The mass production paradigm is at an end; the AI era has arrived. Here’s what it means for consumer marketing
We are in the midst of a great transition: from mass production and marketing of just about everything, to hyper-personalized and targeted services. Understanding consumers is at once easier and harder: we have more data and feedback to work with, yet value increasingly comes from complex relationships in which straightforward features and benefits don’t dominate purchasing decisions.
Despite the changing consumer landscape, the fundamental strategic question remains the same. What do customers actually value and are we delivering it? The question sounds simple. In practice, it is fiendishly difficult to answer. Customers frequently cannot accurately articulate what drives their decisions. Companies frequently confuse what is easy to measure with what is strategically important. The result is product and service design that optimizes for the wrong things, leading to overinvestment in attributes customers don’t value, and underinvestment in those they do.
Understanding what consumers value
A rigorous approach to categorizing the attributes of your product or service, which I call attribute mapping, offers a path through the confusion. For any given customer segment, there will be attributes that are positive – things they are willing to pay for. Inevitably, there will also be negatives – things they would prefer to do without. And for many customers, there will be neutral attributes, which they simply don’t care about. The emotional relevance of these attributes varies. Some are basic, meaning they are taken for granted. Some are discriminating: customers notice differences between different providers’ offerings. And some are energizing – the kind that customers stand outside in the rain overnight to get their hands on.
Basic positive attributes are non-negotiables: baseline expectations that any credible offering must meet. A bank’s mobile app must allow balance checking; a hotel must provide a bed. These attributes don’t win customers, but their absence loses them instantly. Companies rarely gain competitive advantage by investing further in non-negotiables once they meet the threshold.
Discriminating positives are differentiators; features where superior performance actively shifts purchase decisions in competitive situations. In the early days of smartphone cameras, for example, photographic quality was a differentiator for different models. Other discriminators operate at the ecosystem level. Having a green text bubble rather than a blue one – signaling the use of a device other than an iPhone – led some teenagers to fear for their social lives if they left the Apple ecosystem. Differentiators are worth investing in, but only to the extent that the investment yields competitive separation. Once competitors match performance, the differentiator migrates toward non-negotiable status, and further investment yields diminishing returns.
Energizing positives are arguably the most strategically interesting. These are exciters – attributes that customers didn’t know they wanted until they encountered them. Financial technology firm Stratyfy, for instance, is using AI pattern recognition to allow banks to confidently offer loans to people who would otherwise be rejected for reasons unrelated to their creditworthiness. It’s energizing for the banks – it opens up a whole new set of customers – and energizing for the customers, giving them access to financial resources. Unfortunately, once an exciter is accepted by customers, such features eventually become discriminators and eventually, through competitive matching and customer adaptation, non-negotiables.
Basic negatives are tolerables – attributes that customers dislike but accept because all competitive options share the same flaw.
Discriminating negatives are dissatisfiers, which can create competitive differentiation: if you can credibly claim to eliminate something customers used to tolerate, it can become an advantage. Amazon’s one-click purchasing experience, for example, made buying far easier.
Energizing negatives are enragers: features that cause customers to abandon an offering. Lululemon’s Get Low yoga pants enraged many customers when they were found to be see-through, inflicting serious damage on the brand.
When using the attribute mapping approach to assess your offering, it’s important not to focus solely on attributes within your own category. As the world becomes more digitized and product categories blur, changes from elsewhere can radically transform how customers respond to the attributes you offer. Consider how the advent of GLP-1 weight loss drugs – now reportedly taken by 1 in 8 Americans – have led to attribute ripple effects across industries. Traditional weight loss firms such as WW (formerly Weight Watchers), which preached discipline, rigorous measurement of portions, and group support, have found those attributes squarely in the negative category, as consumers warm to the new attributes of GLP-1 drugs – such as their ability to ‘eliminate food noise.’ Food producers and restaurants are also having to adjust, offering high-protein formulations and smaller portions.
Having rigorously assessed your offering, you should invest in discovering new exciters and differentiators, work urgently to neutralize dissatisfiers and enragers, bring costs down on attributes that are becoming non-negotiables, and ruthlessly eliminate neutrals that add cost without adding value to any segment you care about. Then, with the attribute mapping framework in mind, consider two big shifts in consumption patterns: the shift from products to services, and consumption activities mediated by digital agents.
From products to services: the feedback revolution
In the mass production paradigm that we are now leaving behind, the moment of sale was also typically the moment of ‘goodbye.’ A manufacturer designed a product, sourced materials, made it, pushed it through a distribution chain, and handed it off to a retailer. What happened next – how the customer used it, what frustrated them, what delighted them – filtered back slowly and imperfectly through market research, customer surveys, and lagging sales data. The product was frozen at the moment of manufacture.
That world is disappearing. In industry after industry, offerings that used to be sold as discrete products are now delivered as continuous services. The implications for understanding consumer attribute maps are profound. Consider the differences between buying a car and subscribing to a mobility service; purchasing software on a disc and paying a monthly SaaS fee; or buying a treadmill and joining a Peloton membership. In each case, the shift from product to service transforms the information architecture of the relationship. The service company receives real-time signals about how consumers actually behave, not how they say they behave in surveys. Every dropped session, every feature ignored, every workflow abandoned is a data point. The gap between stated preference and revealed preference – one of the great frustrations of market research – begins to close.
The mass market paradigm brought wonders to consumers. Economies of scale meant that companies could sell hard-to-manufacture goods affordably. Mass media, such as paid television ads, became a staple of marketing. Brands represented quality and idiosyncratic positioning in the minds of consumers. But now, economies of scale are giving way to hyper-local, digitized and personalized offerings. Mass media is being supplanted by social media, where it is possible – even probable – that no two people see exactly the same messages. Brands no longer carry the clout they once did.
All of this represents a fundamental reconception of what a company is selling. Increasingly, the answer is outcomes, not objects. Customers of cloud software don’t want servers; they want computing capacity on demand. Patients using digital therapeutics don’t want an app; they want managed health outcomes. When the product is the outcome rather than the artifact, the company has the potential to remain permanently in the feedback loop – if it designs its systems appropriately.
The strategic implication is significant: companies that complete this shift gain a continuous, iterative capability to map attributes and refine what they offer. Those that remain in a product mindset are operating with one hand tied behind their back. In competitive markets, that lag is increasingly fatal.
When AI does the shopping
The shift to services changes how companies can adapt the attributes they offer. The rise of agentic AI changes something even more fundamental: who the ‘consumer’ is. For a growing category of purchases, the answer will be a machine.
Amazon’s Subscribe & Save program offers an early glimpse of the direction of travel. Consumers set parameters – brand preference, quantity, frequency – and the system handles replenishment automatically. The consumer’s role in the recurring transaction is essentially zero. Multiply that logic across the full range of routine, predictable, low-involvement purchases – household staples, personal care products, standard office supplies, recurring B2B inputs – and the implications are massive. For a substantial share of commerce, the consumer making the purchasing decision may not be a human being at all.
Agentic AI systems, now emerging from every major technology platform, extend this logic further. These agents don’t merely automate repetitive buying; they can evaluate options, compare prices at speed, assess suppliers, negotiate terms and execute transactions – all without human involvement. Even more interesting from a game-theory perspective is that agents will likely be interacting with other agents, each trying to outsmart the other on behalf of the humans who set them loose. There is also the revealed preference issue: what consumers tell AI they want may not be what actually matters to them. Perhaps AI will be smart enough to figure that out!
The assumptions underlying how commerce works will be utterly undone. Marketing has long been in the business of influencing human psychology: structuring choice architectures, arousing aspirations, triggering emotions, and creating brand affinity. Those levers work on people. They don’t work on algorithms. An AI purchasing agent evaluating laundry detergent is not susceptible to aspirational advertising or choice architectures that capitalize on human bias. It is evaluating a structured set of criteria: price per unit, delivery reliability, environmental rating, compatibility with previous purchases, and whatever other parameters its human principal has established.
The second-order effects cascade quickly. Brand loyalty as traditionally understood may matter far less. What will matter is performing well on the attributes that agents are programmed to optimize. A company whose product scores well on those attributes wins the transaction; one focused on brand sentiment may find itself bypassed entirely.
This is not the end of brand or emotion in commerce. For high-involvement, high-consideration purchases – a car, a vacation, a piece of furniture – human judgment, aspiration and feeling will remain central. But for the vast categories of routine purchases, the rules are changing. Companies need to understand, urgently, which of their sales are vulnerable to agent disintermediation, and redesign their offerings accordingly.
The human and the algorithm
What makes the attribute mapping framework especially relevant for the AI age is that it applies with equal force to the two very different consumers companies must now design for. For the human consumer making a considered, high-involvement purchase, exciters matter enormously. Experience, emotion and surprise are strategically powerful. Investments in discovery – understanding what delights customers before they can articulate it – pay off in the form of genuine competitive separation. Yet for the AI agent handling routine procurement, what matters is performance on a structured set of criteria. The relevant question is: what attributes will the agent optimize for, and how does our offering perform against those parameters? Companies that understand this can design explicitly for agent-legibility – ensuring their products are rated, certified and structured in ways that score well on the criteria that agents apply.
The companies best positioned for what comes next will be those that maintain simultaneous mastery of both. They will use real-time data to close the gap between what customers say and what their behavior reveals they value. They will design routine offerings to be visible and attractive to AI agents, while investing in the experiential and emotional dimensions of high-involvement categories where human judgment remains central. And they will maintain the discipline to keep asking, for every attribute in their offering: is this a non-negotiable, a differentiator, an exciter – or a cost we’re carrying for no competitive reason?
The consumer has never been more empowered, more fragmented, or more technologically augmented. The companies that understand what ‘consumer’ now means – and design their offerings accordingly – will find that the AI age is not a threat to customer centricity, but its most demanding expression.
Rita Gunther McGrath is professor of management at Columbia Business School and a Duke CE educator
