Agentic Search Displaces Brand-Owned Discovery as Commerce's Primary Channel
Executive Summary
AI-referred traffic to commerce properties grew 150–428% every quarter between August 2025 and May 2026, while traditional search referral fell 15% and brand-property discovery fell 7% — across 1.5 billion shoppers. The discovery layer has already migrated. Brands whose product catalogs, pricing logic, and promotional structures are not machine-readable are invisible in the channel now capturing the fastest-growing share of purchase intent. Audit your data infrastructure against agent-readability criteria before Q1 2027 budget locks.
The Signal
Salesforce's Fourth Edition State of Commerce report — drawn from surveys of 3,450 commerce professionals and 4,690 consumers, plus behavioral data from 1.5 billion shoppers — shows agentic search growing 200% year over year as the first step in the purchase journey. AI-referred traffic grew between 150% and 428% every quarter measured, while discovery through brand-owned properties fell 7% and traditional search fell 15% between August 2025 and May 2026. Despite the consumer shift, only 28% of commerce organizations currently deploy agentic AI, though 44% of non-adopters plan to within six months. Among current adopters, 35% are already scaling across functions — only 4% remain in pilot. Talon.One, via a sponsored Modern Retail guide, is positioning its loyalty and promotions infrastructure as agent-ready, citing Bain & Company projections that agentic AI will account for 15–25% of total U.S. e-commerce by 2030.
What Changed
AI assistants now function as the primary discovery layer between consumer intent and brand exposure — operating before a shopper reaches any brand-owned property. Commerce teams can instrument product catalogs, promotional logic, and loyalty structures to be machine-readable and agent-surfaceable, making products retrievable by AI agents conducting end-to-end shopping on a consumer's behalf. Discovery, evaluation, and purchase initiation are now automatable at the agent layer, entirely outside brand-controlled environments.
Why It Matters
The structural advantage in commerce is shifting from those who own the most traffic to those whose products are most legible to machines. That is the commercial core of what the Salesforce data is showing, and it has immediate budget implications. What becomes newly possible: brands that instrument their product catalogs, promotional logic, and loyalty structures for machine readability can now participate in purchase decisions they would previously have been excluded from entirely — the AI-mediated consideration set that forms before a shopper ever opens a browser tab. That is not a marginal improvement in top-of-funnel efficiency; it is access to a discovery layer that did not exist two years ago. What is being devalued faster than most commerce teams appreciate: investment in brand-owned traffic acquisition. A 7% decline in brand-property discovery and a 15% drop in traditional search referral, measured across 1.5 billion shoppers, is not a rounding error. It is the early signal of structural demand migration. Brands that have been compounding investment in SEO, paid search, and owned-channel optimization are watching the returns on that infrastructure erode in real time. The new competitive pressure is data-layer quality, not content volume. Brands with fragmented product data, inconsistent pricing logic, or promotional structures that cannot be queried by an API will simply fail to surface in agent-mediated journeys — not because their products are inferior, but because their infrastructure is unreadable. The Bain projection of 15–25% of U.S. e-commerce flowing through agentic channels by 2030 is the destination; the question is which brands have negotiated the on-ramp before that volume concentrates and the cost of entry rises sharply.
Marketing Impact
ecommerce
Product catalog management becomes a machine-readability problem, not a merchandising one. Teams that cannot expose structured product data, real-time pricing, and promotional logic via queryable APIs will simply fail to appear in agent-mediated consideration sets — regardless of how strong the underlying product is.
media
Paid search and owned-channel acquisition budgets are being systematically devalued by a channel that does not accept bids. Media teams need to reallocate toward GEO instrumentation and agent-feed optimization, and begin building measurement frameworks that capture AI-referred conversion, which current attribution stacks largely cannot see.
crm
Loyalty and promotions infrastructure must be rebuilt to be agent-readable. If an AI shopping agent cannot query a brand's loyalty status, eligible offers, or member pricing in real time, those incentives are invisible at the moment of decision — eliminating the retention advantage CRM investment was meant to create.
marketing ops
The 27% of organizations with fully unified customer data are now structurally better positioned for agentic commerce than the 73% who are not. Marketing ops teams face a forcing function: data unification is no longer an efficiency project — it is the prerequisite for participating in the fastest-growing discovery channel.
The Exploit
Opportunity
Commerce teams that instrument product catalogs, pricing logic, and promotional structures to be machine-readable and API-queryable can enter the AI-agent consideration set before competitors do. The window is Q3–Q4 2026: 44% of non-adopters plan deployment within six months, meaning the field compresses sharply by early 2027. Brands with clean data layers capture disproportionate agent-referred volume while acquisition costs remain low.
Risk
Agent ranking logic is opaque and unstable; optimising for today's retrieval criteria may require costly re-instrumentation as models and agent frameworks evolve. Over-indexing on agent channels before attribution is mature also creates measurement blind spots that complicate board-level ROI reporting.
The Move
By October 2026, the VP of Commerce Technology should audit the full product catalog against agent-readability criteria — structured data completeness, API queryability, real-time pricing and promotional logic exposure — and close the top three gaps before Q1 2027 planning locks budget. Success checkpoint: agent-referred sessions measurable and trending above 5% of total discovery traffic within 90 days of instrumentation.
First-Mover Advantage
Gains
Early adopters occupy agent consideration sets before volume concentrates and agent-platform negotiating leverage shifts to incumbents — the same dynamic that rewarded early Google Shopping adopters with structurally lower CPCs before the market priced in competition.
Risks
Agent ranking logic is opaque and unstable; optimising for today's retrieval criteria may require costly re-instrumentation as models and agent frameworks evolve. Over-indexing on agent channels before attribution is mature also creates measurement blind spots that complicate board-level ROI reporting.
Window
The advantage window runs approximately 18 months — until mid-2028, when agent-readiness tooling commoditises and major platforms standardise catalog ingestion requirements. The closing signal is when Shopify and Adobe Commerce ship native agent-optimisation modules at scale.
Winners & Losers
Winners↑
Commerce teams with machine-readable product catalogs and structured data infrastructure
As AI agents become the primary discovery layer, brands whose product attributes, pricing logic, and availability data are structured, queryable, and feed-ready will surface in agent-mediated consideration sets that competitors with legacy catalog infrastructure will simply miss. The Salesforce data confirms the mechanism: AI-referred traffic grew 150–428% quarter over quarter while brand-property discovery fell 7% — meaning the addressable market is already migrating to a channel gated by data quality, not ad spend. The immediate move is a catalog audit against the query patterns AI assistants actually use, prioritising natural-language attribute completeness over visual merchandising.
Loyalty and promotions platform vendors with agent-ready API infrastructure
When an AI shopping agent executes a purchase on a consumer's behalf, it will factor in available promotions, loyalty redemptions, and personalised offers — but only if those mechanics are exposed via clean, real-time APIs the agent can query mid-session. Vendors like Talon.One that have pre-positioned their infrastructure as agent-compatible gain an urgent sales narrative: the alternative for brands is having their promotional economics invisible at the exact moment purchase intent is highest. Commerce leaders should pressure-test whether their current loyalty and promotions stack can surface offer logic to an external AI agent without a human intermediary.
Commerce organisations that have already unified customer data across sales, marketing, and service
Only 27% of organisations report fully unified customer data, and the Salesforce data shows those that have unified it report measurably better AI outcomes, improved customer retention, and cross-functional alignment — the precise capabilities required to personalise agent-mediated journeys at scale. Fragmented data doesn't just slow AI deployment; it makes personalisation impossible when an agent is operating autonomously at the point of purchase without a human review loop. Organisations that completed data unification ahead of the agentic wave are now structurally positioned to deploy contextual, account-level offers into agent channels while competitors are still solving identity resolution.
Losers↓
Performance marketing teams over-indexed on paid search and brand-owned traffic acquisition
A 15% decline in traditional search referral and a 7% drop in brand-property discovery, measured across 1.5 billion shoppers between August 2025 and May 2026, is a direct return-on-investment signal for teams whose budget is concentrated in the channels now losing structural share. The mechanism is substitution, not supplementation: consumers are redirecting discovery intent to AI assistants, and that channel is not bought through auction-based media. The defensive move is to reweight measurement frameworks to capture AI-referred conversion separately and make the case internally for reallocating a defined share of paid search budget into catalog and data infrastructure before the volume concentration makes entry prohibitively expensive.
Retailers with inconsistent pricing, fragmented promotional logic, and unsynchronised inventory across channels
The Salesforce report identifies inconsistent pricing and promotions (41% of multichannel organisations) and unsynchronised real-time inventory (40%) as the most prevalent omnichannel failure points — and both are precisely the data fields an AI shopping agent queries before committing to a purchase recommendation. A brand whose promotional data conflicts across endpoints, or whose inventory state is stale by the time an agent reads it, will either be excluded from agent recommendations or generate a failed transaction that degrades its agent-channel reputation. The structural fix requires investment in real-time data synchronisation infrastructure that was previously a customer-experience priority but is now a channel-access prerequisite.
Strategic Outlook
The adoption curve here follows a familiar but compressed pattern: a 200% year-over-year growth rate in agentic search, combined with 44% of non-adopters planning deployment within six months, means the window between early-mover advantage and table-stakes parity is collapsing faster than most planning cycles can accommodate. By Q2 2027, agent-mediated discovery will be a standard line item in commerce infrastructure budgets, not a test-and-learn allocation. The Bain projection of 15–25% of U.S. e-commerce flowing through agentic channels by 2030 implies a volume concentration event: brands that have established machine-readable catalog and promotions infrastructure early will compound that advantage as agent traffic scales, while late entrants face both higher integration costs and a discoverability gap that widens with every quarter of delay. The structural analogy is mobile commerce circa 2012 — the brands that treated it as a channel to instrument, not just a trend to monitor, captured disproportionate share before the window closed.