt, N. (2016). The Netflix Recommender System. ACM Transactions on Management Information Systems. Link

‍