On this page
- How it works
- AI Mode vs AI Overviews
- Shopping Graph integration
- Agentic checkout and Universal Commerce Protocol
- Universal Cart
- Merchant requirements and optimisation
- Google Merchant Center
- Structured data
- Business Agent and Direct Offers
- Traffic and SEO impact
- Zero-click and click-through rate
- Ecommerce-specific traffic impact
- Search Agent trajectory
- Key terms
- What practitioners report
- Benchmarks (as-of 2026-05-19 unless otherwise noted)
- Frontier links
AI Mode (Google)
AI Mode (Google)
AI Mode is Google's dedicated conversational search workspace, powered by Gemini, that lets users have multi-turn, follow-up conversations with Google Search rather than issuing discrete keyword queries. Launched in the United States on May 20, 2025 at Google I/O, it represents a fundamentally different interaction paradigm from both traditional search and AI Overviews: users must actively opt in by selecting the AI Mode tab, and within it they receive Gemini-generated conversational answers backed by live Shopping Graph data and real-time web signals. For ecommerce, AI Mode is the primary surface on which agentic checkout, the Universal Commerce Protocol (UCP), and the Universal Cart operate — making it the most consequential Google product launch for retail since the introduction of Shopping ads.
How it works
AI Mode runs on Gemini (Gemini 3.5 Flash as the default model as of Google I/O 2026) and handles queries that are, on average, three times longer than traditional Google searches — reflecting the kind of complex, considered questions users bring to it. (Google Blog, I/O 2026, 2026-05-19, as-of 2026-05-19)
Unlike AI Overviews, which surface automatically within the standard SERP for informational queries, AI Mode requires an explicit tab selection. Within the tab, users can continue a conversation across multiple turns — asking follow-up questions, narrowing down product options, checking local stock availability, and initiating checkout — without returning to a keyword search. (Cyberclick, 2026; Evertune, 2026)
AI Mode pulls live signals — including real-time Shopping Graph data — whereas AI Overviews are generated from Gemini's pre-indexed crawl. This live-data distinction is what enables AI Mode to surface current prices, real-time stock levels, and nearby store availability within its interface.
[!unverified] The claim that AI Mode pulls live Shopping Graph data while AI Overviews use pre-indexed crawl is reported by third-party sources (niara.ai, 2026) but has not been confirmed verbatim in primary Google documentation in this harvest pass. Treat as plausible but unverified at source level.
By May 2026, AI Mode had surpassed 1 billion monthly users, with queries more than doubling every quarter since launch. (Google Blog — Search I/O 2026 update, 2026-05-19, as-of 2026-05-19)
AI Mode vs AI Overviews
| Dimension | AI Overviews | AI Mode |
|---|---|---|
| Entry point | Automatic, within standard SERP | Opt-in tab selection |
| Interaction model | One-shot answer in SERP | Multi-turn conversation |
| Checkout capability | None | Yes — agentic checkout via Universal Commerce Protocol (UCP) |
| Data freshness | Pre-indexed Gemini crawl | Live signals including Shopping Graph |
| Ad formats | Sponsored units in SERP | New Gemini-powered conversational ad formats |
| Appearance frequency | 48% of all queries; 14% of shopping queries (Ahrefs / Search Engine Land, as-of March 2026) | Opt-in only — frequency not disclosed |
Shopping Graph integration
The Shopping Graph is the product knowledge graph that powers AI Mode's product discovery. Google describes it as containing more than 50 billion product listings, with more than 2 billion refreshed every hour. (Google Blog — I/O 2026 Shopping, 2026-05-19, as-of 2026-05-19)
AI Mode shopping flows combine Gemini models with this Shopping Graph to move shoppers from natural-language questions to purchase decisions across a series of back-and-forth prompts. The interface surfaces product panels with images, prices, and reviews embedded directly in the AI-generated answer. (Google Blog — I/O 2026, 2026-05-19; Think with Google, 2026)
Products not registered in Google Merchant Center cannot enter the Shopping Graph and therefore have no path to appearing in AI Mode product surfaces. This is the primary merchant entry requirement — not structured data markup (see Contradiction 1 below). (paz.ai, 2026)
AI Mode also supports Virtual Try-On: shoppers can upload a photo and virtually try on apparel from the billions of listings in the Shopping Graph. First announced at Google I/O 2025 and subsequently expanded into AI Mode. (Google Blog — Shopping AI Mode + Virtual Try-On, 2025-05-20)
Agentic checkout and Universal Commerce Protocol
The defining capability separating AI Mode from all prior Google search surfaces is agentic checkout: the ability for a Google agent to initiate and complete a purchase on the shopper's behalf, using Google Pay, with the shopper's explicit confirmation required at each step. (Google Blog — Agentic Checkout, 2025; Google Blog — NRF 2026, 2026-01-11)
The infrastructure enabling this is the Universal Commerce Protocol (UCP), announced at the National Retail Federation 2026 conference (January 11, 2026) as an open-source standard for agentic commerce. UCP establishes functional primitives enabling seamless commerce journeys between consumer surfaces, businesses, and payment providers — covering discovery, buying, and post-purchase support. (Google Blog — NRF 2026, 2026-01-11; Google Developers Blog — UCP technical, 2026)
UCP: open-source vs open standard. The Google Developers Blog technical post describes UCP as "open-source." Marketing blog posts and trade press consistently describe it as an "open standard." The distinction matters for merchant and developer implementation: if open-source, the specification and reference implementation are available for inspection and contribution; if merely an open standard, the spec may be licensed differently. Source A: Google Developers Blog — UCP technical. Source B: Google Blog — Agentic Shopping NRF 2026.
Alongside UCP, Google introduced the Agent Payments Protocol (AP2) as a payment-layer building block, and the Universal Cart as the shopper-facing aggregation layer. (Google Developers Blog — UCP, 2026)
Universal Cart
The Universal Cart, announced at Google I/O 2026 (May 19, 2026), is described as "an intelligent shopping cart and new hub for shopping on Google." Items added to the cart in any Google surface — Search, Gemini app, YouTube, or Gmail — are synchronised across all surfaces. The cart also runs background functions: tracking price drops, monitoring price history, sending back-in-stock alerts, and flagging incompatible products proactively. (Google Blog — I/O 2026 Shopping, 2026-05-19, as-of 2026-05-19)
UCP expansion milestones (as-of 2026-05, Google Marketing Live 2026):
- January 2026 (NRF): Launch with select U.S. merchants — Nike, Sephora, Target, Walmart, Wayfair, Ulta Beauty, Fenty, Steve Madden
- May 2026 (Google Marketing Live / I/O): Expansion to Canada and Australia confirmed; UK "in the coming months"
- May 2026: UCP announced for YouTube (U.S.); expansion to hotel booking and local food delivery ("coming soon")
PayPal confirmed support for Google's AI Checkout in 2026, enabling PayPal-credentialed purchases within Google's agentic commerce flows. (PayPal IR filing title, 2026 — PDF not fully fetched; treat as unverified pending primary confirmation)
[!unverified] Nike's June 2026 integration with Google's agentic commerce stack — described as a multi-item Universal Cart experience inside the Gemini app and AI Mode tied to FIFA World Cup 2026 — is reported by stellagent.ai (2026) but no primary Google or Nike source was found to confirm it in this harvest pass.
Merchant requirements and optimisation
Google Merchant Center
Presence in Google Merchant Center is the threshold requirement for any product to enter the Shopping Graph and therefore appear in AI Mode product surfaces. There is no confirmed structured data requirement beyond what standard Merchant Center feeds already expect. (paz.ai, 2026; Google Search Central AI optimization guide, updated 2026-07-10)
New Merchant Center features announced at Google Marketing Live 2026 (as-of 2026-05):
- Conversational Attributes: six optional data fields (exact names not confirmed in primary source in this pass) that complement existing product feeds and help AI systems "understand product nuances" for conversational shopping behavior. Added via supplemental feeds or Merchant API. (Search Engine Land, 2026-05)
- AI Performance Insights: a new reporting feature giving merchants visibility into their share-of-voice across AI Mode, AI Overviews, and the Gemini app, with competitor comparison. Rolling out to Australia, Canada, India, New Zealand, and the U.S. from May 27, 2026. (Search Engine Land, 2026-05-27, as-of 2026-05-27)
- Ask Advisor: an AI collaborator inside Merchant Center that shares goal-tailored insights and connects across Google Ads and Google Analytics. (Google Blog — Marketing Live 2026, 2026-05)
Structured data
Structured data requirement — official vs practitioner positions. Multiple third-party practitioners (ppc.land, paz.ai, productsup.com, 2026) frame schema.org Product markup and specific Merchant Center attributes as near-requirements for AI Mode visibility. Google's own Search Central AI Optimization Guide (updated 2026-07-10) states explicitly: "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add" — while still recommending existing structured data "as part of your overall SEO strategy" for rich result eligibility. Source A (requirement framing): ppc.land — 8 Merchant Center attributes for AI Mode. Source B (official guide): Google Search Central AI Optimization Guide.
The Search Console Generative AI performance report (launched August 31, 2026 per AI Overviews concept page) provides a first-party measurement surface for AI Mode and AI Overviews impressions and clicks, separate from traditional organic. (Google Search Central, 2026-07-10, as-of 2026-07-10)
Business Agent and Direct Offers
Business Agent allows brands to have a conversational AI representative active directly within Search and the Gemini app — described as "like a virtual sales associate" — where shoppers can ask product questions, compare options, and receive brand-level guidance. (Google Blog — NRF 2026, 2026-01-11)
Direct Offers in Google Ads let retailers inject exclusive discounts directly into AI Mode recommendation flows, surfacing promotions to shoppers "ready to buy" without requiring a click-through. (Google Blog — NRF 2026, 2026-01-11)
Gemini-powered conversational ad formats inside AI Mode were announced at Google Marketing Live 2026, described as "contextual and interactive" with embedded AI-generated explainers. Separate "Direct Offers" Shorts format also surfaced in practitioner trade press (YouTube, 2026). (Google Blog — Marketing Live 2026, 2026-05; ppc.land, 2026)
Traffic and SEO impact
All traffic impact figures below carry high stale-risk or source-quality caveats. Read notes carefully before citing.
Zero-click and click-through rate
AI Mode sessions: 93% of AI Mode sessions end without a single click to an external website, according to Semrush data cited by third-party sources as of September 2025.
This figure (93% zero-click) is from September 2025 Semrush data, cited via pasqualepillitteri.it (2026) — not a direct Semrush primary source. As-of 2025-09. Potentially superseded by 2026 data.
AI Overviews click reduction: A field study cited by Search Engine Journal (2026) found AI Overviews reduced outbound organic clicks by 38% on queries where they appeared, with Zero-Click Search rising from 54% to 72%. (Search Engine Journal, 2026, as-of 2026)
Zero-click scope — AI Mode vs AI Overviews: The 93% zero-click figure is attributed to "AI Mode sessions" (Semrush, Sept 2025, via pasqualepillitteri.it); the 38% click-reduction figure applies to "AI Overviews on triggered queries" (SEJ, 2026). These measure different surfaces and different baselines. They are frequently conflated in trade press as a single "Google AI" traffic impact figure. Source A (93% AI Mode): pasqualepillitteri.it/en/news/811/. Source B (38% AI Overviews): searchenginejournal.com/ai-overviews-cut-organic-clicks-38-field-study-finds/573145/.
CTR drop magnitude from AI Overviews — conflicting measurements: Ahrefs reports a 58% CTR drop for top-ranking pages when AI Overviews appear. Semrush-sourced figures describe a 30–50% average reduction across positions 1–3. Both are 2026 figures; methodologies differ (GSC data vs SERP-level analysis). Source A (58%): omnibound.ai/blog/ai-seo-statistics (citing Ahrefs). Source B (30–50%): quickseo.ai/blog/google-ai-overviews-statistics-2026-60-data-points (citing Semrush).
Ecommerce-specific traffic impact
Ecommerce and retail lost an estimated 11.6% of organic traffic due to AI Mode and AI Overviews — framed as among the least-affected industries because transactional queries still require a click to convert. (First Page Sage 2026 report, as-of 2026)
AI-driven referral traffic to U.S. retail websites increased 393% year-over-year in Q1 2026, suggesting that while total organic volume declined, AI-referred visitors showed significantly different (higher-value) behaviour. (First Page Sage 2026 report, as-of Q1 2026)
Separately, brands cited inside an AI Overviews|AI Overview earn 35% more organic clicks and 91% more paid clicks than uncited brands on the same SERP — suggesting citation inside AI surfaces, rather than traditional ranking position, is the primary visibility goal in the AI search era. (Semrush data cited by quickseo.ai, 2026 — treat as third-party aggregation)
The 35%/91% citation lift figures are from a third-party aggregator (quickseo.ai) citing Semrush. The primary Semrush research URL was not verified in this pass. As-of 2026 but source quality unverified.
Search Agent trajectory
BrightEdge data (2026) reports that AI agent requests have reached 88% of human organic search volume — described as a "tipping point" where most online customers will effectively be AI agents by end of 2026. (BrightEdge press release, 2026, as-of 2026)
BrightEdge is a vendor with a conflict of interest on AI search severity claims. The 88% agent-to-human traffic ratio is from a press release (open access) — the underlying dataset is likely gated. Treat with caution.
Key terms
| Term | Meaning |
|---|---|
| AI Mode | Google's opt-in conversational search workspace, powered by Gemini |
| AI Overviews | Automatic Gemini-generated summaries appearing within the standard SERP (separate from AI Mode) |
| Shopping Graph | Google's product knowledge graph — 50B+ listings, 2B updated/hour — that powers AI Mode shopping results |
| Universal Commerce Protocol (UCP) | Open-source/open-standard (contested) agentic commerce infrastructure enabling checkout across Google surfaces |
| Universal Cart | Cross-surface shopping cart (Search, Gemini, YouTube, Gmail) with background deal-monitoring |
| Agentic checkout | Ability for a Google AI agent to complete a purchase on a shopper's behalf with explicit confirmation |
| Business Agent | Brand's AI-powered virtual sales associate inside Search/Gemini app |
| Direct Offers | Merchant-configured exclusive discounts injected into AI Mode recommendation flows |
| Conversational Attributes | Optional Merchant Center feed fields (six, as-of May 2026) for optimizing product listings for conversational AI |
| AI Performance Insights | New Merchant Center report tracking brand share-of-voice across AI surfaces |
| Agent Payments Protocol (AP2) | Payment-layer building block alongside UCP |
| AEO / GEO | Answer Engine Optimisation / Generative Engine Optimisation — practitioner disciplines for ranking inside AI surfaces |
What practitioners report
Practitioners (YouTube metadata, trade press, 2026) have converged on two parallel optimisation disciplines for AI Mode visibility — AEO (Answer Engine Optimisation) and GEO (Generative Engine Optimisation) — distinct from traditional SEO. The core practitioner shift is that citation inside an AI-generated answer, not ranking position 1 on a traditional SERP, is the primary visibility goal.
Specific merchant strategy signals (from Shopify and SEJ, 2026):
- Ensure Google Merchant Center feed is clean, complete, and structured for conversational queries
- Use the new Conversational Attributes fields once available in Merchant Center
- Monitor AI Performance Insights share-of-voice vs competitors
- Register for Direct Offers if eligible (Google Ads account required)
- Prioritise the Search Console Generative AI performance report over standard organic reporting
AI Mode's impact on Shopify merchants is covered separately in YouTube metadata-only content (channel: unknown, URL: youtube.com/watch?v=i3BqEp3HG9I, published: 2026 approx) — transcript unavailable (Apify unavailable in this session).
Benchmarks (as-of 2026-05-19 unless otherwise noted)
| Metric | Value | Source | Date |
|---|---|---|---|
| Monthly users | >1 billion | Google Blog — I/O 2026 | 2026-05-19 |
| Query length vs standard search | 3× longer | Google Blog — I/O 2026 | 2026-05-19 |
| Query volume growth | Doubling every quarter | Google Blog — I/O 2026 | 2026-05-19 |
| Shopping Graph size | 50B+ listings | Google Blog — I/O 2026 | 2026-05-19 |
| Shopping Graph refresh rate | 2B listings/hour | Google Blog — I/O 2026 | 2026-05-19 |
| AI Mode zero-click rate | 93% of sessions | Semrush via pasqualepillitteri.it | 2025-09 ⚠ stale-risk |
| AI Overviews on all queries | 48% | Ahrefs via eseospace.com | 2026-03 |
| AI Overviews on shopping queries | 14% | Search Engine Land | 2026 early |
| Organic traffic decline (ecommerce) | −11.6% | First Page Sage 2026 | 2026 |
| AI-referred retail traffic growth | +393% YoY Q1 | First Page Sage 2026 | Q1 2026 |
| AI agent requests vs human search | ~88% | BrightEdge (vendor) | 2026 ⚠ conflict-of-interest |
Frontier links
Zero-Click Search · Shopping Graph · Google Merchant Center · Direct Offers (Google) · Business Agent (Google) · Universal Cart (Google) · Gemini (Google) · AI Max (Google) · Search Agents (Google) · AEO (Answer Engine Optimisation) · GEO (Generative Engine Optimisation) · Agent Payments Protocol (AP2)