Zero-click commerce: 73% of shoppers already use AI to buy
What is zero-click commerce?
Zero-click commerce refers to a model where l’achat se finalise directement dans l’interface d’un assistant IA, without the consumer visiting the merchant site. The user formulates a need in natural language (« I’m looking for women’s running sneakers, size 39, max budget €120 »), and the agent manages the entire process: search, comparison, selection, payment.
The concept emerges from a convergence between two forces:
- Les zero-click searches: for several years, a growing share of Google searches have been resolved directly in the SERP (featured snippets, knowledge panels). The consumer gets their answer without clicking.
- Les agents IA transactionnels : les LLM (ChatGPT, Copilot, Perplexity) ajoutent désorMays des capacités d’achat intégrées. Le « zero-click » passe du search au commerce.
The consequence for brands is direct: the website remains essential as a source of data, Mays it ceases to be the sole place of the transaction. The catalog must become readable by AI agents, structured to be interpreted and recommended in interfaces where the huMayn buyer no longer visits the product sheet.
This transformation is already operational. According to Commercetools et Stripe (presented at Shoptalk 2026), agentic commerce could redirect between 3 et 5 trillions de dollars of retail spending by 2030, including nearly 1 trillion de dollars from the United States alone.
Why do 73% of shoppers already use AI?
Le chiffre est frappant : 73% des consommateurs en ligne use at least one generative AI tool during their purchasing journey (Digital Commerce 360 / eMarketer). Adoption is massive, and it is spread across the entire funnel.
Les 4 usages dominants
Public data from Digital Commerce 360 reveals clear segmentation:
- 45% utilisent AI pour generate product ideas — « Quel cadeau pour un adolescent qui fait du skate ? »
- 37% s’en servent pour summarize customer reviews — instead of reading 200 reviews, AI produces a structured summary
- 32% l’utilisent pour comparer les prix between several brands in a single request
- 28% check the availability and delivery times via AI
These uses share one thing in common: the buyer delegates tasks to AI that previously took between 15 and 45 minutes. The perceived value is saving decision-making time.
From assisted research to delegated purchasing
Aujourd’hui, 13% of consumers have already completed a purchase via an AI assistant (Digital Commerce 360). Ce chiffre semble modeste, Mays la dynamique sous-jacente est puissante : 70% des consommateurs say they are comfortable with the idea of an AI agent making a purchase on their behalf (Invisibletech, 2026).
The gap between intention (70%) and action (13%) represents the window of opportunity for prepared brands. As soon as technical barriers (payment integration, confidence in returns) are lifted, transactional adoption accelerates.
2026: the pivotal year
Selon eMarketer, 1 acheteur sur 4 utilisera des chatbots IA in its shopping journey in 2026. The transition from assisted search to assisted commerce is underway. Brands that structure their catalog for AI agents are taking a position in a growing acquisition channel.
Comment fonctionne un achat sans clic ?
The zero-click purchasing journey is broken down into 5 steps, all orchestrated by the AI agent in a single conversational interface.
Step 1: Formulation of the need
The user expresses his request in natural language. The precision of the prompt determines the quality of the results: “Bluetooth headphones with noise reduction, battery life 30 hours minimum, less than €200”. The agent interprets the constraints and transforms them into filtering criteria.
Step 2: Crawl and Extract Product Data
L’agent interroge plusieurs sources : structured product flows (Schema.org Product, Google Merchant Center), API marchandes (Shopify Storefront API, WooCommerce REST API) et indexed web content. Structured data is preferred because it allows direct reasoning on attributes (price, size, availability, delivery time).
Step 3: Comparison and Ranking
The agent classifies the products according to the user’s criteria. It weights the technical attributes, the preuve sociale (AggregateRating, nombre d’avis), les conditions commerciales (livraison gratuite, politique de retour) et la consistency of information between sources. A product whose attributes are complete and standardized obtains a better ranking than a competitor with fragmented data.
Step 4: Recommendation and validation
The agent presents a selection (generally 3 to 5 products) with a comparative summary. The user validates or refines their request. The conversation is iterative: “Rather black or white? », “With charging case included? « .
Step 5: Transaction
On platforms that integrate payment (Copilot Checkout with Stripe and PayPal), the user confirms the purchase in the conversational interface. The merchant site delivers the order, Mays the commercial point of contact has moved to the AI agent.
This diagram transforms the role of the e-commerce site: it becomes a structured data provider for agents, more than a showcase for huMayns browsers. The two functions coexist, Mays the share of trade mediated by AI grows every quarter.
Which platforms already offer AI agent purchases?
Zero-click commerce is operational on several major platforms. Here is the state of play in April 2026.
Integrated directly into the ChatGPT interface. The user describes what he is looking for, the agent displays product recommendations with prix, images, avis et liens d’achat. Active on Etsy and hundreds of marketplaces. The model is based on structured merchant data and product feeds indexed by GPTBot.
The most advanced integration yet. Copilot Checkout connects Shopify, PayPal et Stripe to allow the complete purchase in the Copilot interface: product selection, choice of variant, secure payment and order confirmation. The user finalizes their purchase in 3 to 4 conversational exchanges.
Augmented comparator which summarizes product sheets, customer reviews and delivery conditions in a single answer. Perplexity cites its sources and allows the user to click through to the merchant to complete the purchase. The model is moving towards direct transactional integration.
Google integrates its AI Overviews to Shopping results. The Shopping Graph (35 billion product listings) powers generative responses with real-time transactional data: price, stock, reviews, delivery options. The Shopping panel appears directly in the AI response.
Secondary platforms are also emerging: Amazon Rufus (AI assistant integrated into the Amazon app), the AI chatbots of Klarna et Shopify Sidekick for merchants. The ecosystem is structured around a principle: whoever provides the best product data wins the recommendation.
Ce que regardent les agents pour choisir un produit
AI agents use measurable criteria to decide which product to recommend:
- Completeness of Schema.org Product markup : name, description, brand, sku, gtin, image, category
- Transactional data in Offer schema : price, priceCurrency, availability, deliveryLeadTime, shippingDetails
- Preuve sociale : AggregateRating avec ratingValue, reviewCount, bestRating
- Consistency of sources : un produit dont les données concordent entre le site, le flux Google Merchant et les marketplaces inspire davantage de confiance à the agent
- Vitesse de crawl : un site rapide (LCP < 2,5 s) est moins coûteux à analyser pour the agent
Comment adapter votre boutique au zero-click commerce ?
Adapter votre boutique au zero-click commerce repose sur un principe : make your catalog readable, structured and complete for AI agents. Voici les 5 actions prioritaires.
Chaque fiche produit doit comporter un balisage Product complet (name, description, image, brand, sku, gtin, category) et un balisage Offer detailed (price, priceCurrency, availability, deliveryLeadTime, shippingDetails, hasMerchantReturnPolicy). Agents use these fields to objectively compare your product to the competition. A product with full markup is confidently recommended by the agent.
Le fichier llms.txt at the root of your site tells LLMs which pages have priority, how to interpret your catalog and what structured data is available. It supplements robots.txt with specific guidelines for AI agents. The main AI crawlers (GPTBot, ClaudeBot, PerplexityBot) read this file.
A daily refreshed Google Merchant Center / Meta Commerce Manager feed ensures that agents have access to prices, stocks and delivery conditions in real time. The agents cross-reference the data from the flow with that of the site. A discrepancy (different price, obsolete stock) degrades the agent’s confidence in your catalog.
FAQs tagged in FAQPage schema are the format most cited by LLMs in their responses. Each product sheet should contain 3 to 5 questions and answers targeting frequently asked questions: size, compatibility, maintenance, warranty. The agent extracts these FAQs to directly answer the user’s questions.
A fast site is a site that agents crawl as a priority. Target a LCP less than 2.5 seconds, a CLS less than 0.1 and an INP less than 200 ms. The performance benefits both Google SEO and the accessibility of your catalog by AI agents.
Exemple de structure llms.txt pour un site e-commerce :
> Catalogue de 1 800 produits mode femme
## Structured data
– Schema.org Product sur chaque fiche
– Schema.org Offer avec prix, stock, livraison
– FAQPage sur les fiches les plus vendues
## Flux produit
– /feed/google-shopping.xml : flux quotidien
– /sitemap-products.xml : sitemap produits
## Main categories
– /categorie/robes/
– /categorie/chaussures/
– /categorie/accessoires/
What impact on your acquisition strategy?
Zero-click commerce changes the acquisition equation at three levels: the channel, the metric and the required skill.
A new acquisition channel to integrate
Until now, e-commerce acquisition channels focused on Google (SEO + SEA), social networks and eMayl. Zero-click commerce adds an additional channel: les interfaces conversationnelles IA. This channel is neither classic search, nor social, nor paid. It is a channel where the quality of structured data determines visibility.
Brands that invest in this data-driven channel are gaining a measurable lead. According to Invisibletech, les boutiques avec un balisage Schema.org complet sont recommended 2 to 3 times more often par les agents IA que celles avec un balisage partiel.
New KPIs to follow
Classic metrics (organic traffic, click-through rate, on-site conversion rate) remain relevant but insufficient. Zero-click commerce requires measuring:
- La couverture Schema.org : pourcentage de fiches produit avec un balisage Product + Offer complet
- Le taux de citation IA: frequency with which your brand appears in LLM responses (measurable via tools like Otterly.ai or systematic prompts)
- AI referral traffic : visites provenant de ChatGPT, Copilot, Perplexity (identifiables dans les logs serveur via les user-agents GPTBot, CopilotBot, PerplexityBot)
- Le score agent-readiness: completeness of structured data on the 5 axes (intelligibility, attributes, information, categorization, transaction)
SEO et GEO convergent
The good news: optimize for zero-click commerce, this is optimize for Google at the same time. Enriched structured data improves rich snippets in SERPs. Expert content powers AI Overviews. Technical performance benefits Core Web Vitals. The strategy is not a choice between SEO and GEO: this is a unified strategy.
L’avantage du premier entrant
Agentic commerce is in its infancy. Structured catalogs are still rare. According to the analyzes of Shopify et Search Engine Land (2026), the majority of e-commerce sites exploit less than half of the potential of their structured data. Product attributes and transactional data are the two most under-exploited areas.
The brands that structure their Mayntenant catalog are taking a position in a market where 1 acheteur sur 4 utilisera des chatbots IA in his shopping journey in 2026 (eMarketer). The first-mover advantage is real: once an agent recommends a product, they tend to recommend it again to subsequent users, creating a virtuous cycle of visibility.
Prepare your catalog for zero-click commerce
Free audit of your store: Schema.org markup, agent-readiness score, concrete recommendations. Results in 48 hours.
Book a free auditFrequently asked questions about zero-click commerce
Qu’est-ce que le zero-click commerce exactement ?
Zero-click commerce refers to a model where the purchase is finalized directly in the interface of an AI assistant (ChatGPT, Copilot, Perplexity), without the consumer visiting the merchant site. The AI agent searches, compares and purchases on behalf of the user.
How many consumers are already using AI to shop?
73% of online consumers use at least one generative AI tool during their purchasing journey, according to Digital Commerce 360 and eMarketer. 45% use it to find product ideas, 37% to summarize opinions, 32% to compare prices.
Which platforms already offer purchasing via AI agent?
ChatGPT Shopping is live on Etsy and hundreds of marketplaces. Microsoft Copilot Checkout integrates Shopify, PayPal and Stripe. Perplexity Shopping offers an enhanced comparator with a summary of reviews and delivery conditions. Google integrates AI Overviews into its Shopping Graph.
Le zero-click commerce remplace-t-il le SEO classique ?
Classic SEO remains essential for visibility in Google. Zero-click commerce adds an additional channel. The two strategies converge: structured data, expert content and technical performance benefit SEO as much as AI agents.
Comment rendre ma boutique compatible avec les agents IA ?
Five priority actions: enrich the Schema.org Product and Offer markup, create an llms.txt file, publish a structured product feed, optimize the Core Web Vitals and set up product FAQs tagged in FAQPage schema.
What is the market potential of agentic commerce?
According to estimates from Commercetools and Stripe (Shoptalk 2026), agentic commerce could redirect between $3 and $5 trillion in retail spending by 2030, including nearly $1 trillion from the United States alone.
Les agents IA peuvent-ils vraiment finaliser un paiement ?
Yes. Microsoft Copilot Checkout already allows you to select a product, choose a variant, pay via Stripe or PayPal and receive confirmation – all in the conversational interface. 70% of consumers say they are comfortable with this model (Invisibletech, 2026).
Faut-il un budget important pour s’adapter au zero-click commerce ?
Most of the work is based on the optimization of existing data: Schema.org enrichment, standardization of product attributes, creation of an llms.txt file. The investment is structural, more than an additional media cost.
