Agentic Commerce Glossary

An Agent Offer Network connects intent inside an AI product with structured offers and the evidence needed to attribute reported commercial outcomes.

AI commerce is the use of artificial intelligence across the commerce journey to help users discover, evaluate, personalize, and move toward products, services, or transactions.

In agentic commerce, this journey can happen inside an AI agent, app, assistant, recommendation system, or conversational product. The AI product interprets the current request and decides which commercial next step helps the user. AON provides the offer discovery, native presentation, and attribution layer that connects that intent to structured merchant Offers and later reported outcomes.

AI commerce infrastructure is the stack an AI product uses to interpret intent, find Offers, present recommendations, hand users to merchants, and measure results. AON provides the offer discovery, attribution, and monetization layer within that stack.

Machine-readable commerce is the practice of making commercial information and actions understandable to software.

AON applies this idea to structured offer discovery and attribution. It does not replace merchant checkout, payment authorization, order acceptance, or fulfillment systems.

AI agent advertising makes merchant Offers available inside relevant AI-agent and app interactions. AON supplies the intent-led offer, attribution, and outcome-based monetization infrastructure for this ecosystem.

Agentic commerce is commerce in which an AI agent takes goal-directed steps toward a commercial outcome. AON supports the offer discovery, native presentation, attribution, and monetization stages of that workflow.

An AI shopping agent is an AI system that understands a shopping goal and uses tools or data to find, compare, or take permitted actions on products and offers.

Conversational commerce uses messaging, chat, or voice conversations to help customers discover, evaluate, buy, or get support for products and services.

Agent-to-agent commerce describes commercial workflows in which agents exchange intent, offer, or event information on behalf of buyers, sellers, or systems.

In AON, agents and supply-side systems exchange structured intent, Offer, and event data. The OfferProvider API supports offer discovery, and postback flows report downstream events. The AgentOffer Protocol roadmap extends this foundation toward more autonomous A2A transaction workflows.

Agentic checkout is the transaction stage where an AI agent helps a buyer move through a merchant-controlled checkout session. In the AON ecosystem, it follows offer discovery and the attributed merchant handoff.

Agentic payments are payment or authorization flows in which an AI agent participates in initiating, approving, or coordinating a payment step.

Within an agentic commerce stack, payment systems handle authorization and payment processing. AON occupies the offer discovery, attribution, and monetization layer around that transaction: it preserves the Offer interaction, correlates conversion or Partner postback events, and supports commission settlement for qualified outcomes.

AI agent monetization is how a product powered by an AI agent earns revenue through user fees, paid promotion, or commission on qualifying referrals.

AI shopping agent monetization earns revenue from buying assistance through user subscriptions or service fees, sponsored recommendations, or commission on qualifying referrals.

AI affiliate monetization uses artificial intelligence to support earning affiliate commissions through content, marketing workflows, or product recommendations.

AI app monetization is the process of generating revenue from an AI-powered application through subscriptions, paid features, services, or commercial outcomes.

AI search products, copilots, assistants, chat interfaces, and recommendation apps can combine subscriptions, paid features, usage-based pricing, advertising, and commercial outcomes. In an agentic commerce use case, AON adds an outcome-based monetization layer: the app requests Offers for the current intent, renders a relevant option in its own interface, and can earn commission from a validated attributed outcome.

AI recommendation monetization is the practice of earning revenue when an AI product presents a relevant commercial recommendation and a defined downstream outcome is attributed to that interaction.

In an AI agent, app, assistant, or conversational product, the recommendation appears inside the user's current task. AON supplies an eligible structured Offer and preserves the tracked action. When the user continues to the merchant, a merchant or Partner can report the defined outcome and AON can evaluate the applicable commission.

Outcome-based monetization pays for a defined, attributed, and validated commercial result. In an AON-powered agentic commerce workflow, that result connects a merchant-reported Goal outcome to the originating AI-product Offer interaction.

Performance marketing is an approach in which marketing spend or compensation is tied to measurable actions or commercial outcomes.

Performance marketing can include click-, lead-, signup-, booking-, or sale-based models across advertising and affiliate channels. AON overlaps with the outcome-based part of performance marketing when an AI product presents a structured offer and a merchant or supply partner later reports a defined CPA or CPS result. AON is not a general media-buying, campaign-bidding, or display-inventory platform.

Affiliate marketing is a performance-based arrangement in which a publisher or partner can earn commission after a referred user completes a defined qualifying outcome.

Intent-based advertising selects ads based on what a person appears to want to do, such as compare products or book a trip.

A direct request or recent research can suggest a goal, but does not prove someone will buy. This differs from targeting a broad interest, such as travel. AON helps AI apps find commercial Offers for the current request; the app chooses what to show and identifies paid placements where applicable. A relevant Offer is not automatically a paid advertisement.

A sponsored recommendation is a recommendation that an advertiser or sponsor pays to include, highlight, or present to users.

A commission model sets how an AI product can earn when a merchant validates a sale, lead, booking, signup, or other commercial outcome.

Cost per action (CPA) is a commission model where compensation is tied to a defined action such as a signup, lead, install, booking, or purchase.

CPA is broader than CPS because the qualifying action does not have to be a sale. In AON, CPA terms should be connected to conversion events and postback rules so the publisher understands what counts as a qualified outcome.

Cost per sale (CPS) is a commission model where compensation is tied to a completed sale or purchase.

CPS is common in affiliate commerce because it connects publisher compensation to revenue-producing merchant outcomes. In AON, a CPS offer needs sale confirmation, order value or commission amount, attribution context, and partner reporting after the user reaches the merchant destination. This makes CPS different from CPA: the payable event is specifically a completed purchase rather than a broader action such as a lead, install, or signup.

A hybrid commission model is a commercial arrangement that combines more than one compensation basis, such as a fixed payment for an action plus a percentage of a later sale.

When an AON Offer recognizes multiple payable outcomes, the documented contract represents them as explicit Goals with CPA or CPS pricing. The commercial terms determine whether and how those outcomes combine.

Cost per click (CPC) is an advertising model in which compensation is tied to a recorded click.

A CPC campaign assigns a price to each qualifying click, regardless of whether the user later completes a purchase or another downstream action. AON currently uses CPA and CPS for Goal pricing, so CPC appears in this glossary as a comparison model.

An impression is a recorded instance of an offer, ad, or recommendation being shown to a user.

Advertising platforms use impressions to measure exposure and calculate metrics such as reach, frequency, and click-through rate. AON's documented attribution workflow begins with tracked clicks and continues through conversion events and attributed outcomes; impression remains a useful comparison metric for advertising models.

Disclosure is a clear label or explanation that tells a user about the commercial relationship behind a recommendation, link, or offer.

In AON, disclosure accompanies offers that are sponsored, commissionable, or commercially connected to a publisher, advertiser, or partner. The SDK can include a disclosure label by default, and publishers should keep it close to the recommendation and make sure it accurately describes the relationship. Specific disclosure requirements vary by market and context.

An affiliate network connects merchants with websites, creators, and apps that can earn commission by referring customers. It supports program access, referral tracking, reporting, and commission administration.

An affiliate subnetwork gives websites, creators, and apps access to multiple merchants' affiliate programs through one intermediary's own affiliate accounts and relationships.

The intermediary joins merchant programs or networks; its participants use those relationships as subpartners. Unlike joining a network directly, a publisher operates through the intermediary's affiliate account. It can manage links, reports, and payments in one place. Commission sharing and merchant restrictions still apply. An authorized business could supply these offers to AON, but connecting the systems does not replace upstream approval.

An affiliate link is a tracked URL used to attribute user traffic or downstream actions to the publisher or partner who referred the user.

Traditional affiliate links are often inserted manually into static content. In an AON workflow, the matched Offer returns a tracked destination dynamically. The publisher can render that recommendation natively and carry the attribution context into the merchant handoff.

An affiliate deep link sends a shopper to a specific merchant product, category, or campaign page while tracking which publisher referred them.

It saves the shopper from searching the merchant's homepage. The merchant program must support the destination and tracking method; an ordinary product URL is not automatically an affiliate link. An AI app using AON should keep the returned tracked link intact so the referral can be connected to a later purchase.

Commerce media is advertising that uses shopping activity or data to reach potential customers and measure results such as purchases.

Retailers, travel services, and other commerce businesses can offer advertising on their sites, in stores, or through partner channels. Retail media is the retailer-focused part of this broader category. AON supports commercial Offer discovery and referral tracking in AI shopping, rather than a complete advertising buying platform. Not every AON Offer is a paid advertisement.

An ad network is a platform that aggregates advertising inventory from publishers and sells or distributes it to advertisers.

Ad networks are usually priced by impressions, clicks, targeting, or auction dynamics, and they are commonly organized around ad inventory. AON is positioned around commerce intent and offer outcomes rather than generic display inventory, because AI product recommendations should be useful, explainable, and tied to user tasks.

Programmatic advertising is the automated buying, selling, or delivery of advertising inventory using software, audience or contextual data, and campaign rules.

Programmatic advertising manages ad inventory, targeting, campaign delivery, and often real-time bidding across demand- and supply-side systems. AON serves a different commerce workflow: it queries structured merchant offers for expressed user intent and connects an Offer interaction to downstream outcomes. An AI product can use both systems for their respective purposes.

Display advertising is visual ad placement such as banners, boxes, popups, or sponsored units shown in publisher inventory.

Display advertising is usually inventory-led and optimized around placement, audience, and creative delivery. AON is positioned around user intent and structured merchant offers inside AI-native workflows, where the offer should fit the task rather than occupy a generic ad slot.

Retargeting is a marketing technique that shows ads to users based on previous browsing or interaction behavior.

Retargeting uses behavior from an earlier session or site visit to choose a later ad. AON uses the intent expressed in the current AI conversation to match a relevant merchant offer. The two approaches use different signals and serve different moments in the customer journey.

AON differs from traditional affiliate networks by focusing on AI-native offer discovery, structured payloads, native rendering, and attribution inside dynamic agent workflows.

Established affiliate networks support program relationships, tracked links, APIs, feeds, reporting, and automated operations. AON focuses on querying eligible structured Offers by current intent, publisher-controlled presentation, and attributable Partner-reported outcomes. An authorized Bridge can connect upstream affiliate supply to AON; these layers can complement each other.

AON and Amazon Associates both relate to commerce monetization, but they operate at different scopes: Amazon Associates is Amazon's affiliate program, while AON is offer infrastructure for AI-native publishers and multiple supply paths.

Amazon Associates is useful when a publisher wants to refer users into Amazon's retail ecosystem through approved affiliate links and placements. AON is useful when an AI product needs to query offers dynamically by user intent, work across multiple merchants or supply partners, render offers inside its own product experience, and preserve attribution through AON tracking and settlement flows.

AgentOffer Protocol v1.0 is the adopted stable contract for exchanging structured offers and attributed outcomes between Agents, AON deployments, and Offer Providers.

An Offer Query is the versioned request and response interaction used to ask AON for structured offers that fit current intent and supported context.

An Offer Goal declares the exact downstream event and pricing basis that can enter attribution and billing evaluation for an Offer.

A multi-goal offer declares more than one possible downstream outcome, with each Goal evaluated under its own event and pricing terms.

Each Goal can identify a different event and pricing basis, such as a qualified action and a confirmed sale. Multiple Goals do not authorize payment by themselves; attribution, validation, and the applicable commercial terms still determine whether an outcome becomes payable.

A Public Offer is the AI-product-facing projection of a commercial offer that AON returns for evaluation and presentation.

A Partner Offer is the supply-side offer object a Partner returns to AON through the OfferProvider contract.

An offer instance is the identity of one particular delivery of an Offer in a Query response.

An empty offer result is a successful offer query response that contains no Offer objects for the current request.

An empty result is a successful response with an empty Offer collection. It occurs when the available supply contains no candidate that meets the current eligibility and relevance conditions. The AI product can continue the user's task, ask for a useful refinement, or present a broader non-commercial answer.

A structured offer schema is a consistent data shape for describing commercial offers to software and AI systems.

In AON, the structured offer schema defines the offer fields publishers and partners can rely on across offer query, rendering, tracking, and attribution workflows. It lets developers avoid scraping landing pages or manually normalizing merchant feeds, while making offers easier for AI systems to summarize, compare, cite, and render safely.

Machine-readable content is information published in a structured, predictable form that software and AI systems can parse and use.

Common formats include JSON responses, JSON Schema, OpenAPI descriptions, structured data, and Markdown metadata. In AON, the Offer schema and API contracts describe commercial data. Resources such as agent.md and llms.txt direct agents to the relevant sources. Together, these formats make AON content easier to retrieve, interpret, compare, and cite.

An offer action is the structured next step supplied with an Offer, such as a web redirect or an app deep link that continues the user toward a merchant destination.

An Offer action can describe a web redirect or app deep link. The applicable delivery contract determines the executable action, and an AON consumer should preserve the returned tracked action and append only attribution parameters supported by the active integration, so the documented handoff and attribution context are preserved. The merchant destination remains authoritative for checkout and current terms.

An offer catalogue is an organized collection of commercial offers that can be searched and retrieved, with the information needed to evaluate their actions and conditions.

An affiliate API is an application programming interface that lets software access affiliate-program data or functions, such as products, tracking links, promotions, and transaction reports.

The OfferProvider API is the Partner-facing Protocol v1.0 contract for exposing eligible supply-side Offers to AON.

Offer Fetch is the AON-to-Partner supply contract used to retrieve structured Partner Offers for a public Query.

The AON category taxonomy is the controlled Protocol v1 system of canonical IDs used to classify Offers and narrow Queries.

A category schema describes the decision factors an AI product can use to clarify a category-specific request before searching for offers.

Location targeting evaluates geographic Offer eligibility using canonical AON location IDs, Offer rules, and supported viewer context.

Location Search is the AON lookup surface for resolving and normalizing supported location identifiers.

Location Search can find canonical IDs, resolve supported external signals, and return a self-to-root location chain. It is a lookup and normalization API, not an Offer search endpoint. Applications can pass resolved viewer location IDs through the documented Query context.user_profile.location_ids field.

A request ID identifies one Offer Query request and helps an integration correlate retries, responses, and conversation rounds.

The active integration defines how request_id participates in idempotency and session continuity. Reuse the same ID only when retrying the same request; assign a new ID when the query changes. Do not share caller-generated IDs across end users under one credential.

Idempotency lets a repeated request or event produce one intended result instead of duplicate business effects.

In AON integrations, a request ID lets an exact Query retry return the prior response. Webhook and postback receivers use durable event identities and stored deduplication records so each business event is applied once. A changed Query receives a new request ID.

A Query helper is a server-supplied follow-up suggestion that helps a product refine the next Offer Query.

A Query helper can include a label, a protocol field patch, and origin metadata. When a user adopts it, the host applies the patch to the next request, echoes the origin in intent.origin, and chains the prior response request ID so the service can preserve conversation continuity.

A match reason is a concise, user-facing explanation of why AON returned an Offer for the current intent.

A match reason is authored by AON for the public Offer response and is tied to the current Query. It helps an AI product explain relevance without exposing private ranking data or chain-of-thought. It is distinct from Partner-authored supply fields and can be omitted when the active response options disable it.

Session continuity is the bounded use of prior Query context to improve a later Offer search without sending a complete conversation transcript.

AON session continuity uses the supported previous request ID and bounded recent topics from the prior round. It helps the runtime understand the relationship between consecutive searches, but it does not create a long-term user profile, change recall by itself, or authorize the product to infer unstated preferences.

Tolerant degradation keeps the core Offer workflow usable when optional guidance or enrichment is unavailable.

In an AON integration, the host can continue through the core search path when optional refinements, follow-up topics, or guidance metadata are unavailable. Required validation, eligibility, and contract fields still govern the Query and its returned Offers. A Placement ID remains necessary for attribution, but anonymous search does not require a key.

llms.txt is a proposed Markdown format that gives AI agents concise site context and curated links to authoritative resources.

An AI crawler is a program that retrieves web content for an AI system’s discovery, processing, or retrieval workflows.

An AI crawler is different from an AI model and from an indexing or citation decision. robots.txt can state crawl rules, while llms.txt can provide a human-readable site map and agent.md can guide an integration; none of these assets guarantees crawling, indexing, ranking, or citation.

agent.md is a Markdown-based integration entry point that gives an AI agent a readable, machine-addressable contract for using a service.

AON publishes agent.md as an Agent integration quickstart with the portable AgentOffer contract, version-selection rules, schema references, and deployment-access boundaries. It is an AON integration asset, not a universal protocol or a replacement for the underlying Query, Offer, MCP, or postback specifications. An agent should use the linked canonical contract sources and the deployment's active endpoint and access rules when implementing the integration.

A Representational State Transfer (REST) API endpoint is an HTTP route that a client calls to read, create, update, or trigger behavior for a resource.

In AON, REST API integration lets teams query offers directly without using an SDK or API key. Queries without a Placement ID run anonymously and earn no attribution or payouts. The API documentation defines the current request, response, and compatibility behavior; integrations should send only the intent and fields documented for their active environment.

API integration connects software systems through application programming interfaces so they can exchange data or request supported actions.

Teams call documented operations directly or through an SDK. API integration describes how systems communicate; embedded integration describes where a capability appears in a product. An embedded component can use an API behind the scenes. MCP lets compatible AI apps discover and call tools that may themselves use APIs. AON's Offer Query API returns eligible offers without an API key; the app controls presentation. A serviceable Placement ID attributes traffic, while credentials for account-scoped operations stay on the server.

A software development kit (SDK) is a collection of software development tools packaged for building applications on a platform, service, or system.

In AON, the SDK wraps offer queries, click reporting, conversion reporting, formatting helpers, context detection, and mock-mode testing for application code. AON provides a TypeScript SDK path through @agentoffernetwork/sdk.

Embedded integration places another service's interface or capability inside an existing website or app, so users can access it there.

Teams can embed a widget, load a JavaScript SDK, or build an interface around a service's API. Embedded integration describes where a capability is used; API or MCP integration describes how software accesses it. These approaches can work together. With AON, an app uses eligible offer data and preserves tracking while controlling its own presentation; embedding does not itself authorize a purchase.

An AI plugin is an extension that adds instructions, tools, or access to outside services to an AI application.

The AI platform defines installation and permissions. Some plugins connect through APIs or MCP; others provide instructions only. A plugin can include a user interface, so it is not simply the opposite of an embedded widget. AON's MCP tools let compatible assistants search eligible commercial offers; the assistant chooses what to recommend.

An embedded widget is a ready-made interface component added to a website or app, such as a chat window or product recommendation panel.

Teams commonly embed widgets with a JavaScript snippet, SDK, or iframe. A widget supplies an interface; an AI plugin extends an assistant's capabilities, and the two can work together. AON Content Assistant is a hosted example: publishers add its web script and configure supported entry points. Custom offer interfaces can instead use the API or SDK. Setup depends on the selected component and integration.

An MCP integration connects an MCP-compatible AI host to external tools, data, or workflows through the Model Context Protocol.

Mock mode is a development or testing mode that replaces live external dependencies with simulated responses.

In AON, developers can use SDK mock mode to test locally without live credentials or network calls. Mock mode uses in-memory sample offers and does not demonstrate live offer availability. The Public Test Sandbox is a separate, shared HTTP path with fixed non-billable data. Live Offer Query and MCP search can also run anonymously, while attributable traffic needs a serviceable Placement ID.

User intent is the goal a person wants to accomplish through a question, request, or action.

In agentic commerce, user intent appears inside an interaction with an AI agent, app, assistant, or conversational product. It usually combines a task with useful context such as product category, budget, location, timing, or preferences. The AI product represents that current goal and supported context in an AON Offer Query, and AON returns relevant merchant Offers for the product to evaluate.

An intent signal is a word, action, or contextual detail that helps a system understand what a user wants.

In an AI agent or conversational product, intent signals can come from the current query, selected filter, clicked category, stated preference, or other supported interaction context. The AI product carries those signals into an AON Offer Query as intent and constraints, giving AON the information it needs to find suitable merchant Offers.

Commercial intent is a user's current goal to research, compare, select, book, or buy a product or service.

In agentic commerce, commercial intent appears inside an interaction with an AI agent, app, assistant, or conversational product. The user expresses a product category, service need, transaction, or buying decision together with useful constraints. The AI product represents that intent in an AON Offer Query, and AON returns eligible structured merchant Offers for the product to evaluate and present.

In agentic commerce, intent matching interprets a user's current request and measures how well candidate Offers fit the expressed goal and requirements. AON applies this step between an AI-product Query and eligible merchant supply.

Agentic offer discovery is how an AI app finds and evaluates commercial offers for a user's request before checkout.

Offer ranking orders eligible offers according to the active relevance and commercial rules for a request.

In an agentic commerce workflow, ranking happens after an AI product sends the user's current intent and supported constraints to the offer discovery layer. AON identifies eligible Offers, orders them by their fit for that request and the applicable commercial rules, and can return a match reason that helps the AI product explain relevance.

Intent provenance records whether information in an agentic commerce request came directly from the user or was inferred by the AI product. The product carries that distinction into the AON Offer Query so the ecosystem can preserve the source of the intent it evaluates.

Offer eligibility determines whether an Offer can be returned for the current agentic commerce request. AON evaluates the supplied intent, supported context, and serving rules before ranking eligible merchant Offers.

A structured offer is machine-readable information that tells an AI product what a merchant is offering and what the user can do next. It is the commercial object AON moves between supply Partners and agentic product experiences.

A merchant offer is a specific commercial proposal from a seller for a product or service.

In agentic commerce, an AI product needs a merchant proposal in a structured form it can evaluate inside the user's current workflow. A merchant offer brings together the seller, product or service, applicable terms, action destination, and commercial outcome. AON turns supply-side offer data into a Public Offer that the AI product can compare, explain, and present natively.

A verified merchant offer is AON's term for a structured Offer whose supply source, Merchant association, and action path are recorded through a configured integration.

Here, verified describes documented supply provenance: the AI product can identify which Merchant or supply Partner provided the Offer and which action path accompanies it. AON preserves that source relationship in the structured Offer so the product can evaluate and present a traceable commercial option. The Merchant destination remains the source of current transaction details at checkout.

An offer payload is a populated, machine-readable representation of one commercial offer returned inside an API response.

Offer freshness describes whether the commercial information in an offer is current enough to use when an AI product considers presenting it.

In an agentic commerce interaction, the AI product may revisit an option after the original Query. Offer information ages as supply status, destinations, prices, availability, and campaign terms change. Freshness helps the product decide when to reuse an AON result and when to submit a new Query before presenting the Offer again.

A real-time offer is an offer retrieved or evaluated from current supply data at or near the time of a user's request.

In an agentic commerce workflow, the AI product requests commercial options as part of the user's current interaction. AON queries or evaluates the latest supply data available through the active integration and returns eligible structured Offers for that request. The merchant destination confirms the transaction details when the user continues.

A live offer is an active production offer that can be considered for eligible user requests.

Within the AON ecosystem, a live Offer belongs to an enabled production supply path and can enter eligibility and ranking for current AI-product requests. Live describes operating status; real-time describes when the system retrieves or evaluates the Offer data. The phrase ‘live-time offer’ may refer to either idea, so the surrounding context determines the intended meaning.

A product data API lets software request product details, such as specifications, images, prices, and availability, from a store or data provider. Some APIs also let authorized teams update their catalogue.

A product feed is a structured collection of product records supplied to another system for discovery, comparison, advertising, or affiliate publishing.

Structured product data organizes product information into named fields, such as brand, size, weight, and seller price, so software can read and compare it.

Product data enrichment adds useful information to a product record, such as dimensions, materials, images, or compatibility details.

It fills gaps so shoppers can understand and compare products. Enrichment adds information; normalization makes existing information consistent. Added facts should come from reliable sources, not invented specifications. Businesses supplying offers to AON can include these details so AI apps can assess whether a product fits a user's needs.

Product schema markup is code on a product webpage that helps search engines and other software understand the product and its selling details.

It commonly uses Schema.org vocabulary in a format called JSON-LD. Product describes the item; Offer describes selling details such as price and availability. Markup must match the visible page and does not guarantee an enhanced Google result. AON Offers serve a separate purpose: exchanging commercial options and referral information with AI apps.

Product data normalization puts product information from different sources into consistent field names, formats, and units so it is easier to search and compare.

It changes how information is represented, not which facts are available or whether two records describe the same product. Teams keep the original meaning and source values. Businesses supplying offers to AON map their data into a shared Offer format, helping AI apps read commercial options consistently.

Product entity resolution determines whether records from different sources describe the same product, including its exact version, such as color or storage capacity.

Teams compare product identifiers, brand, model, and version details rather than relying on similar names. GTINs, or Global Trade Item Numbers, can help identify products across sellers. Even when an AI app identifies the same product in two AON Offers, each merchant's price, terms, and tracked link remain separate.

Product mention extraction finds product names in writing or conversation, such as a headphone model mentioned in a review.

Some tools also identify brands or product categories. Finding a name is different from identifying the exact catalogue item or deciding whether someone wants to buy it. An AI app can extract mentions before asking AON for related commercial options; AON retrieves Offers, while the app handles the extraction.

In AON, a publisher is an AI agent, app, assistant, or website that presents commercial options to its users and can earn revenue from attributed outcomes.

In an AON agentic commerce workflow, the publisher interprets the user's current request, represents that intent in an Offer Query, and decides whether and how a returned Offer fits the experience. AON retrieves and ranks eligible merchant Offers, preserves the interaction for attribution, and evaluates later reported outcomes under the applicable commission terms.

A merchant is the business that sells the product or service and owns the destination where the customer completes the transaction.

In an AON agentic commerce workflow, the Merchant supplies the commercial destination and remains responsible for current product information, price, availability, checkout, payment, order acceptance, and fulfillment. A Merchant may run its own Offer program as the Advertiser or make its Offers available through a supply Partner.

An advertiser is the commercial party that funds or owns an Offer program and defines the terms under which a user outcome can earn commission.

In an AON agentic commerce workflow, the advertiser defines the Offer, its eligible Goal, and the applicable commercial terms for distribution through AI agents, apps, and assistants. The advertiser may also be the Merchant, but the roles remain distinct: the advertiser owns the commercial program, and the Merchant owns the product or service and transaction experience.

A partner, also called an offer supply partner, is an organization that connects Merchant offer inventory and downstream outcome signals to AON.

In an AON agentic commerce workflow, a Partner can aggregate Offers from multiple Merchants, return eligible inventory through the OfferProvider path, and report later outcomes for attribution and commission evaluation. A Partner can be an affiliate network, marketplace, merchant aggregator, commerce platform, or another supply-side system. Its supply integration is separate from the Publisher integration used by AI products.

An attribution layer records which product, surface, click, or conversion should receive credit for a commercial outcome.

In agentic commerce, a recommendation can begin inside an AI conversation and convert later on a merchant surface. AON is the attribution and monetization layer that connects those stages: tracked destinations and identifiers preserve the originating Offer interaction, conversion events or Partner postbacks report the outcome, and reporting presents the result to publishers and advertisers.

Agentic offer attribution connects a reported commercial outcome to the AI product, Offer instance, and tracked interaction that preceded it. AON preserves that evidence path from the agentic experience to the merchant-reported result.

AI commerce attribution assigns credit for a purchase or other commercial action to an earlier AI interaction using observed evidence and defined rules.

A tracked destination is the URL or routing target a user follows from an AON-rendered offer so the click and later outcome can be attributed.

The tracked destination is the clickable destination itself: a tracked URL, redirect route, or routing target returned with the offer payload. Publishers should preserve it instead of replacing it with a raw merchant URL, because it may carry identifiers or route through tracking infrastructure needed for click attribution and downstream conversion matching.

A tracking identifier is an ID value carried in an offer payload, tracked URL, event, or postback to connect activity to the correct publisher, surface, and offer context.

A tracked destination is the URL or route a user follows. A tracking identifier is the underlying value used to match and report activity across the Offer, click, and conversion lifecycle. The integration carries this implementation data through the supported tracking fields.

Sub-ID attribution carries a publisher’s own reference through a tracked interaction so a later conversion can be associated with its user, channel, or campaign.

AON Offer Wall accepts a publisher's sub_id reference on the hosted URL or documented tracked link and returns it in conversion and earning status webhooks. Optional attribution slots can distinguish channels or campaigns. These references supplement AON tracking IDs; they do not prove a conversion or guarantee payment. Publishers verify and deduplicate the updates, then use the largest sequence per earning_id rather than adding snapshot amounts. On rejection, reverse only rewards already credited; AON does not pay the application's end users on its behalf.

A tracking macro is a placeholder in a Partner landing-URL template that AON replaces with a click or pseudonymous value.

AON currently supports {CLICK_ID} for per-click attribution and {PSEUDO_USER_ID} for a best-effort click-time pseudonymous value. A macro is expanded only when the exact placeholder appears in the configured template, and the inserted value is URL-component encoded; it is not an account ID, credential, or guarantee of anonymity.

Click attribution records that a user clicked a tracked offer from a specific product surface.

In AON, click attribution connects an AI product interaction to downstream merchant activity. The Offer supplies a tracked destination or tracking identifier, and the publisher uses that value when the user follows the recommendation. A later conversion can then be matched to the originating product, surface, and Offer.

A conversion event is a downstream action that may qualify for attribution, reporting, or commission settlement after an AON offer interaction.

In the AgentOffer Protocol postback path, event_name is evaluated against the Goal events declared by the served Offer. An exact Goal match is required for billing evaluation, while a valid unmatched event may be recorded as non-billing. AON uses attribution evidence and partner reporting to connect the event to the originating offer, publisher, and surface; successful event matching does not by itself guarantee that commission is payable.

In affiliate marketing, a postback is a server-to-server notification that reports a downstream conversion to the system responsible for attribution.

A Developer webhook is AON's signed server-to-server delivery of a supported conversion or earning status event to a Developer Application endpoint.

Webhook signature verification checks that an incoming event was sent by the expected AON delivery path and was not altered in transit.

A receiver verifies the exact request target, raw request body, timestamp, key identifier, and HMAC signature before parsing or applying a conversion event. Signature verification is only one part of webhook security: the receiver must also enforce timestamp windows, durable idempotency, payload validation, and the documented retry behavior.

Conversion attribution connects a downstream action to the offer interaction that influenced it.

AON SDK and API materials distinguish click reporting from conversion reporting. In production, conversion attribution maps a qualified action, such as a sale, lead, booking, or signup, back to the original offer, click, publisher, and surface without forcing the publisher to own merchant-side checkout data.

A commission outcome is the user action or downstream event that can trigger payable revenue for a publisher.

In AON, the commission outcome varies by Offer and commercial agreement. Each Goal names the downstream event and defines its CPA or CPS pricing. An Offer can declare multiple Goals when its commercial terms recognize more than one payable outcome, such as a qualified signup followed by a subscription purchase.

A qualified outcome is a reported event that matches the served Offer's Goal and satisfies the applicable attribution, validation, and commercial conditions before it can become payable.

A click or raw conversion report is not automatically a qualified outcome. AON and the applicable commercial parties evaluate the event against the served Offer, attribution evidence, deduplication, validation, and the declared terms.

Commission settlement is the process of determining whether a commission is payable and, if so, calculating and reconciling the amount.

In AON, settlement begins after an offer interaction and a later outcome are matched. New conversions use automatic confirmation by default; sources configured for manual confirmation retain their confirmation process. Confirmed acknowledges the conversion, available means the earning is available for payout, and paid out records AON's payment rather than receipt in a bank account. Validation and applicable commercial terms still determine payable commission.

Cookieless affiliate tracking connects a referral with a later purchase using methods that avoid some or all browser cookies.

A cookie can help a website remember who referred a shopper after a click. Alternatives include referral codes or account-linked records. Some systems avoid third-party cookies but still use first-party cookies. Server-to-server reporting can also use cookies; it describes how purchase data is sent. AON matches reported outcomes with tracking identifiers, which does not prove the whole shopping journey is cookie-free. Privacy requirements still apply.

An attribution window is the time limit for connecting a referral or ad interaction with a later purchase or other result.

It answers how long a referral remains eligible, not how long commission approval takes. A purchase within the window must still meet other rules and may be credited to another publisher. AON connects tracked interactions with reported outcomes; the applicable program rules determine whether a purchase qualifies.

Revenue share is an agreement to give a participant a percentage of a defined income amount, such as sales revenue or affiliate commission.

The key question is: a percentage of what? Sharing sales income differs from sharing commission or profit. In AON, the commission payable by a supply Partner is not the publisher's final share. Fees, refunds, and other adjustments can affect payment, and estimated earnings are not the same as money received.

AON Content Assistant is an embedded website assistant that uses page content and visitors' questions to help them discover relevant commercial offers.

Publishers add the aon.js script with a Content Assistant Placement ID. AON hosts the assistant and matches offers to the page and the visitor's request, without a manually curated offer list. The web script supports a floating button and an end-of-content entry point; the latter requires a static slot on a public HTTPS page. The integration guide lists current styles and page-access limits.

An offer wall is a browsable list of commercial offers that users can explore and act on within a website or app.

In AON, Offer Wall is a placement type with an AON-hosted page and an API option for your own interface. Both use the same placement and support attribution through the publisher's sub_id reference. Publishers can use conversion and earning updates to manage their own user rewards. Click links have no time-based expiry, but require a serving application, an active placement, and an offer that remains selected. Refresh the list when viewer context changes. An offer listing or click does not by itself guarantee a reward or payable commission.

Native rendering displays a structured offer inside an AI product's own experience instead of forcing a generic ad unit or external widget.

An inline text component is an AON offer surface rendered as a linked phrase or short recommendation inside an AI-generated answer.

Inline text components render a structured Offer as part of the answer text instead of using a separate card or banner. The Offer is selected for the current intent and includes the attribution fields needed for the linked recommendation.

View visual example

Inline text component

A merchant card is an AON offer surface that displays a recommended merchant offer with enough context for a user to act.

Merchant cards are useful when the user needs to compare options, see price or merchant context, and make a decision without leaving the AI workflow too early. Unlike a traditional display ad card, the card is rendered by the publisher from AON's structured offer payload and should appear only when it fits the user's current task.

View visual example

Merchant card

An actionable banner is an AON offer surface that gives the user one clear next step from a matched merchant offer.

Actionable banners are useful for high-confidence offers, post-answer prompts, or persistent placements where a full card would be too heavy. Unlike generic banner ads, an AON banner should be tied to a specific user intent and can use payload fields such as merchant, price, destination, and tracking data.

View visual example

Actionable banner

A custom presentation surface is an AON-powered offer component designed and controlled by the publisher.

Custom surfaces use the same structured offer payload as inline text, cards, and banners, but the publisher decides the layout. They differ from third-party ad widgets because the publisher owns the interaction design while AON supplies offer data, tracking, and attribution context.

View visual example

Custom presentation surface

The Agentic Commerce Protocol (ACP) is an open specification for programmatic commerce and checkout interactions between AI agents and sellers.

ACP defines agent-facing interfaces for commerce operations such as checkout-session management and secure payment-credential relay, while sellers retain control of pricing, inventory, order acceptance, payment processing, and fulfillment. It can complement AON’s offer-discovery and attribution layer, but it is a separate protocol from AgentOffer Protocol.

Universal Commerce Protocol (UCP) is an external commerce protocol intended to standardize interactions among buyers, businesses, and AI agents.

UCP describes commerce capabilities such as catalog, checkout, and order interactions. AON focuses on intent-led offer discovery, native presentation, attribution, and commission evaluation. A commerce workflow can use each system for its respective part when the merchant and integration support both.

Agent2Agent Protocol (A2A) is an open protocol for communication and coordination between AI agents.

Agent2Agent Protocol (A2A) provides an interoperability layer for agents to discover capabilities, exchange messages, and coordinate tasks. In a commerce workflow, A2A can coordinate the participating agents and an AON-compatible service can supply structured Offers and attribution events.

Agent Payments Protocol (AP2) is Google's open protocol for secure agent-led payments and transaction authorization.

Relationship to AON: Agent Payments Protocol (AP2) is a payment and authorization layer, while AON is an offer discovery, native rendering, attribution, and commission infrastructure layer. The two can be complementary when an agentic commerce flow needs both offer matching and a separate payment protocol.

A Placement ID is a public identifier for an AON offer surface that connects served traffic to its owning Developer Application.

An Application that is enabled to serve offers owns placements for its product surfaces. A Placement ID attributes Offer Query and MCP search traffic to that Application; anonymous search earns no attribution or payouts. The ID is public and is not an API key. Keep credentials for account operations such as Reporting and Offer lookup on your server.

An API key is a credential for AON account-scoped capabilities such as Reporting; live Offer Query and MCP search use a Placement ID when traffic should be attributed.

Query and MCP search can run anonymously without a key or Placement ID, but anonymous traffic earns no attribution or payouts. Keep Reporting keys server-side, out of repositories and logs, and use the active API documentation for the required scope.

A Developer Application is the application-owned identity and access boundary used by supported AON Developer integrations.

A serviceable Application owns Placements and related configuration in Developer Portal. Its Placement IDs attribute live Query and MCP search traffic; API keys serve account-scoped capabilities such as Reporting.

The Public Test Sandbox is a shared, query-only environment for inspecting the AgentOffer Protocol v1.0 Offer Query shape without production access.

The Sandbox uses fixed non-billable test data and does not create an Application identity, enable live inventory, attribute traffic, or prove MCP, Reporting, webhook, or production access. It is a contract and response-shape check, not a live integration completion signal.

A rate limit is a runtime control that bounds how frequently a client can call an AON endpoint or tool surface.

Rate limits depend on the endpoint and request identity, including anonymous or Placement ID traffic. An integration should read the documented response or error, back off according to the active contract, and avoid treating a rate limit as proof that the underlying Query or Offer contract is invalid.