AI platforms do not rank incomplete product pages lower. They skip them. ChatGPT Shopping, Google AI Overviews, and Perplexity crawl your PDPs, extract structured data, compare attributes across competing products, and pick whichever one they can describe most confidently.
That is a different game from SEO, and it has a different failure mode. There is no position 11. Your product either has the data an engine needs to make a claim about it, or it is absent from the answer entirely. Here are 15 changes, ranked from highest impact to lowest.
The 15-point checklist
01Add complete Schema.org Product markup
Implement JSON-LD Product schema including name, description, image, brand, SKU, offers (price, currency, availability), and aggregateRating. AI platforms parse structured data before they read your marketing copy. Without complete Product markup your PDP is effectively invisible to AI shopping engines, however good the page looks to a human.
02Include clear pricing with currency and sale indicators
Your schema and on-page content must show exact price, currency code, and any active sale pricing. AI platforms compare prices in real time. If your pricing is missing, ambiguous, or buried in JavaScript that crawlers cannot render, the engine skips you and recommends a product with verifiable pricing.
03Display real-time availability status
Mark products InStock, OutOfStock, or PreOrder using the Offer schema availability property. Platforms will not recommend a product they cannot confirm is purchasable right now. Stale availability, such as showing "in stock" for a sold-out item, damages trust signals across every AI surface.
04Add GTIN and MPN identifiers
Global Trade Item Numbers and Manufacturer Part Numbers let platforms cross-reference your product against manufacturer databases, marketplace listings, and review aggregators. These identifiers confirm the listing is legitimate and matches the exact item being asked about. Products without GTINs lose the cross-reference matching engines use to build confidence.
05Implement review aggregation with AggregateRating markup
Add AggregateRating schema reflecting your real review count and average score. Platforms treat verified, recent reviews as evidence when deciding which product suits a specific use case. Brands with structured review data get materially more mentions in AI shopping responses.
06Write descriptive image alt text with product identifiers
Use alt text including product name, colour, material, and view angle. "Women's merino wool crewneck sweater in navy, front view" gives image models far more than "IMG_4827.jpg." Over half of ecommerce sites fail at informational alt text, so fixing this puts you ahead of most of your category.
07Build structured attribute tables AI can parse
Present specifications in clean HTML tables with consistent headers: attribute name, value. Platforms extract tabular data far more reliably than specs buried in paragraph copy. A table covering dimensions, weight, materials, and compatibility gives engines exactly the format they need for detailed conversational queries.
08Add FAQ sections with Q&A schema
Add an FAQ section to each PDP built from real support-log questions and on-site search data, marked up with FAQPage schema. Platforms prioritise Q&A content because it mirrors how people actually ask. "Does this jacket work for hiking in light rain?" matches directly against your FAQ answers.
09Include brand and manufacturer information in structured fields
Add Brand and manufacturer properties to your Product schema. Platforms use brand identity to disambiguate products, connect items to reputation signals, and match queries like "best [brand] for [use case]." Missing brand data excludes you from brand-specific recommendations entirely.
10Create comparison-ready specification tables
If shoppers compare your product to alternatives, address "vs" queries directly on the PDP. Platforms pull from pages offering honest, structured comparisons. Acknowledging trade-offs builds trust with AI systems, because it signals the content is informational rather than purely promotional.
11List ingredients or materials for research-driven categories
For beauty, skincare, supplements, food, and home goods, include the complete ingredient or material list. Platforms serving health-conscious or sustainability-focused shoppers need this to answer "is this moisturiser fragrance-free?" or "does this contain organic cotton?" If it is not on your page, they recommend the product that does provide it.
12Add size and fit guides with structured data
For fashion and apparel, include size charts and fit guidance with structured markup. Assistants increasingly handle "what size should I get?" by extracting measurements and fit descriptors from PDPs. A page with structured sizing answers directly. A page without it gets passed over.
13Show delivery estimates with geographic specificity
Include shipping timeframes and delivery estimates that specify regions or postcode ranges. Platforms answering "can I get this by Friday?" or "does this ship to Canada?" need delivery data on the page. Without it you drop out of every response where delivery speed matters.
14State return policy information explicitly
Add return window, conditions, and process directly on the PDP, not only in a site-wide policy page. Platforms extract return information when shoppers ask "can I return this?" during research. Clear return policies earn more confident recommendations because they reduce purchase risk.
15Write conversational meta descriptions
Rewrite meta descriptions to read as a concise answer to "what is this product and who is it for?" rather than stuffing keywords. Platforms use them as a quick summary signal when deciding how to present your product.
Scoring this at catalog scale
Running this checklist by hand across thousands of SKUs is not practical. Alhena scores your PDPs against these exact criteria automatically, analysing each page, extracting every attribute, and showing which elements are missing, which are incomplete, and how each gap affects visibility across ChatGPT, Perplexity, and Google AI. Instead of guessing which products need work, you get a prioritised list.
SKU-level monitoring then tracks whether platforms are recommending the right products with correct specs, not just whether your brand gets mentioned. Tatcha's AI concierge drove 3x conversion and a 38% AOV lift after this work. Go live in under two days, no engineering tickets required.
Every missing element is a reason to recommend your competitor
AI platforms do not penalise incomplete PDPs with lower rankings the way search engines do. They skip your product and recommend one that provides the data they need. Every absent schema field and unstructured specification is a reason to choose someone else.
Brands treating PDP work as a revenue channel rather than an SEO task will capture the fastest-growing discovery channel in ecommerce. The rest will keep losing recommendations they never knew they were eligible for.
See where your PDPs fall short
Get a full AI visibility audit across your catalog, scored against all 15 items.
Frequently asked questions
It crawls each PDP, extracts structured data, attributes, schema markup, and content completeness, then scores every element against what ChatGPT, Perplexity, and Google AI surfaces require. You get prioritised reports showing which fields are missing or incomplete and how each gap affects your score, so teams can sequence PDP work by revenue impact.
Complete Product schema, verified pricing with currency codes, real-time availability, and GTIN or MPN identifiers. Those four are the foundation tier for a reason: without them the engine cannot form a confident claim about your product, so it moves on. This checklist covers the on-page execution. For the field-by-field ranking of the underlying catalog data, see the six product data fields that decide AI recommendations.
Yes. SKU-level monitoring tracks whether ChatGPT, Perplexity, and Google AI are recommending the right products with correct specs and pricing, not just whether your brand is mentioned. That catches inaccurate recommendations and missing products before they cost revenue.
Most brands see measurable change within two to four weeks of implementing the foundation tier: complete Product schema, pricing markup, and availability status. Tracking your score over time is what ties specific fixes to increases in AI-driven recommendations.
Yes, and that is the efficiency argument for doing it. The same structured data that earns recommendations from ChatGPT also powers on-site assistants. One set of improvements pays out on both channels. For what your on-site assistant specifically needs, see catalog data quality and the accuracy ceiling.
