Google Shopping guide

Prepare your product data for conversational shopping.

Conversational shopping illustration showing a natural-language question connected to structured product attributes and relevant product choices.

Last updated:

Conversational shopping lets shoppers describe what they need in natural language and receive relevant product recommendations. This guide shows ecommerce teams how to audit, map, enrich and validate Google Merchant Center product data for these AI-assisted shopping experiences.

Start with a free auditBook a practical conversation
Merchant Center attributesConversational shopping readinessAI-assisted product discovery

What is conversational shopping, and why does product data matter?

Conversational shopping is a shopping experience in which a customer asks a natural-language question or describes a need, and an AI system identifies, compares or explains relevant products. The quality of the response depends on product information that is accurate, structured and specific enough for the system to understand.

Google describes conversational attributes as optional product data that can help AI systems and conversational agents understand your products’ specific nuances. They complement your existing Merchant Center data; they do not replace the core product specification. Use the Merchant Center Excellence Report guide to assess that core product-data foundation before adding conversational attributes.

The Google Shopping Graph is Google’s data set of the world’s shopping information. It brings together product details such as colours, sizes, materials, reviews and availability so Google Search, including AI Mode, can understand products and respond to more specific shopping needs. Conversational attributes give your products richer, more structured context to contribute to that experience alongside the information you already provide through Google Merchant Center and Google Ads.

Google has published a retailer example featuring lululemon, where conversational attributes were incorporated into relevant AI Mode product recommendations during testing and helped provide shoppers with more precise answers.

Read Google’s conversational attributes guidance for the current field definitions and submission options.

Supported attributes

Conversational attributes you can prepare.

  • question_and_answer: product-specific questions and answers.
  • item_group_title: a shared title for products in the same variant group.
  • variant_option: the option represented by a particular variant.
  • document_link: approved product documentation links.
  • related_product: native complementary or alternative product relationships from supported platforms, plus mapped custom relationships.
  • popularity_rank: an agreed ranking signal with a defined scale and freshness expectation.

Step 1 · Start with the audit

See which conversational attributes could improve your products.

Use the FeedOps audit to identify which conversational attributes are recommended but missing. This gives you a practical starting point for improving the product information Google can use in Search and AI Mode.

1. Check recommendations

Review the audit for conversational attributes that could improve your product data but are not yet present.

2. Choose the opportunity

Prioritise the attributes that are most relevant to your products, variants and shoppers’ questions.

3. Move into implementation

Use the recommendations to decide which website fields to map and where the FeedOps playbook can add useful context.

Step 2 · Map

Map your website data to the recommended attributes.

  1. Map the relevant built-in or custom fields from your website into the attributes recommended by the audit.
  2. Confirm which products qualify, how values are normalised, and how product and variant relationships should be represented.
Related products are supported natively. Supported platform connections can automatically provide complementary and alternative product relationships through the related_product field. This includes Shopify, Magento, BigCommerce, WooCommerce and more. If your relationships are stored in custom fields, FeedOps can map those too.

Step 3 · Enrich

Use the FeedOps playbook to enrich the mapped facts.

Once your website data is mapped, the FeedOps playbook makes it more useful without inventing product facts. It identifies useful questions, writes concise answers from approved evidence, standardises group titles, and expresses variant options consistently across a product family.

Questions and answers

Identify the questions shoppers are likely to ask, then answer them from approved product facts and specifications, documentation or seller expertise.

Item group titles

Create a clear shared title for a product family, preserving the group identity while keeping individual product titles and IDs intact.

Variant options

Normalise the option name and value for each variant—such as colour, size, memory or width—so Google can distinguish the choices within the group.

Google’s example shows these fields working together: an item group title identifies the product family, variant options describe the specific version, and question-and-answer pairs provide decision-making context.

Review the generated feed output and resolve any mapping or content issues before rechecking the audit.

Quality and consistency

Keep every product detail clear and dependable.

FeedOps validates conversational attributes against your product data so the information presented to Google is consistent, explainable and useful to shoppers. Clear rules for eligibility, formatting, relationships and product variants help you scale quality across your catalogue.

  • Keep output consistent and explainable from the product data on your website.
  • Use approved product facts to create useful answers and variant details.
  • Resolve URL, product identity and attribute issues, then confirm relationship and range checks for the relevant attribute.
  • Use the audit to identify any remaining opportunities to improve coverage.

Step 4 · Recheck

Use the audit to confirm the rollout.

When your mapped fields have been enriched by the playbook, run the FeedOps audit again. Check that the recommended conversational attributes are now present, valid and supported by the product data on your website.

If an attribute is still missing, return to the mapping step and check whether the source field is available, whether a custom field needs to be mapped, or whether the product is eligible for that attribute.

Readiness checklist

Before you roll out.

  • The audit recommendations for your products are understood.
  • Relevant built-in and custom website fields are mapped.
  • Questions and answers, item group titles and variant options have been enriched from approved product data.
  • The generated feed output has been reviewed.
  • The audit has been rechecked and remaining gaps have an explanation or next step.

Common questions

Conversational Shopping questions, answered.

Short answers to the questions ecommerce teams most often ask about conversational shopping product data.

What is conversational shopping?

Conversational shopping is a shopping experience in which a shopper describes a need or asks a question in natural language and an AI system uses structured, accurate product data to identify and explain relevant products.

What are Google conversational attributes?

Google conversational attributes are optional product data fields that help Google understand product nuances for conversational shopping experiences, alongside the product data already provided through Merchant Center.

Do I need custom fields to use conversational attributes?

No. Supported platform connections such as Shopify, Magento, BigCommerce, and WooCommerce can provide available built-in fields and related-product relationships automatically. You can also map custom fields from your website when they contain useful product or variant information.

Which conversational attributes does the FeedOps playbook enrich?

The FeedOps playbook can enrich question and answer content, item group titles, and variant options from mapped product facts and approved product evidence.

Do conversational attributes replace standard Merchant Center attributes?

No. Conversational attributes complement the core product data already submitted to Google Merchant Center, including identifiers, titles, descriptions, price, availability and category-relevant attributes.

Do conversational attributes guarantee visibility in Google AI shopping experiences?

No. They provide additional product context, but they do not guarantee that a product will appear or be recommended. Product eligibility, data quality, relevance, availability and other Google systems still apply.