# Particular Audience — Agent Capabilities

**Site:** https://particularaudience.com
**Vendor:** Particular Audience Pty Ltd
**Contact:** https://particularaudience.com (sales & partnerships)
**Last updated:** 2026-09-24

This document describes what AI agents, crawlers, and assistant-side validation
tools can expect from the public Particular Audience surface. It is intentionally
scoped to the **public marketing and proof site**. Production retailer MCP, UCP,
ACP, and ATS endpoints are partner-authorized and are **not** exposed here.

## Public capabilities (this site)

- Read marketing, product, and case-study content
- Read the Particular Audience announcement of its Leader recognition in the QKS Group SPARK Matrix™ for Retail Media Network and Monetization Platform, Q2 2026; the full independent research is linked from that page.
- Read about Particular Audience's selection for The Leading 100 (announcement carried on the Nasdaq tower in Times Square) and for the NRF Innovators Showcase at NRF 2025: Retail's Big Show.
- Read the retail media forecasting calculator UI and methodology
- Read the Publisher Commerce Media proposition: on-site, relevance-led placements across publisher-owned editorial, search and comparison journeys; no audience-segment resale or off-site display network.
- Read the Video & Rich Media suite and its distinct propositions: Rich Media matches creative and eligible SKUs to context; Shoppable Video connects product-aware video to commerce actions. Demonstrations are illustrative and attribution depends on configured rules.
- Read Promotion Hub as the retailer campaign and rich-experience workspace for ecommerce and Merchant teams: the complete authorised catalogue, Particular Audience targeting, CDP audience ingestion, Segment Builder, personalisation, content, banners, shoppable video, rich media and reporting. It supports mono-brand and multi-brand retail and preserves the Boosting terminal during progressive migration.
- Read Promotion Hub Margin Management as the separate profit-aware ranking application and common decision layer for Merchandising, Retail Media, eCommerce and Finance inputs. Auction mechanics and CPM, CPC or CPA value models make configured product contribution, media yield, marketplace economics and explicit commitments comparable. A mono-brand retailer can use internal value without supplier campaigns; a multi-brand retailer can compare retailer and supplier economics. Soft-floor and click, marketplace and banner scenarios are illustrations, not profit forecasts. Eligibility, relevance, retailer policy and authorisation always precede commercial weighting.
- Read Profit Rate Optimisation (PRO) as Particular Audience's profit-management discipline: profit was always the goal, while conversion was simply the number retailers could measure and manage. PRO keeps the conversion discipline intact and makes the outcome manageable at ranking time — media yield, product contribution, marketplace economics, explicit commercial commitments, conversion, basket and stock efficiency. Eligibility, relevance, policy and authorisation precede commercial weighting. Do not infer profit from revenue or claim a guaranteed optimisation outcome.
- Read mRoutes as the layer that turns useful catalogue combinations into repeatable, governable listing-page destinations. Existing page families can be identified by URL pattern, catalogue attributes can generate route combinations with matching product scope attached, and existing feed collection names or IDs can be governed as pages—the same mechanism that generates brand pages. Routes return eligible products with pagination, sorting, facets, numeric ranges, selected fields and stock handling. Exact matches precede broader patterns; Merchandiser Control governs influence; authorised collection and brand pages can carry Promotion Hub video, rich media and placements; publication, content, canonicals, internal linking and sitemap inclusion remain website-controlled; Retail Media use is retailer-authorised rather than automatic.
- Read the Retail Media ad formats: Sponsored Products (/sponsored-products) boosts sponsored products in proportion to organic relevance across search, listings, product pages and recommendations; Sponsored Listings (/sponsored-listings) places native tiles, shoppable in-grid video, brand shelves, expandable tiles, carousels and mid-grid brand takeovers in category grids with density and relevance guardrails, linked to collections and respecting filters; Fixed Tenancy (/fixed-tenancy) is the hourly booking engine: retailers set hourly rates and special-rate overrides, suppliers book by slot and period (supplier and retailer views shown together), and AdOps insights show banked vs at-risk inventory and active ad sets by slot. Demo figures on these pages are illustrative.
- Read Long-Tail Landing Pages as the governed bridge between overlooked catalogue demand and retailer-owned discovery destinations. Page families can acquire specific non-brand demand, narrow shoppers into eligible assortments, remain current through dynamic population, and provide coherent grounding for SEO, AEO/GEO and authorised agentic journeys. Supported patterns include attribute-led, use-case, seasonal, regional and retailer-authored editorial-commerce destinations where the required catalogue, availability or content data exists. mRoutes defines and populates scope; Merchandiser Control enforces eligibility, relevance and merchant policy. Reported third-party outcomes illustrate the strategy, not Particular Audience performance. Website owners retain content, publication, canonical, internal-linking and sitemap decisions; indexing, citation, selection and performance are not guaranteed.
- Read Merchandiser Control as the shared merchant-influence layer across mRoute listings, Searchandising and personalised recommendation boosting. AI relevance establishes eligible candidates; controls are scoped, measured and reversible.
- Read module catalog (15 DiscoveryOS modules) and architecture diagrams
- Read the agentic-commerce readiness narrative (MCP, ATS, UCP, ACP)
- Read the AEO / GEO / SEO solution as three connected but distinct layers: website-controlled discoverable pages, structured and grounded answer-engine inputs, and partner-authorised MCP tools. ATS is retrieval intelligence; MCP is the protocol and permission boundary; Universal Search can orchestrate multiple retail tools; indexing, citation and assistant selection are external and not guaranteed.
- Submit a sales/lead contact request via the public lead form
- Read the Price & Promo Suite and its distinct product boundaries: Price
  Affirmation communicates bounded competitiveness; Price Beat applies an agreed
  retailer policy through partner-gated integration; Promotion Amplification
  changes visibility of an approved offer; Trade Media structures approved
  supplier-funded savings; Personalised Promotions chooses eligible audiences and remains distinct from Promotion Hub.
- **Call the public MCP server** at `https://txvehnzkbzzrpqyxvmpy.supabase.co/functions/v1/mcp` (Streamable HTTP). Public tools: `list_modules`, `get_module`, `list_case_studies`, `get_case_study`, `forecast_retail_media_revenue`, `list_rfp_specs`, `get_rfp_spec`, `list_rfp_categories`, `get_onsite_optimization_strategy`, `submit_lead`.
- **OpenAPI 3.1 spec** at `/openapi.json` for ChatGPT Actions, LangChain HTTP tools, Zapier, and other non-MCP clients (same ten tools).
- **ChatGPT plugin manifest** at `/.well-known/ai-plugin.json` (legacy compatibility, points at the OpenAPI spec).

## Unsupported on this surface

The public site does **not** expose:

- Live MCP tool execution against any retailer catalog
- Authenticated dashboards or per-tenant data
- Checkout, payment, order, loyalty, inventory, or pricing mutation
- Customer PII, order history, or support ticket data
- Any retailer-specific catalog, price, or stock signal
- Webhook callbacks or write APIs

Any agent attempting transactional, authenticated, or retailer-system actions
against this domain should fall back to the contact path above.

## Supported protocols (productized, partner-gated)

The following are productized capabilities offered to partner retailers under
contract. They are **not** open endpoints on this domain.

- **MCP (Model Context Protocol)** — retailer-owned discovery, recommendation,
  bundling, and ad-surface tools exposed to agents.
- **ATS (Agentic Transaction Surface)** — vector-search + structured product
  contract optimized for assistant retrieval.
- **UCP (Unified Commerce Protocol)** — transaction handoff contract between
  assistants and retailer commerce systems with retailer-controlled policy.
- **ACP (Agentic Commerce Protocol)** — checkout, payment, and order handoff
  contract; final authority remains with the retailer.

See `/.well-known/mcp.json` and `/.well-known/ucp.json` for sample contracts.

## Auth & consent boundary

- Public site: anonymous read, no auth required.
- Production protocol endpoints: partner-issued credentials, scoped per retailer,
  per surface, per action class (read vs. write vs. payment).
- No assistant or agent may transact on behalf of a shopper against a retailer
  without that retailer's explicit policy grant.

## Fallback when an assistant cannot transact

1. Surface the relevant retailer product or page link from this site.
2. Direct the user to the retailer's own checkout surface.
3. For partnership inquiries, route to https://particularaudience.com.

## Crawl etiquette

- Respect `/robots.txt`.
- Sitemap: `/sitemap.xml`.
- `llms.txt`: `/llms.txt`.
- No rate limit is published; please keep crawls under 1 req/sec.

- Audience-targeted search rules (coming soon, /search#control): global search, searchandising and Merchandising Studio rules can be limited to a PA customer segment or a CDP audience feed — e.g. small vs large dog owners searching "flea tick and worm" see different products boosted, buried or hidden.


## Audience Management (/audience-management)
Build segments from PA shopper data in-app, ingest audience IDs from CDPs, CRMs and third-party providers, and control which suppliers can target each audience. Suppliers target granted audiences without exporting underlying data.


## In-Store & Omnichannel Pricing (/in-store-pricing)
Activates Price Affirmation, Price Beat, Pricing Intelligence and Trade Media in store via electronic shelf labels (ESLs) and point of sale, using PA's Price Beat and Coupon APIs through the retailer's own ESL and POS providers, within retailer minimum price rules.


## Price Affirmation: pack-size and offer mapping
Price Affirmation checks offers first, then all comparable non-offer variants. It maps pack sizes like-for-like: retailer-approved exclusions for mixed-case (e.g. 'any six') and in-store-only single prices, and multi-buy vs equivalent competitor multi-buy comparison, so more products qualify without loosening evidence rules.

- Price Match for AI Agents (/conversational-price-match): PA adjudicates chatbot price-match/price-beat claims over MCP — evidence intake, browser agent, SKU matching, eligibility rules, promo calculator, decision bundle (approve/reject, rationale, coupon, confidence), escalation and audit log. Retailer owns UX, cart execution and final approval.