Shopify Agentic Storefront Product Data Checklist for Indian D2C Brands

What product data does an agent need?

Shopify's product-discovery guidance identifies Shopify Catalog as the primary method for agentic storefronts to receive product data. It lists core information such as titles, descriptions, options, images, price and availability.

Audit each SKU against the buyer's real decision:

| Data group | Minimum useful fields | Common D2C failure | | --- | --- | --- | | Identity | Stable SKU, product title, brand, category | Decorative title that omits the actual product type | | Variant | Size, colour, material, pack or capacity | Options embedded only in images or theme scripts | | Commercial | Current INR price, sale price, availability | Feed, schema and visible page disagree | | Fit and use | Dimensions, compatibility, intended use | Generic description with no selection criteria | | Visual | Clear primary image and useful alternate views | Text-heavy creative used as the only product image | | Fulfilment | Delivery, shipping, return and exchange facts | Policies exist but are difficult to associate with the item | | Trust | Accurate claims, care, ingredients or warranty | Unsupported superlatives and missing limitations |

For apparel, a title such as “Aura” may express the collection but not help an agent distinguish a cotton kurta from a synthetic dress. Preserve the brand voice in merchandising copy, but make the product identity and selection attributes explicit.

How should custom fields and metafields be handled?

Many mature Shopify stores keep key facts in metafields or custom theme sections. A shopper can see them, but a sales channel may not automatically use the intended field.

Shopify advises using Catalog Mapping when product information such as title, description or category is stored in custom fields. Create a field map with:

1. the source-of-truth system;
2. Shopify standard field or metafield;
3. Catalog destination;
4. validation owner;
5. refresh frequency; and
6. exception process.

Do not duplicate the same fact into several fields without ownership. Price, availability and variant relationships change. A stale duplicate creates a contradiction that is harder to detect than a missing value.

Do agents.md or llms.txt solve product discovery?

No. Shopify documents automatic discovery URLs including /agents.md, /llms.txt and /llms-full.txt for stores. It also says these discovery files are separate from Shopify Catalog and do not replace it.

Treat the files as a discovery aid. The higher-value work remains:

• complete and accurate Catalog data;
• crawlable product pages;
• consistent visible content and structured data;
• fast, usable mobile pages;
• clear shipping and return information; and
• stable inventory and pricing operations.

The Shopify AEO checklist covers answer structure and crawlability. This article focuses on SKU-level product data and agentic commerce operations.

How should product pages and structured data stay consistent?

Google's product structured-data guide recommends combining page markup and Merchant Center data where appropriate. It can use fields such as price, availability, shipping, returns and variants to understand a purchasable product.

For each priority product, compare four representations:

1. visible product-page content;
2. Shopify product and Catalog data;
3. Product and Offer structured data in the initial HTML; and
4. external merchant or channel feeds.

Google Merchant Center's landing-page requirements say the title, description, image, price, currency, availability and buy action should refer to the same product or variant as the submitted data. The same consistency discipline is useful for any machine-mediated shopping surface.

Use the AEO Readiness Checker for crawl and answer diagnostics, then perform a separate SKU sample because a site-level score cannot confirm every product variant.

What does a worked product-data audit look like?

Consider a fictional Indian apparel brand with 120 active SKUs. The figures are synthetic.

| Check | Passing SKUs | Coverage | Priority | | --- | ---: | ---: | --- | | Clear product type in title | 78 | 65% | High | | Complete material and fit fields | 66 | 55% | High | | Variant-specific image and availability | 102 | 85% | Medium | | Page, schema and Catalog price match | 116 | 97% | Urgent for four failures | | Shipping and return facts accessible | 120 | 100% | Maintain | | Stable SKU and variant identifiers | 118 | 98% | Urgent for two failures |

Method: export active products and variants, crawl each canonical product page, extract visible fields and JSON-LD, compare them with Catalog values, then weight failures by the prior 90 days of product revenue. A mismatch receives higher priority than a missing optional attribute because it can mislead the buyer.

Suppose the 20 products producing 62% of revenue include six unclear titles and three incomplete material fields. Fix those nine high-exposure records first, then update the shared template or data-entry rule so the defect does not return.

The score is a backlog tool, not a predicted visibility uplift. Do not claim that 97% completeness guarantees placement in an AI answer.

Which operational controls prevent catalog drift?

Create a weekly exception report for:

• active products with missing descriptions or categories;
• variants without a specific image;
• price or availability mismatches;
• duplicate or changing identifiers;
• products blocked from intended discovery;
• expired promotions still described in copy;
• missing shipping or return information; and
• AI-generated product content requiring applicable disclosure or review.

Google's product-data optimisation guidance stresses detailed titles, high-quality images, stable identifiers and current price and availability. Those are durable catalog practices even when a specific AI channel changes.

Assign one owner for catalog health across merchandising, development and growth. If every team owns one field but nobody owns the product representation, contradictions persist.

How should agentic storefront performance be measured?

Report direct outcomes separately from broader visibility:

| Layer | Metric | What it proves | | --- | --- | --- | | Eligibility | Products active and free of known data defects | Catalog readiness only | | Discovery | Channel impressions or queries where available | Observable visibility, not a sale | | Traffic | Sessions and engaged sessions by channel or referrer | Attributable visits | | Commerce | Orders, net revenue, returns, contribution margin | Recorded business outcome | | Incrementality | Controlled lift or credible holdout comparison | Additional outcome caused by the channel |

The GA4 AI Assistant measurement guide explains referral classification. The LTV:CAC cohort guide shows why acquisition quality should be reviewed beyond the first order.

For an implementation example, see HML's Shopify analytics case study. The CRO glossary is useful when product-data fixes must also improve the human purchase path.

What should a Shopify team do next?

Export the top 20 products by net revenue and audit their Catalog data, visible page, variants and structured data side by side. Fix contradictions first, then missing decision attributes, then lower-exposure catalog gaps. Record the owner and verification date for every rule.

If catalog structure, theme output and channel attribution cannot be reconciled, request a Shopify product-data and measurement review. HML's website development service can connect the data model, storefront output and conversion instrumentation before the brand expands AI-channel distribution.

Sources

1. Shopify Catalog and product discovery for agentic storefronts - Shopify Help Center; accessed 2 September 2026.
2. Shopify agentic storefronts - Shopify Help Center; accessed 2 September 2026.
3. Product structured data - Google Search Central; accessed 2 September 2026.
4. Tips to optimise your product data - Google Merchant Center Help; accessed 2 September 2026.
5. About landing page requirements - Google Merchant Center Help; accessed 2 September 2026.