Agentic Commerce Glossary: 50 Terms Every Merchant Should Know

Core Agentic Commerce

Agent An autonomous AI system that understands customer intent and executes commerce tasks without human intervention. Agents can discover products, negotiate pricing, process refunds, and complete transactions.

Agentic Storefront An ecommerce architecture where customers interact with AI agents instead of traditional product pages. Agents query commerce APIs to fulfill customer requests. Shopify agentic storefronts are the market reference implementation.

Agentic Commerce The broader shift toward AI-driven customer interactions across the commerce journey. Includes product discovery, personalization, checkout, customer service, and returns—all handled autonomously by agents.

Natural Language Understanding (NLU) The AI capability that translates customer intent into structured actions. If a customer says "I need a waterproof jacket for hiking in Colorado," NLU parses this as: product_type=jacket, material=waterproof, use_case=hiking, location=Colorado.

Intent Recognition The process of identifying what a customer actually wants, even if they don't say it explicitly. "I'm training for a half-marathon" signals high-performance running shoe intent, even though the customer didn't mention shoes.

Semantic Search Product discovery based on meaning, not keywords. Instead of "matching" the search term to product titles, semantic search understands context. "Shoes that won't hurt my wide feet" returns wide-fit shoes, even if that phrase doesn't appear in product titles.

Multi-Turn Conversation A series of back-and-forth exchanges between customer and agent. The agent remembers context across turns. Turn 1: "I need a winter coat." Turn 2: "With lots of pockets." Turn 3: "And it should be waterproof." The agent maintains context and refines recommendations.

Hallucination When an AI agent makes up information that isn't true. An agent that claims your store carries a product you don't sell is hallucinating. Modern agentic storefronts have guardrails to prevent this.

Agent Orchestration The middleware layer that routes customer requests to appropriate systems. If a customer asks a question, orchestration directs it to FAQ data. If they want to buy, it routes to product and pricing APIs. If they want to return, it routes to refund logic.

Agent Runtime The execution environment where agents operate. Shopify's agent runtime handles authentication, API routing, error handling, and conversation management. Think of it as the operating system for agents.

Product and Inventory

Product Attributes Structured data fields describing product characteristics. Examples: material, size, color, waterproofing_rating, insulation_type, weight, fit_width. Agents rely on attributes to match customer intent to products.

SKU (Stock Keeping Unit) A unique identifier for a specific product variant. A blue M t-shirt and a red L t-shirt have different SKUs. Agents use SKUs to manage inventory and pricing at the variant level, not the product level.

Inventory Real-Time Sync Live inventory updates flowing from your warehouse/WMS to Shopify. When you sell one unit offline, your online inventory decreases immediately. Agents see accurate stock in real-time.

Stockout When inventory for a product reaches zero. Modern agents handle stockouts gracefully: "We're out of stock, but we expect more in 3 days. Should I notify you?"

Overstock When inventory levels are high relative to expected demand. Agents can trigger dynamic pricing to clear overstock. "We have 500 units. Can I offer you $15 off to help?"

Demand Forecasting Using historical sales data and external signals (seasonality, trends, marketing activity) to predict future demand. Agents use forecasts to guide inventory recommendations and dynamic pricing.

Just-in-Time Inventory (JIT) An inventory model where products arrive just before they're needed, reducing carrying costs. Agentic systems pair well with JIT because agents match inventory to real-time demand.

Inventory Hold A temporary reservation of stock made by the agent when quoting a product to a customer. The hold lasts 5 minutes. If the customer confirms, the hold converts to an order. If they don't, it expires and stock releases.

Pricing and Discounts

Dynamic Pricing Real-time price adjustments based on demand, inventory, customer value, and other factors. An agent might quote $89 to one customer and $79 to another for the same product based on their history and context.

Price Quote A guaranteed price offered by the agent at a specific moment. Quotes are time-limited (typically 5 minutes). This prevents "bait and switch"—the customer sees a price and doesn't watch it change mid-conversation.

Loyalty Discount An automatic discount applied to returning customers based on their purchase history or loyalty tier. A customer with 10 previous purchases might get 10% off automatically.

Volume Discount A discount triggered by quantity. "Buy 10, get 15% off. Buy 50, get 25% off." Agents understand volume thresholds and recommend bulk purchasing when it's economical for the customer.

Promotional Code / Promo A discount token (e.g., SAVE20) that reduces price. Modern agents can validate and apply promo codes automatically. "I have a code SAVE20." Agent: "Applied. Your price drops from $150 to $120."

Margin The profit retained after cost of goods. Dynamic pricing tries to optimize margin while remaining competitive. If a product costs $40 and normally sells for $100, the margin is $60. Dynamic pricing might lower it to $85 to drive volume while maintaining reasonable margin.

Price Elasticity How sensitive demand is to price changes. If a 10% price increase causes 20% demand drop, elasticity is high (customers are price-sensitive). Agents use elasticity data to optimize pricing.

Tiered Pricing Different prices for different customer segments or quantities. VIP customers pay one price, standard customers another. Agents apply tiered pricing based on customer status.

Customer Data and Personalization

Customer Lifetime Value (CLV) The total profit expected from a customer over their entire relationship with your brand. A customer with CLV of $10K gets treated differently (more lenient returns, proactive offers) than a customer with CLV of $50.

Recommendation Engine The system that suggests products based on customer data. Agentic recommendation engines are semantic (based on intent and context) rather than algorithmic (based on similar users).

Personalization Customizing the customer experience based on individual data. A returning customer sees prices they can afford, recommendations aligned with their taste, and service levels matched to their CLV.

Customer Data Platform (CDP) A system that centralizes customer data from all sources: website, email, CRM, purchases, support interactions. Agents query CDPs to personalize experiences.

First-Party Data Data you collect directly from customers: purchase history, email signups, website behavior, survey responses. First-party data is reliable and actionable. Agents use it to personalize.

Behavioral Data Data about what customers do: which products they view, how long they spend on pages, what they abandon, what they buy. Behavioral data signals intent.

Psychographic Data Data about customer values, lifestyle, and preferences. "Eco-conscious consumer," "price-sensitive," "quality-focused." Agents use psychographics to recommend products that align with customer values.

Checkout and Transaction

Headless Checkout A checkout experience decoupled from your website. Instead of a website checkout page, customers complete purchases through agents, apps, or other channels. Headless checkouts are API-driven, not UI-driven.

One-Click Checkout A frictionless checkout where the customer approves one transaction and it completes without re-entering payment/address. Agents enable one-click checkout by handling those details automatically.

Buy Now Pay Later (BNPL) A payment option where customers buy now and pay later in installments. Agents can offer BNPL automatically to qualifying customers. "Pay $80 today, $80 in 30 days."

Payment Authorization Verification that a customer has funds or credit to complete a purchase. Agents check payment authorization before confirming an order. If declined, they offer alternatives: different card, BNPL, buy later.

Order Confirmation The final step where the customer approves a transaction. An agent asks: "Shipping 1 unit to [address] on 2-day delivery, total $87.50. Confirm?" Customer says yes, order locks in.

Transaction Logging Recording every transaction in your system. Agent-created orders appear in your admin and fulfillment systems exactly like manually-created orders. Logging is real-time.

Customer Service and Returns

Conversational Commerce Commerce conducted through conversation (chat, voice, messaging) rather than traditional website UI. Agents are the primary interface for conversational commerce.

Intent Routing Directing customer requests to the appropriate system. "I want to buy a shirt" routes to product discovery. "I want to return my order" routes to returns logic. "When will my order arrive?" routes to order tracking.

Support Ticket A formal customer service request logged in your ticketing system. Modern agents reduce support tickets by 60% by handling routine questions and requests autonomously.

Self-Service Return A return initiated by the customer themselves without agent help. Agents enable self-service by handling return logistics automatically. "Send it back, and I'll refund you once it arrives."

Reverse Logistics The process of moving products from customer back to you. This includes pickup, transportation, inspection, and restocking. Agents can manage reverse logistics proactively.

Refund Money returned to customer. Agents can issue refunds automatically once the returned item is received and inspected. "Your return arrived. Refunding $89 now."

Chargeback A customer dispute with their credit card company, claiming they didn't authorize a transaction. Agent-created orders have the same chargeback risk as customer-created orders.

Metrics and Analytics

Conversation Rate The percentage of website visitors who start a conversation with an agent. If 100 visitors come to your site and 15 start agent conversations, your conversation rate is 15%.

Agent-Assisted Conversion Rate The percentage of agent conversations resulting in a purchase. If 100 customers chat with an agent and 32 complete a purchase, your conversion rate is 32%.

Average Order Value (AOV) The average revenue per transaction. Agents typically increase AOV by 25–40% through intelligent upselling and bundling.

Repeat Purchase Rate The percentage of customers who buy more than once. Customers who interact with agents have 45% higher repeat rates because personalization builds loyalty.

Customer Satisfaction (CSAT) A survey metric asking "How satisfied are you?" Agentic systems score higher on CSAT because response time is instant and recommendations are relevant.

Net Promoter Score (NPS) A loyalty metric asking "How likely are you to recommend us?" Agentic systems improve NPS because agents remove friction and increase perception of the brand as modern.

Cost Per Conversion The cost to acquire a paying customer. Agentic systems lower cost per conversion by improving conversion rate with the same marketing spend.

Return on Ad Spend (ROAS) Revenue generated per dollar of ad spend. Agentic systems improve ROAS by converting more clicks into purchases.

Technical and Infrastructure

API (Application Programming Interface) A set of rules for how software systems talk to each other. Agentic storefronts use APIs to let agents query products, inventory, pricing, and customer data.

REST API A common API architecture using HTTP requests (GET, POST, PUT, DELETE). Shopify's agentic APIs are REST-based.

Webhook A real-time alert sent from one system to another when something happens. When inventory drops below 10, a webhook notifies your agent system.

Authentication Verification that an agent (or user) is who they claim to be. Shopify agentic APIs use OAuth 2.0 authentication for security.

Rate Limiting A cap on how many API requests an agent can make per second/minute. Rate limits prevent abuse and ensure fair resource allocation. Most Shopify agents have rate limits of 1000 requests per minute.

Latency The time delay between a request and a response. For agents, latency should be under 500ms for product searches and under 300ms for pricing. High latency makes conversations feel slow.

Uptime The percentage of time a system is operational. Agentic systems should have 99.9%+ uptime. Any downtime means your agents can't operate.

Scalability The ability to handle increased traffic and transactions without performance degradation. Agentic systems are designed to scale horizontally (add more agents without adding servers).

Key Takeaways

• Agentic commerce fundamentally shifts control: customers no longer navigate; agents navigate your systems on their behalf.
• Product data is the foundation. Attributes make agents smart. Unstructured data makes agents blind.
• Dynamic pricing and inventory management are real-time. Agents negotiate in the moment.
• Conversion rates improve 10–15x because agents remove friction.
• Metrics that matter: conversation rate, agent-assisted conversion, AOV, repeat rate, CSAT.

Frequently Asked Questions