In boardrooms across the financial services industry, a familiar refrain echoes: How can we leverage artificial intelligence (AI) to optimize pricing? The promise is seductive. Algorithms that analyze millions of data points, detect patterns invisible to human analysts, and deliver hyper-personalized price recommendations in milliseconds. Banks are investing billions in this pursuit, convinced that AI will unlock unprecedented efficiency and customer satisfaction.
This article argues for a more tempered view. While AI offers genuine value in certain pricing dimensions, the banking industry’s rush to algorithmic pricing overlooks a fundamental truth: price is not a number. It is a perception. And perception, shaped by emotion, culture, psychology, and individual circumstances, resists the neat categorization that machine learning requires.
The irreducible subjectivity of value
Consider a successful professional in Mumbai. She takes a loan to purchase a home worth ₹5 crore (~USD 0.5 million) without hesitation. The property represents security, status, family legacy, and emotional fulfilment. The same woman walks out of a store rather than pay ₹ 2000 (~USD 20) for a pair of jeans she considers overpriced. The jeans, in her calculus, are a fleeting fashion statement while the home is a life anchor.
No algorithm examining her income, spending patterns, or demographic profile would predict this behavior. The price sensitivity is not about the amount to be spent but it is about what the ‘purchase’ means to the buyer.
This phenomenon scales infinitely. One person spends $2000 on a bottle of wine at dinner, savoring the experience as a reasonable indulgence. Another invests $700K in a high-performance bicycle, viewing it as an extension of identity and discipline. Each would likely regard the other’s spending as absurd, even wasteful. Yet both are acting rationally within their own frameworks of value.
Neither is foolish. Both are simply pricing the world through their personal lenses.
The Psychology That Algorithms Cannot Parse
Behavioral economics has long recognized that price perception operates beyond rational calculation. The phenomenon of charm pricing where $999 feels meaningfully cheaper than $1,000 despite a difference of a single dollar, illustrates how deeply irrational our relationship with numbers can be. The left-digit effect, as it is formally known, has been documented for decades, yet it continues to influence purchasing decisions across every demographic and income level.
Morgan Housel, in The Psychology of Money, captures this beautifully when he observes that financial decisions are driven less by spreadsheets than by personal histories. A person who grew up in scarcity may hoard cash despite inflation eroding its value not because they cannot calculate returns, but because the feeling of liquidity provides a psychological security no investment can match. Another person, raised in abundance, may take risks that appear reckless but reflect a fundamentally different emotional relationship with wealth.
Consider the investor who sleeps soundly with 7% annual returns from fixed deposits while another, earning 15-20% in equities, lies awake worrying about volatility. The first is not less sophisticated, they have simply priced peace of mind into their investment calculus in a way no algorithm can quantify.
The Cultural Dimension
Add to this the layer of cultural conditioning, and pricing complexity deepens further.
Stereotypes, while reductive, often contain grains of behavioral truth that influence financial decisions. Cultural conditioning of a few communities, for instance, is associated with an almost visceral reluctance to part with money unnecessarily. An inherent disposition toward frugality, negotiation, and value extraction from every penny. Another community, by contrast to the frugal community, may celebrate generosity, visible success, and a certain flamboyance in spending.
These are not fixed rules, but they are real tendencies that shape how individuals within these communities perceive value, respond to pricing, and engage with financial institutions. A single interest rate change, a single fee structure tweak, a single promotional offer will land differently across these cultural contexts- not because of income or credit score, but because of inherited frameworks for thinking about money.
AI can segment by geography, language, and spending history. It cannot segment by worldview.
What AI Can and Cannot Do
Let us be precise about where artificial intelligence adds value in pricing.
AI excels at pattern recognition across scale. It can identify which customers are likely to respond to which products, predict churn, detect fraud, and optimize the timing of offers. It can process real-time market data to adjust rates dynamically. In competitive analysis, in operational efficiency, in identifying broad behavioral clusters, AI delivers genuine advantages.
Relationship-Based Pricing (RBP), now common in banking, uses AI to adjust price points and rates based on a customer’s total relationship with the institution: deposits held, products used, tenure, and transaction patterns. This is valuable. A customer with a $5 million in deposit and three product relationships deserves a preferential pricing conversation than a new customer with a single savings account.
But RBP, even in its most sophisticated forms, operates from the bank’s perspective of ‘What is this customer worth to us?’ True hyper-personalization requires inverting the question: What is this product worth for this customer, given how they think about money?
This second question demands understanding, not just behavior but motivation, not just transactions but meaning. It requires deciphering whether a customer sees a home loan as an investment, a burden, a rite of passage, or a risk. It requires knowing whether a credit card limit feels like freedom or temptation, whether a fee waiver signals value or suggests desperation.
This is not a data problem. It is an interpretation problem. And interpretation, particularly of emotion and context, remains stubbornly human.
The Limits of Micro Hyper-Personalization
The industry’s current trajectory points toward ever-finer segmentation of micro hyper-personalization, where each customer receives pricing calibrated to their unique profile. The vision is compelling: an AI that knows you so well that it can predict exactly what price point will maximize both your satisfaction and the bank’s profit.
But this vision collides with a fundamental challenge: Two customers with identical data profiles may have entirely different internal relationships with money.
Two thirty-year-old professionals in Chicago, earning similar incomes, with similar credit histories, similar family structures, may respond to the same loan offer in opposite ways. One grew up watching his father’s business fail due to debt carries an emotional aversion to borrowing that no amount of favourable terms will overcome. The other was raised to see leverage as a tool for growth and he will eagerly accept terms that might alarm the first.
No amount of transactional data captures this. The algorithm sees two nearly identical profiles. The humans inhabiting those profiles see entirely different propositions.
The Case for Human Judgment
This article is not an argument against AI in banking. Technology is too powerful, and its applications too valuable to dismiss. But it is an argument for humility about what AI can achieve in pricing and for continued investment in the human judgment that algorithms cannot replicate. The best combination would be to utilize AI to decode data and let it decipher and present different options. Picking and choosing the best option should still be left to the best judgement of the human mind.
Pricing, at its core, is an act of communication. It signals value, respect, and understanding. A price that feels right to a customer reflects not just fair exchange but recognition. This institution understands who I am and what I need.
That recognition requires:
- Emotional intelligence: the ability to read unstated concerns, to sense hesitation, to understand what a customer is really asking beneath the surface question
- Contextual judgment: knowing when to bend a rule, when to offer a concession, when to stand firm
- Cultural fluency: understanding how different communities and backgrounds shape financial behavior
- Ethical navigation: recognizing when hyper-personalized pricing crosses from value creation into exploitation
These are not capabilities that can be encoded in a model. They are capabilities that experienced bankers develop over careers spent in conversation with customers.
The Path Forward
The most effective pricing strategies will not replace human judgment with AI but will augment human judgment with AI-generated insight.
A recent McKinsey study states that for global banking, AI technologies could potentially deliver up to $1 trillion of additional value each year.1AI can surface the data regarding customer’s transaction patterns, their product usage, their likely sensitivity to rate changes. But the final pricing decision particularly for significant products, for relationship critical moments, for complex or ambiguous situations should remain with humans who can integrate that data with everything the algorithm cannot see.
This is not inefficiency. It is wisdom.
In a world obsessed with automation, where every CXO seeks to demonstrate technological sophistication, there is courage in acknowledging limits. The banks that thrive in the coming decade will not be those that automate the most, but those that automate wisely. Deploying AI where it excels and preserving human judgment where it remains irreplaceable.
Pricing, that most intimate of financial conversations, is precisely such a domain.
Conclusion
The AI conundrum in pricing is not a technical problem awaiting a technical solution. It is a reminder that banking, for all its quantitative sophistication, remains a fundamentally human enterprise.
Customers do not experience prices as numbers. They experience them as judgments about their worth, their relationship, their respect. Those judgments are filtered through emotional histories, cultural inheritances, and personal philosophies that no dataset fully captures.
AI will continue to improve. Its role in pricing will expand. But the irreducible subjectivity of value and the fact that the same $ 999 means something different to every person who encounters it ensures that pricing will never be fully algorithmic.
The banks that recognize this will earn something no algorithm can optimize: TRUST.
Price is what you pay. Value is what you feel. And feeling, for now, remains human terrain.