Half the questions we get from Shopify Plus merchants right now are about the same thing: "Are we visible to ChatGPT? Are our products showing up in Perplexity? Does Google AI Mode cite us?" The honest answer is that most stores are not optimised for any of this, and the fixes are concrete and technical — not the vague "update your meta descriptions" advice that dominates the first page of search.

This is what we actually change on client stores in April 2026, in the order we change it.

Product Data Is the Whole Game

AI shopping assistants do not rank pages — they retrieve structured product facts. Your visibility in ChatGPT Shopping, Perplexity, Gemini, and Google AI Mode is almost entirely a function of how complete and unambiguous your product data is. The single biggest win is filling every recommended attribute on every product: GTIN or MPN, brand, material, colour, size, gender, age group, Google product category, item condition.

Shopify Metafields are the mechanism. Define a metafield set under product.custom that maps every attribute an AI engine might want to a structured field — and then enforce it with bulk editor or a small admin app. Do not let merchandisers leave fields blank when a product ships.

The Attribute Baseline We Enforce

Required on every product

GTIN or MPN plus brand
Google product category (full path)
Colour, size, material
Product identifier metafield
Canonical URL

Required on apparel and beauty

Gender, age group, pattern
Fit / shape
Country of origin
Allergen and ingredient lists
Care instructions

Structured Data That AI Engines Actually Read

Every product page needs a Product JSON-LD block with the same attribute fidelity as your metafields. The difference between "indexed" and "cited" in AI results is usually this block. Include offers with priceValidUntil, itemCondition, availability, and a real priceCurrency that matches the market. Include aggregateRating only if it is real — AI engines have started detecting fabricated ratings and suppressing citations.

Add FAQPage schema to PDPs with answers to the three questions customers actually ask about that product. LLM retrieval is heavily biased towards pages that already present information in Q&A form because that is the format the engine outputs in.

Merchant Feed Hygiene

Shopify syncs product data to Google Merchant Center, Bing Merchant Tools, Meta Catalogues, and (from late 2025) the Shopify Agentic Commerce channel that ChatGPT uses. Broken feeds are invisible to AI shopping. Weekly checklist:

Google Merchant Center: zero disapprovals, zero "limited performance" warnings on bestsellers.
Bing Merchant Tools (feeds ChatGPT): same zero-disapproval rule.
Meta Catalogue: not just connected, but verified — Instagram Shop and ChatGPT's commerce partners use Meta's identity graph.
Shopify Markets: correct regionCode on every product for multi-region availability.

Content That Gets Cited (GEO)

AI engines cite pages that directly answer specific questions in the first two paragraphs. They do not reward introductions, lore, or "in this guide we'll cover…" scaffolding. For every high-intent query you want to win, publish a page whose first 150 words state the answer, followed by the evidence. We retrofit our own posts with this pattern and watch citation share in Perplexity climb within three weeks.

The practical implication for Shopify stores: PDP descriptions should lead with the five product facts a buyer needs before scrolling (material, fit, size range, price, return policy). Category pages should lead with a one-paragraph definition of the category plus a decision framework. "Buying guide" content pages should open with a TL;DR the model can lift.

Agentic Commerce — What Is Real in 2026

Shopify shipped Agentic Storefronts in late 2025. Merchants on Plus get a channel that exposes their catalogue to ChatGPT, Perplexity, and a growing list of agent partners through a single connection. Orders land in the normal admin with channel = "Agentic" on the order. Attribution flows back cleanly and you keep customer ownership.

What is not real yet: Instagram and TikTok's AI discovery layers are still closed betas in the UK and EU, Apple Intelligence is iOS-only and US-first, and the "AI-native storefront" vendors advertising on LinkedIn are mostly re-skinning Algolia or Klevu. Stick with Shopify's native integration first.

The Unsexy 80%

If you only do three things: complete every product metafield, ship valid Product JSON-LD on every PDP, and keep Google/Bing Merchant feeds at zero disapprovals. Everything else in this article is secondary until those three are in place. Most stores we audit fail on at least one of them.