Best Practices for Shopping Agent

This guide outlines recommended practices for building high-quality Shopping Agents and configuring Knowledge Bases in retail and ecommerce environments. You can utilize the provided examples and technical settings to align your conversational experiences with brand objectives, increase purchase confidence, and enhance customer satisfaction.

Shopping Agent Roles and Customization by Industry

An effective Shopping Agent blends brand personality with sales enablement and customer experience goals. Tailor each agent’s tone, expertise, and engagement depth according to the customer journey stage from inspiration to purchase.

The Shopping Agent can be configured in any language supported by the language model, but it performs best in English.

Retail & Ecommerce

Use this setup as a reference for designing agents that drive online product discovery and conversion.

The configuration focuses on blending friendly, brand-aligned guidance with strong product expertise to help customers find the right items faster and shop with confidence.

• Agent Name: ShopMate

• Agent Identity: “You are a digital shopping assistant helping users explore, compare, and select products across categories. You provide helpful, brand-aligned recommendations that make online shopping simpler, faster, and more enjoyable. Whether users are looking for seasonal trends, bestsellers, or personalized suggestions, you guide them with accurate, relevant, and inspiring advice.”

• Communication Style: Friendly and Informative

• Message Length: Standard – Balanced level of detail

• Custom Instructions:

  • You are a helpful, brand-aware Shopping Agent designed to assist users with product discovery and purchase decisions. Adhere to the following behavioral rules at all times:

  • Stay within the shopping context. Focus on product discovery, feature comparison, price clarity, and shopping assistance. Use customer context (e.g., style preferences, occasion, season, budget) to tailor recommendations while respecting data privacy. Only reference verified product data, catalog details, or information included in the knowledge base. If product information is missing, acknowledge it and suggest an alternative or the next best step. If users ask about checkout, delivery, or refunds, redirect them to the official support channels. Never reference unavailable items, third-party websites, or competitors.

Knowledge Base Configuration by Industry

A structured Knowledge Base ensures that the Shopping Agent provides accurate, data-backed, and up-to-date product information. Below is an example configuration designed for retail and ecommerce use cases.

To set up a Knowledge Base for your Shopping Agent, refer to Create a Knowledge Base for Agents.

Retail & Ecommerce

This example illustrates how to structure a Knowledge Base for extensive product catalogs and frequently updated retail sites.

It emphasizes clean data collection, clear exclusion rules, and well-defined selectors to ensure your Shopping Agent delivers accurate, current, and brand-consistent information across all conversations.

• KB Name: ShopBase

• KB Description: Centralized Knowledge Base containing product details, pricing information, size guides, care instructions, category descriptions, and shopping FAQs. Supports conversational product discovery and comparison experiences.

• Source Name: Product & Content Web Crawl

• Crawling Options:

  • Main URL: https://www.example-shop.com

  • Proximity: 2

  • CSS Selectors to Wait to Load: .product-detail, .faq-section

  • CSS Selectors to Exclude: .cookie-banner, .popup, .footer, .advertisement

  • File Extensions to Exclude: .mp4, .pdf, .jpg, .gif

  • Headers – Key/Value:

    • Accept-Language: en-US

    • User-Agent: ShopAgentBot/1.0

  • Rule1: /faq/* And /careers/*Or /about-us/*

  • Rule2: /faq/*And /careers/*Or /about-us/*

Example Domain Applications

The following examples illustrate how Shopping Agents can be adapted for different retail domains.

Each domain highlights a distinct conversational focus, guiding you on how to balance product expertise, personalization depth, and knowledge coverage when designing your agent.