Imagine a customer asks an AI assistant: “Find a cabin-size suitcase under 3 kg, with a warranty, for less than 5,000 baht.”
In a traditional Customer Journey, a business may focus on ranking its Product Page so the customer clicks through and evaluates the product.
With AI-assisted shopping, an additional stage appears.

The system may compare several products before the customer visits a store, using attributes such as Weight, Dimensions, Price, Warranty and Availability.
If your page only says: “Premium suitcase designed for every journey”
while a competitor clearly provides Weight, Dimensions, Material, Warranty, Current Price, Availability and Delivery Information, the competitor provides more information that can support comparison.

This does not mean Structured Data guarantees an AI recommendation.
It means Product Information is becoming a more important part of the infrastructure behind AI-assisted Customer Journeys.

How well AI can compare your product partly depends on how clearly your business describes it
Product discovery is no longer limited to typing a Keyword and opening several websites.
A customer can ask an AI system: “Find running shoes for wide feet under $100.”
or: “Which coffee machine is suitable for a small apartment and easy to clean?”
ChatGPT Shopping can currently present Product Information and shopping options based on user intent, with information such as Price, Reviews and Availability contributing to the experience. OpenAI also supports structured Merchant Product Feeds designed to keep Product Information such as Price and Availability current.
Google similarly uses Product Structured Data and Merchant Data to understand information including Price, Availability, Variants, Shipping and Returns.
For SMEs, the strategic question is therefore not only: “How do we get AI to mention our brand?”
It is: “If an AI system had to compare our product with competitors today, have we provided enough accurate information for that comparison?”

AI Commerce shifts the question from “Can customers find us?” to “Can systems understand and compare us?”

Traditional Search might begin with:
“Women's running shoes”
AI-assisted Shopping can begin with:
“Find lightweight women's shoes for a 10K runner with wide feet, under $120, preferably available for immediate delivery.”
That requires understanding several attributes:

• Product Type
• Intended Use
• Fit
• Size
• Weight
• Price
• Availability
• Delivery

OpenAI's current Commerce Product Feed supports core data such as Product ID, Title, Description, URL, Image, Brand, Price and Availability, as well as Product Category, Material, Dimensions and Variant information.

Product Data therefore increasingly needs to support three tasks at once:
Human Understanding
Search Discovery
Machine Comparison

Layer 1: Make it clear what the product actually is

Start with a consistent Product Identity:

• Product Name
• Brand
• Product Category
• SKU or Stable Product ID
• GTIN / Barcode where applicable
• MPN where applicable
• Product Variant

Google recommends providing accurate unique identifiers such as GTIN, Brand and MPN where they exist because they help its systems understand which product a merchant is offering.

OpenAI's Product Feed similarly supports stable Item IDs, GTINs, MPNs and Brand information.
For an SME manufacturing its own products without GTINs, this does not mean the product cannot be represented.
It does mean the business should maintain stable internal Product IDs and consistent naming.
Avoid referring to one product inconsistently as:
Eco Bottle Pro
Bottle Pro Eco
EcoBottle Pro Model
EB-Pro
without a clear system showing that they are the same product.

Layer 2: Product Descriptions should explain differences, not only make claims

Statements such as:
“Premium quality”
“Perfect for your lifestyle”
“The ultimate choice”
provide little comparison information.
‍
If a customer asks “Which model fits on a small desk?”
Dimensions are more useful than the word Premium.
A decision-useful Product Description should explain:
What is the product?
Which Use Cases is it designed for?
What are its important Features?
What limitations matter?
How is it different from another model?

Marketing Copy still matters. But its job is different from Product Facts.
Marketing Copy builds interest. Product Facts support comparison.

Layer 3: Put important Specifications into actual data fields

Decision-relevant information may include:

• Dimensions
• Weight
• Material
• Capacity
• Compatibility
• Size
• Color
• Model
• Warranty
• Technical Specifications
• Usage Requirements
• Package Contents
• Verifiable Certifications

OpenAI's Commerce feed supports physical attributes including Material, Dimensions and Weight. Google's Merchant product attributes also support fields for Size, Material, Product Dimensions and Product Weight in relevant contexts.

If critical specifications appear only inside an infographic or packaging image, consider making the same information available as text or Structured Product Data.
A useful principle is: Information customers use for comparison should be represented as data where practical.

Layer 4: Treat Variants as distinct purchase options

One Product may vary by:
Color
Size
Capacity
Material
Package
Model

A Product Page might say “1,990 baht In Stock”
while the real situation is:
Black / Size M = 1,990 baht and In Stock
White / Size XL = 2,290 baht and Out of Stock
Without clear Variant Data, the wrong Price or Availability may be attached to the wrong option.

OpenAI's Google-compatible Product Feed requires one Product or Variant per row and supports grouping information for attributes such as Color, Size and Material.
Google also supports Product Variant Structured Data so its systems can better understand products that belong to the same parent Product.
Product ≠ Variant and:
Variant A Availability ≠ Variant B Availability

Price needs to be the current Price, not a Price that used to be correct

Price changes quickly, and inconsistent pricing can damage trust.
Google Merchant Center states that submitted Price and Currency should match the Landing Page, Structured Data and Checkout.
OpenAI's Product Feed supports Regular Price, Sale Price and Promotion Effective Dates.
Check consistency across:
Product Page
Shopping Feed
Structured Data
Marketplace Feed where relevant
‍

Checkout
If the Website says 990 baht while the Product Feed says 890 baht, the underlying information system has a problem regardless of which platform displays it.
AI Commerce readiness therefore requires both: Data Completeness and Data Freshness + Consistency

Availability can be as important as the Product Description

A customer may ask “Which option can arrive tomorrow?”
A product with ideal Specifications but no Stock may no longer match the request.

Google Merchant Center supports Availability statuses including In Stock, Out of Stock, Preorder and Backorder and requires Merchant Feed Availability to align with the Landing Page and Structured Data.
OpenAI's Product Feed also requires Availability and supports several stock states, including In Stock, Out of Stock, Pre-order and Backorder.
For SMEs with fast-moving inventory, ask How quickly do the Store System, Website, Feed and Inventory Data synchronize?

Shipping is part of the Product Value, not just an operational detail

Consider two products with the same Price.
Product A ships free and arrives tomorrow.
Product B costs 300 baht to ship and arrives in seven days.
For some Customer Needs, those are not equivalent offers.

Google Merchant Listing Structured Data supports Shipping Cost and Delivery Time information, while Merchant Center and Merchant APIs can also carry shipping rules and related information.
Google's 2026 Merchant Center specification also added additional Product-level shipping attributes, including Handling Cutoff Time and Minimum Order Value.
At minimum, clarify:

• Shipping destinations
• Shipping cost
• Free-shipping conditions
• Processing time
• Estimated delivery
• Store pickup where available

Returns and Warranty may become part of the comparison

Two products priced at 3,000 baht may appear equal until the customer learns that:
One can be returned for 30 days.
The other is Final Sale.
Or:
One includes a two-year Warranty.
The other has no clearly stated Warranty.

Google supports Merchant Return Policy information in Structured Data, including Return Windows, Methods and Fees.
In 2026, Google's Merchant API also added more granular Offer-level return fields covering Return Windows, Methods, Outcomes and Fees.
Replace vague policies such as: “Returns subject to company conditions”
with information customers can actually use:
How many days?
Which items qualify?
Who pays return shipping?
Is Exchange available, or Refund only?

Your Website, Structured Data and Product Feed should not describe different realities

Think of Product Information in three layers:

1. Human-readable Content
   What customers read on the Product Page.
2. Structured Data
   Information represented in machine-readable fields.
3. Product Feed
   Information sent to Commerce and Shopping Platforms.

Google recommends Product Structured Data for information including Price, Availability, Shipping, Returns and Variants and also supports Merchant Center as a product data channel.
OpenAI accepts Structured Product Feeds to help ChatGPT index products with current Price and Availability information.

Avoid situations such as:
Website = In Stock
Structured Data = Out of Stock
Feed = Pre-order
or:
Product Page = 1,590 baht
Feed = 1,790 baht
Consistency needs to become an operating process, not a one-time website project.

AI needs to understand not only “What is it?” but also “Who is it for?”

Customers will not only ask: “What features does this product have?”
They may ask: “Is it suitable for me?”
Useful information may include:

• Intended Use
• Suitable User
• Compatibility
• Size / Fit
• Skill Level
• Environmental Requirements
• Usage Context
• Limitations

For example, a Laptop should clearly provide facts such as:
RAM
Storage
Weight
Display
Ports
Battery specifications where supported
rather than relying on an unsupported statement such as:
“The best laptop for everyone who works remotely.”
“Best” requires a Comparison Base and Criteria.
Clear Product Facts are easier to evaluate.

Reviews can provide evidence, but businesses should not manufacture evidence for AI

Reviews can help customers understand how other users experienced a Product, and Google Product Structured Data supports Review and Aggregate Rating information under its eligibility and content requirements.
However:
Average Rating ≠ Product Fit for everyone
Review Frequency ≠ Population Opinion
Most Mentioned Feature ≠ Most Important Purchase Driver

And fabricated or misleading Reviews create more risk than sustainable advantage.
The stronger strategy is to improve the provenance and context of Customer Evidence rather than generating text that merely looks like evidence.

Do not make the Website unreadable just to make it “AI-friendly”

Machine-readable Product Data does not require turning the Product Page into a technical database.
A useful page can still present:
Product Name
Direct Description
Key Benefits
Key Specifications
Comparison Information
Price
Availability
Shipping
Returns / Warranty
FAQ addressing real purchase questions
with Structured Data operating underneath.
The goal is not AI-readable instead of Human-readable
It is Human-readable + Machine-understandable + Evidence-consistent

SMEs can start with a Product Data Audit instead of a major AI project

If you sell 50 products, you do not need to begin with a complex AI Commerce transformation.
Select your Top 10 Products and review:

1. Identity
   Are Product Name, Brand, SKU and Variants clear?
2. Description
   Can someone understand what the product is and what Use Case it serves?
3. Specifications
   Are the attributes customers actually compare available?
4. Price
   Is it consistent across Channels?
5. Availability
   Can current Stock Data be trusted?
6. Shipping
   Are Cost and Delivery Time clear?
7. Returns / Warranty
   Can customers understand the post-purchase risk?
8. Structured Data / Feed
   Can key information be provided in formats Commerce Platforms understand?
9. Freshness
   Who owns the update process?
10. Consistency
    Do the Website, Feed and Checkout tell the same story?

Start Product Information Readiness with the Business Question, not the Schema

Do not begin with: “Which Schema fields should we implement?”
Start with the Customer Question.

For example “How would a family choose a water purifier?”
Then identify the Decision Criteria:
Capacity
Filter Type
Replacement Cost
Installation Requirement
Dimensions
Warranty
Price

Then ask:
Which information do we already have?
Which should appear on the Product Page?
Which should become a structured attribute?
Which Claim still lacks evidence?
‍

This keeps Product Data connected to the Customer Decision rather than turning it into an isolated technical exercise.

Complete Product Data does not guarantee that AI will recommend you

This limitation matters.
OpenAI currently states that ChatGPT Shopping selects products based on Relevance to the user's intent and context and may consider factors such as Price, Reviews, Availability and Ease of Use. Merchant presentation can also consider factors such as Availability, Price, Quality and merchant characteristics, and the system may evolve over time.

Google similarly states that eligible Structured Data does not guarantee a particular Search enhancement will be displayed.
Therefore:
Structured Data ≠ Guaranteed Visibility
Product Feed ≠ Guaranteed Recommendation
AI Citation ≠ Guaranteed Purchase
The objective is to improve the quality of the information available for both systems and customers to make a decision.

The takeaway: In AI Commerce, Product Information becomes part of the Product Experience

As customers use AI to assist with product selection, competitive readiness will involve more than having a Website or ranking for traditional Search Keywords.
Another question becomes important: How well can a system understand your Product?

Google and OpenAI both currently operate Commerce systems that use structured Product Information such as Product Identity, Description, Price, Availability, Variants and purchase-related information to help index, match or display products in Shopping Experiences.
A useful SME framework is: Identify → Describe → Specify → Differentiate → Update → Verify
Identify
Make the Product identity clear.
‍
Describe
Explain its Use Case and Value.
‍
Specify
Provide attributes customers can compare.
‍
Differentiate
Separate Variants and Model differences accurately.
‍
Update
Keep Price, Availability and Policies current.
‍
Verify
Check consistency across the Website, Structured Data, Product Feed and Checkout.
‍
AI Commerce interfaces and platforms will continue to change.
A more durable principle is simpler: If a system is going to help a customer decide, your Product Information needs to make it clear what the product is, who it is for, how it differs, what it costs, and whether it can actually be purchased under the stated conditions.