When the Buyer Has an Agent: What Happens to Beauty’s Brand Premium?

AI shopping agents could weaken inherited advantages, open new demand spaces and make successful relationships more valuable. The decisive question is who captures the learning—and the economic return.

A woman has bought the same prestige moisturiser for twenty years. She likes the texture, trusts the name and knows what to expect. Then she gives an AI shopping agent a different instruction: find something that meets her needs for half the price.

How much are those twenty years worth now?

The agent has no nostalgia for the department-store counter. It has never admired the jar on a dressing table. But the woman has. And if the product has worked reliably for her, that history contains information an unfamiliar alternative does not possess.

The question is how much of the premium survives when someone starts examining what she is paying for.

That examination could redistribute existing beauty purchases, bring previously unserved buyers into the market or help businesses develop better offers. It could also transfer the most valuable customer knowledge to the agent provider.

Consumer benefit, market growth and an individual brand’s profitability can move in different directions. Agentic commerce may improve all three. There is no reason to assume it will.

Efficiency gets the order through. What makes value compound?

In his recent essay on Estée Lauder, Erich Joachimsthaler distinguishes the efficient enterprise from the compounding enterprise. It is a useful framework for understanding what shopping agents might change. [1]

The efficient enterprise makes its existing operations work better: lower costs, faster processes, more accurate forecasts. Joachimsthaler sees a continuous ambition running from process reengineering through digital transformation to AI. Each generation brings more powerful tools to the task.

His compounding enterprise begins outside the company, with demand forming in people’s lives. It follows that demand through the interactions that shape choices, the investments that serve those needs and the economic profit available for reinvestment. Its three loops are interactions, economic profit and learning.

His historical example is Youth-Dew. Introduced in 1953 as a bath oil that doubled as a perfume, it helped move fragrance from an occasional gift into a woman’s own daily ritual. In his account, Estée Lauder changed the occasion, the purchasing decision and the pattern of demand. [1]

Shopping agents raise a comparable strategic question. Automating checkout can make an existing purchase easier. Understanding a need that the existing product range serves badly could lead somewhere more valuable.

But repeat orders alone do not establish compounding. Something must accumulate—knowledge, trust, better fit or an advantage in serving customers—and improve subsequent returns. Efficiency can fund that process. It does not establish that the process exists.

Rearranging the deckchairs—or opening a demand space?

Imagine an agent replacing an $85 cream with a $25 alternative the buyer considers satisfactory. The challenger gains a customer. The incumbent loses one. The buyer keeps $60.

That is a meaningful consumer benefit. On this purchase, however, beauty industry revenue has fallen. An agent has redistributed spending and reduced its cost.

Now consider someone who has avoided skincare because the choices seem confusing, expensive or unsuitable. An agent helps identify a manageable routine within a realistic budget. A previously unserved need becomes a purchase.

A third buyer asks for an outcome: a simpler routine that fits changing circumstances, preferences and spending limits. The alternatives may span established categories, products already owned, professional advice and buying nothing additional.

Here the demand space becomes more useful than the category. A serum manufacturer may discover that its relevant competition extends well beyond other serums. It may also discover an opportunity for a different product, bundle or service.

BCG and WWD’s 2026 study of approximately 5,000 US adults illustrates the broader context. It identifies an “Optimizer” segment spanning traditional beauty, aesthetic procedures, and performance and longevity offerings. About 75% of that segment uses AI for research. This is a highly engaged group, not three-quarters of all beauty buyers. [2]

The research establishes interest in AI-assisted decisions. It does not demonstrate the market-wide effects of autonomous purchasing. The strategic opportunity is to investigate needs that cross company categories, then establish whether serving them creates additional demand and an attractive return.

The threat from Korea is bigger than dupes

Beauty is an instructive test because its prices reflect a mixture of performance, reassurance, pleasure, identity and status. Their relative importance changes with the buyer and the occasion.

McKinsey and The Business of Fashion’s 2025 research describes consumers scrutinising value and product performance more closely. Agents would enter a market where brands already face pressure to justify their prices. [3]

Affordable alternatives, including products marketed as “dupes,” could benefit when buyers ask for comparable benefits at lower prices. An agent can widen the shortlist and reduce the work involved in considering an unfamiliar name.

Korean beauty, however, should not be reduced to imitation. Its competitive threat includes brands with independent identities and offers. Amorepacific’s 2025 results report growth in the Americas and EMEA and describe expansion involving LANEIGE, AESTURA and COSRX. These are established international competitors, not evidence of an AI-created phenomenon. [4]

The deeper threat to an incumbent is an offer that meets the buyer’s requirement convincingly without accepting the famous brand as the reference point.

A buyer asking for a particular result within a budget gives an agent a different starting point from “find an alternative to my usual prestige brand.” The incumbent may lose its position before comparison even begins.

Equivalence still has to be established. Shared ingredients do not prove equivalent performance: formulation, concentration, stability, testing and user experience can matter. The FDA’s testing guidance distinguishes ingredient information from additional testing that may be needed for finished products; safety evidence itself does not establish equal cosmetic results. [5]

An agent that collapses those differences could recommend badly. Both a prestige premium and a challenger’s value claim deserve scrutiny.

Loyalty, preference and the cost of looking elsewhere

Does twenty years of brand loyalty mean anything to an agent?

It should mean something when it tells the agent what matters to its buyer. It should carry less weight when it merely records a purchasing habit that the buyer now wants to reconsider.

Three instructions make the distinction clear:

“Reorder my usual cream from a trustworthy seller.”

“Find a cheaper alternative, but I am reluctant to change something I know suits me.”

“Start from scratch. Prioritise functionality and value; I do not care about the label.”

The first preserves the choice. The second tests the price of switching. The third gives the incumbent much less protection.

Repeat purchasing can reflect proven satisfaction, emotional preference or the effort of investigating alternatives. Those sources of loyalty have different prospects when an agent intervenes.

A record of satisfactory personal use supplies relevant information. Pleasure, fragrance, texture and associations can also be legitimate buyer preferences. An agent that ignores them would misunderstand its assignment.

The more exposed premium is the one sustained because comparing alternatives has been too much trouble. If an agent makes that comparison easier and credible, the protection weakens.

An established brand therefore needs to understand what its customers are loyal to. A name may express trust earned through experience. It may evoke something the buyer values. Or it may be the default that has never faced a serious challenge.

For a buyer explicitly seeking value and functionality, accumulated fame alone provides a limited answer to a current requirement.

The relationship may compound for the intermediary

Agents could also make successful choices harder to displace.

An agent that knows a buyer is satisfied, dislikes experimentation and values reliable replenishment may keep recommending the existing product. Challengers could gain an opening during initial comparison but encounter a higher hurdle once the agent has learned a successful routine.

That possibility deserves testing alongside predictions of effortless switching. More capable comparison does not imply constant reconsideration.

It also raises a question for brands: where does the learning reside?

The buyer’s agent could accumulate knowledge of preferences, budgets, disappointments and successful purchases while the merchant receives only an order. The buyer increasingly trusts the intermediary to maintain the routine and decide when a change is worthwhile.

The compounding asset may belong to the agent provider.

A beauty company can gain agent-mediated sales while losing visibility into why people choose it. Participation in the transaction does not guarantee participation in the learning.

Nor is every agent an unconstrained evaluator. Its available sources, merchant access, platform rules and commercial arrangements may shape the alternatives considered. Brands need evidence about particular agents under particular buyer instructions.

AIVO measures the choice. Assist supports the interaction.

Joachimsthaler’s framework helps executives ask which demand deserves investment. AIVO’s role is to contribute evidence about how those needs translate into AI recommendations and merchant interactions.

AIVO examines conversations as they move towards a recommendation. A brand can appear initially and disappear when the buyer adds a budget, questions the evidence or introduces a requirement. Measuring final recommendations helps establish where it wins or loses. Controlled follow-up tests can investigate the reasons.

The starting question matters. Testing only “which prestige serum?” excludes much of the demand space before the conversation begins. Research grounded in the buyer’s circumstances can reveal a wider set of alternatives and unmet requirements.

AIVO Assist supports the subsequent merchant interaction. It gives visiting AI shopping agents an on-site product expert they can question about facts, supporting sources, suitability and limitations, including when a product is a poor fit.

This provides a way for a prestige brand to explain a relevant difference and for a challenger to demonstrate its value. The visiting agent can ask follow-up questions rather than rely on an isolated claim.

The connection is practical: AIVO measures where recommendations are won or lost; Assist helps resolve the questions influencing those decisions; subsequent testing can establish whether improvements changed the result.

Recurring questions could then inform the next business decision. A demand for a simpler routine might suggest clearer guidance, a different bundle or product development. A repeated objection might reveal missing evidence—or a limitation better acknowledged than obscured.

Capturing and validating those signals is a further learning opportunity. It requires permitted records, separation of genuine customer needs from automated noise, and merchant judgement. A transcript collection does not become strategic intelligence by itself.

Whose economic profit grows?

This returns us to Joachimsthaler’s economic profit loop. Additional orders can consume value if discounting, returns, intermediary costs and capital requirements outweigh the return.

The evidence of compounding would be improvement that carries forward: better fit, profitable retention, lower avoidable costs and stronger returns on the capital invested. New demand must also be distinguished from sales transferred between brands or channels.

Those benefits may accrue unevenly. The buyer could spend less. A challenger could acquire customers. The agent provider could deepen its relationship. The incumbent could lose margin. A growing demand space does not guarantee growth for the company that once dominated the category.

Beauty’s established names retain assets worth defending: research, trusted products, distinctive experiences and customer knowledge. Their task is to establish how those assets serve the needs buyers now bring to agents—and to learn where a different offer is required.

The brands that do so may find a route from efficient transactions to more valuable relationships. Others may discover that the transaction has become efficient chiefly at their expense.

For a buyer seeking functionality and value, the agent’s hardest question may be simple:

What, exactly, am I paying extra for?

Sources

1. Erich Joachimsthaler, essay on Estée Lauder, the efficient enterprise and the compounding enterprise, November 2026. Original article.
2. BCG and WWD, Consumers Are Shaping a New Age of Optimization in Beauty, 13 May 2026.
3. McKinsey and The Business of Fashion, State of Beauty 2025: Solving a Shifting Growth Puzzle, 9 June 2025.
4. Amorepacific, Amorepacific Group 2025 Earnings Summary.
5. US Food and Drug Administration, Product Testing of Cosmetics.

The purchasing examples are hypothetical. The article’s prospective mechanisms are analysis, not measured market-wide outcomes of autonomous commerce. No equivalence between named products is asserted.