From Keywords to Intent: Why Shopping Search Is Starting to Feel More Human

E-commerce has long relied on a structural compromise between how people think and how systems retrieve information. Consumers learned to translate needs into keywords, filters, and navigation paths, while retailers attempted to infer intent from clicks, dwell time, and conversion proxies. The arrangement scaled efficiently, but it imposed cognitive work on users that rarely matched how purchasing decisions are actually formed. Most shopping journeys begin with partial information, evolving constraints, and uncertainty about tradeoffs rather than with clearly articulated product specifications.

This mismatch created friction that became normalized. Users adapted to search boxes by iterating queries, opening multiple tabs, and cross-referencing external sources. Over time, the burden of research shifted almost entirely onto the consumer. The fact that this process became habitual did not make it efficient; it merely made inefficiency familiar.

Traditional Search vs AI-Mediated Discovery

Source: McKinsey & Company; Adobe Digital Insights

The current discovery shift represents a structural realignment between human intent and machine interpretation. AI-driven interfaces move beyond keyword matching toward intent resolution, meaning systems infer goals, constraints, and priorities from natural language rather than discrete terms. These systems maintain conversational context across turns, incorporate clarifying questions, and update recommendations as preferences evolve. McKinsey’s 2025 consumer research indicates that roughly half of consumers actively seek out AI-powered search tools, with a substantial share identifying them as a primary source of insight relative to traditional search.

This shift changes where decisions are made. When an AI layer synthesizes category information, frames tradeoffs, and resolves objections before a shopper reaches a merchant site, a large share of evaluation occurs upstream. McKinsey estimates that by 2028, approximately $750 billion in U.S. revenue will flow through AI-powered search interfaces. The implication is not the disappearance of conventional e-commerce, but a redistribution of influence toward environments that shape expectations before transactional intent is expressed.

Lowering Search Costs and Cognitive Load

From a behavioral economics perspective, search is a cost borne by consumers. Time, effort, and uncertainty influence how broadly people explore options and how confident they feel in eventual choices. E-commerce increased access and variety, but it also increased cognitive burden. Large assortments often create choice overload, particularly when evaluation criteria are ambiguous or difficult to compare across products. Under these conditions, additional options do not reliably improve satisfaction and can suppress both confidence and conversion.

AI-mediated discovery reduces these costs by restructuring the work of search itself. Instead of forcing consumers to externalize intent into filters and technical attributes, AI systems translate conversational input into constraints and priorities. They narrow option sets dynamically, explain why certain products are excluded, and surface tradeoffs that would otherwise require manual comparison. This process resembles guided deliberation rather than simple retrieval.

Adobe’s analysis of more than one trillion visits to U.S. retail sites provides empirical evidence that this restructuring is already influencing behavior. Between November and December 2024, traffic from generative AI sources to retail sites increased by more than an order of magnitude year over year, with particularly sharp spikes during peak shopping periods. By mid-2025, generative AI-driven retail traffic had expanded several thousand percent compared with the prior year.

Engagement metrics suggest that this traffic behaves differently from traditional search traffic. Adobe reports that visitors arriving from generative AI sources spend more time on site, view more pages per visit, and are less likely to bounce. Consumer survey data shows that a significant share of users employ generative AI primarily for research and comparison rather than immediate purchase. Together, these indicators suggest that AI compresses exploratory effort into fewer, more purposeful interactions, reducing uncertainty before consumers commit attention or money.

Closing the Gap Between Research and Purchase

E-commerce friction is most visible at the point of abandonment. Baymard Institute’s synthesis of dozens of studies places the average cart abandonment rate at just over 70 percent. While pricing, shipping costs, and payment issues contribute, abandonment frequently reflects unresolved informational gaps. Consumers hesitate when they are uncertain about suitability, policies, or tradeoffs that remain poorly understood.

AI affects conversion indirectly by relocating cognitive effort earlier in the journey. When consumers resolve questions about durability, compatibility, policy terms, or use constraints during discovery, fewer uncertainties remain at checkout. Adobe’s data indicates that although AI-originating traffic historically converted at lower rates than non-AI traffic, the gap has narrowed materially as discovery systems mature. Revenue per visit from AI-driven traffic has improved substantially over time, reflecting higher purchase readiness.

This pattern aligns with behavioral models of regret aversion. Consumers delay or abandon purchases when they cannot justify decisions to themselves or anticipate post-purchase dissatisfaction. AI-assisted research consolidates fragmented information into a coherent narrative, enabling decisions that feel defensible rather than tentative. The downstream effect extends beyond conversion. Better expectation alignment at purchase can reduce returns, disputes, and dissatisfaction, outcomes that impose costs on both consumers and retailers.

In practice, this shift is visible in how large retailers deploy guided discovery tools in high-consideration categories such as electronics, appliances, and furniture. Rather than maximizing exposure to the full catalog, these systems emphasize constraint satisfaction, surfacing products that meet spatial, compatibility, or maintenance requirements before highlighting marginal feature differences. The objective is confidence rather than speed.

Shift in Consumer Effort Along the E-Commerce Funnel

Source: Adobe Digital Insights; Baymard Institute

Discovery Layers as Behavioral Infrastructure

As AI mediates discovery, it becomes a form of behavioral infrastructure rather than a neutral interface. Discovery layers determine which attributes are emphasized, how alternatives are framed, and when a conversation appears complete. Unlike ranked lists, conversational outputs can imply resolution even when uncertainty remains.

Economic research on consumer search and steering has long demonstrated that information architecture shapes choice paths. In digital markets, steering historically occurred through ranking, placement, and default sorting. AI introduces a subtler mechanism, embedding influence within summaries, comparative explanations, and selective emphasis rather than explicit promotion.

Consumers interpret conversational guidance differently from advertising. Fluent, coherent responses convey authority and completeness, particularly when uncertainty is high. As AI becomes a routine interface for shopping decisions, influence concentrates in platforms that control inclusion, framing, and attribution within discovery layers. Industry analysis suggests that brands may experience meaningful erosion of traditional search traffic as AI-mediated discovery expands.

From a consumer protection perspective, this raises familiar concerns in a new form. Steering can occur not only through what is shown, but through what is omitted. Attributes that are harder to summarize, less standardized, or less commercially prioritized may receive less exposure. The interface may feel more helpful while becoming less contestable, particularly for smaller merchants or differentiated offerings.

AI Discovery Mechanisms and Consumer Protection Considerations

Source: OECD; Federal Trade Commission; European Commission

Trust Calibration in AI Mediated Commerce

Trust in e-commerce has traditionally been anchored in brands, reviews, and platform reputation. AI discovery recalibrates that trust toward systems that mediate information rather than the sources that produce it. Consumers increasingly rely on synthesized outputs instead of primary pages, consistent with broader zero-click behavior observed in digital search.

This shift has practical limits. In commerce, inaccuracies carry direct financial consequences. Misinterpreted policies, incomplete comparisons, or incorrect recommendations translate into returns, disputes, and dissatisfaction. As AI usage expands beyond early adopters, trust becomes conditional on accuracy, consistency, and explainability rather than novelty.

Survey data indicates that generative AI use is moving into the mainstream. As usage becomes habitual, tolerance for error declines. Systems that fail to meet expectations risk disengagement. For merchants, this creates reputational exposure that is indirect but consequential. Brand perception may increasingly be shaped before any direct interaction occurs, mediated by how AI systems describe, compare, or exclude products.

Participation in AI discovery therefore becomes a reputational as well as a commercial consideration. Retailers are affected not only by whether they are surfaced, but by how they are framed.

Strategic Implications for Commerce

AI-aligned discovery rewards clarity, structure, and interpretability. Retailers that provide accurate, well-structured product information are more likely to be interpreted correctly by AI systems. Investment shifts away from narrow keyword optimization toward data quality, transparent policies, and validation signals that machines can surface reliably.

Measurement practices must evolve accordingly. If AI resolves early-stage questions before a click occurs, traditional attribution models undercount the influence of discovery. Industry research indicates that relatively few brands systematically track performance in AI-powered search environments, creating a visibility gap as discovery channels reorganize.

Policy frameworks will shape the boundaries of this shift. The European Union’s Digital Markets Act and related consumer protection initiatives focus on contestability and fair presentation in digital ecosystems. As AI discovery layers become central to commerce, scrutiny of framing, defaults, and selective summarization is likely to intensify. The distinction between helpful guidance and undue steering will increasingly define both platform design and competitive strategy.

Key Takeaways

• AI-driven discovery reduces consumer search costs by translating intent into structured, interpretable comparisons.
• Evidence indicates that AI-mediated research produces more engaged and increasingly purchase-ready traffic.
• A growing share of decision-making occurs before consumers reach merchant sites, shifting strategic value upstream.
• Discovery layers function as behavioral infrastructure, concentrating influence in platforms that control framing and visibility.
• Retail strategy must adapt through data clarity, trust calibration, and updated measurement models as regulatory scrutiny grows.

Sources

• McKinsey & Company; New front door to the internet Winning in the age of AI search; – Link
• McKinsey & Company; The agentic commerce opportunity how AI agents are ushering in a new era for consumers and merchants; – Link
• Adobe Digital Insights; Generative AI Powered Shopping Rises with Traffic to U.S. Retail Sites; – Link
• Baymard Institute; 50 Cart Abandonment Rate Statistics 2025; – Link
• Deloitte; 2025 Connected Consumer Innovation with trust; – Link
• Journal of Marketing Research; Choice overload A conceptual review and meta analysis; – Link
• University of Chicago Press Journals; Consumer Search Steering and Choice Overload; – Link
• Nocke and Rey; Consumer Search and Choice Overload working paper; – Link
• Federal Trade Commission; Bringing Dark Patterns to Light report; – Link
• OECD; Six dark patterns used to manipulate you when shopping online; – Link
• European Commission; Digital Markets Act official portal; – Link