Product Feeds in 2027: Why Does Product Data Now Matter Beyond Google Shopping?

12min.

Comments:0

17 November 2026

Product Feeds in 2027: Why Does Product Data Now Matter Beyond Google Shopping?
Just a few years ago, a marketer could think of a product feed as the file that powers Google Shopping. It was a simplification, but it reflected how many companies worked fairly well. The feed contained the product name, ID, price, image, link and availability, and the main question was: are the products displaying correctly in Merchant Center? In 2027, it will increasingly be a structured source of data about the entire catalog, used by Google, ads and AI. Now the question is: is the data about your products good enough for different systems to understand, compare and recommend them? This changes the way we think about the feed, and with it, who in the company is responsible for it. At the end of the article, you'll find a checklist to help you check whether your feed is ready for this new environment.

12min.

Comments:0

17 November 2026

Key Takeaways

  • A feed is now more than a file for Google Shopping. The same product catalog powers campaigns, search and AI answers, so its role goes beyond advertising.
  • For AI to recommend a product, it first has to know about it. It needs information about the offer, price, availability, variants and seller, and the feed is one of the places it can get that information from. In ChatGPT, direct feeds are currently available only to approved partners.
  • Inconsistent data costs more the more systems use it. A different price in the feed than on the page or in the schema means conflicting information about the same product.
  • Complete fields are only the beginning. What counts is whether the data is complete, accurate, consistent and up to date, and whether the title, description and identifiers make it possible to tell the product apart from similar ones.
  • The whole organization is responsible for the feed, not just the advertising team. Errors in prices, variants or GTINs originate in e-commerce, IT and the product database, so that’s where they need to be fixed.

What Exactly Is a Product Feed in 2027?

A product feed is a structured set of product data.

It can include, among other things:

  • an identifier,
  • a title,
  • a description,
  • the brand,
  • the category,
  • the price,
  • availability,
  • images,
  • the variant,
  • the GTIN,
  • the MPN,
  • shipping information,
  • promotions.

Its role is changing, though. In 2027, it’s worth treating the feed as a structured source of data about the entire catalog that Google Search, Shopping, ad campaigns, AI Shopping, ChatGPT, agentic systems and other commerce platforms can use.

A feed can be a file, a data source, an API or another form of structured catalog delivery. OpenAI currently accepts product data both through file uploads and through an API.

Why Was the Feed Equated With Google Shopping for So Long?

Historically speaking, for most stores, Google Merchant Center was simply the place where product data generated the most revenue. Because that’s the channel where the main advertising budgets went, many marketers came to see the idea of a feed as inseparable from the Google ecosystem.

Structured product data was the main fuel for:

  • product ads,
  • Performance Max campaigns,
  • free product listings,
  • dynamic remarketing.

What Has Changed? AI Needs Product Data

AI can answer the question “which TV should I buy?”, but to do it well, it needs access to information about:

  • which products exist,
  • how much they cost,
  • whether they’re available,
  • how they differ,
  • which variant is being sold,
  • who the seller is.

The growth of AI Shopping therefore means machine-readable product data is becoming more important.

Google itself stresses that AI-powered shopping experiences are fueled by the core product data merchants provide, and that an incomplete or disorganized feed can limit how easily products can be found.

AI doesn’t create product data. It has to receive, read or verify it from somewhere. The feed is one of those sources.

The Google Shopping Graph Shows How Much Product Data Matters

One example of infrastructure built on product data is the Google Shopping Graph, a database of product information used in Google’s shopping features.

Google has reported more than 50 billion product listings in the Shopping Graph, covering prices, inventory and reviews, among other things. More than 2 billion of those listings were said to be refreshed every hour. Google later put the figure at more than 60 billion listings.

The size of this database shows how much up-to-date product information matters. If the price, availability or variant is wrong, the problem can carry over into later stages of the shopping experience.

This also matters from the perspective of the purchase process. When AI helps users narrow down their choices, compare offers and find a seller, it needs data that lets it tell products apart and assess whether specific offers are current.

Does ChatGPT Also Use Direct Product Feeds?

Yes, because OpenAI is currently developing product feeds as part of Agentic Commerce. The official documentation states plainly that a structured product feed helps ChatGPT show the right products with up-to-date information.

OpenAI lets merchants send data such as:

  • the product ID,
  • the title,
  • the description,
  • the URL,
  • the brand,
  • the price,
  • availability,
  • variants,
  • media,
  • seller information.

There’s an important limitation to keep in mind, though: sending product feeds directly to ChatGPT is currently available to approved partners. This doesn’t mean every store can connect its catalog to ChatGPT on its own.

Interestingly, OpenAI Also Supports Google-Compatible Feeds

OpenAI officially describes the option of using a data profile based on fields that are familiar from Google feeds. This option requires standard attributes such as ID, title, description, link, image_link, availability, price and brand.

What does this mean?

Well-organized product data is increasingly becoming a portable business asset.

If your store already has:

  • correct IDs,
  • good titles,
  • complete attributes,
  • up-to-date prices,
  • correctly set up variants,

it’ll be much easier and faster to adapt your product catalog to every new platform that comes onto the market.

Is the Product Feed Becoming the “Single Source of Truth”?

You need to be careful with this term, because the feed isn’t always the single source of truth. Depending on the store’s architecture, the primary data source may be an ERP, PIM, CMS or another product database. Above all, the feed should consistently reflect information coming from the right source.

The same product can be shown at the same time in the online store, in Google Merchant Center, in the feed, in the structured data on the page, on Meta, on marketplaces and in systems used by AI platforms. If the information doesn’t match, there’s a problem.

For example:

  • the price on the product page is $799,
  • the feed sends a price of $899,
  • the structured data shows $849,
  • the page shows the product as available, but in the feed it has an out of stock status,
  • Merchant Center, meanwhile, reports it as in stock.

In this case, the individual systems receive conflicting information. The more places use product data, the more its consistency matters.

Why Are Price and Availability Becoming Real-Time Data?

Price and availability are the pieces of information that change fastest in e-commerce. Google puts a strong emphasis on keeping them strictly consistent and warns explicitly that relying solely on dynamically generated Product Schema can lead to less frequent and less reliable crawls by Google Shopping. That’s a critical risk for data that has to reflect the current state of the offer.

The makers of ChatGPT take an equally strict approach. OpenAI clearly requires product data to be updated as soon as a price or stock level changes.

Ultimately, a product feed stops being a file that’s updated “every now and then.” For a large e-commerce business, it has to be part of a continuous data flow that ensures all advertising systems and AI assistants always work with the store’s actual offer.

How Often Should You Update Your Feed?

There’s no single update frequency that works for every store.

How often you update it should depend on:

  • the size of your catalog,
  • how often prices change,
  • stock turnover,
  • the promotions you run,
  • product availability.

OpenAI’s documentation recommends updating full datasets at least once a day. The Agentic Commerce Protocol (ACP) documentation also describes a model in which the full feed is sent, for example, once a day, while changes that need to be reflected faster are sent through the API during the day.

This approach can be especially useful when the offer changes dynamically. In consumer electronics and home appliances, the availability of a specific model can change much faster than its description or technical specifications. Likewise, the price may change because of a promotion or an update to the pricing policy.

Feed Quality Is No Longer Just About Filled-In Fields

Traditionally, a marketer checks: are the required attributes filled in? In 2027, you also need to ask: does the data actually help the system understand the product? Look at these examples:

Feed element Technically correct Information that helps explain the product
TV title Samsung TV 65 Black Samsung OLED 65″ S95F 4K 120 Hz Smart TV
Smartphone title Galaxy Samsung Galaxy S26 256 GB Black
Description The perfect TV for any home. 65-inch 4K OLED TV with a 120 Hz panel, HDMI 2.1 and support for gaming features.

The Product Title Is Taking On a New Role

When the feed mainly serves Search and Shopping ads, the title is often built around the brand, product type, model and key feature. That still makes sense, but with AI, the title should above all make it clear what the product is and which variant it is. We cover the details of product identification in our guide to product pages for consumer electronics and home appliances.

Descriptions Should Describe, Not Advertise

In the description, we’re not looking for slogans like “A revolutionary smartphone for the most demanding users!” but for specifics: “Smartphone with a 6.4-inch OLED screen, 256 GB of storage and a 50 MP main camera.” In its best practices, OpenAI recommends concise, factual descriptions that help explain the product.

Why?

Because the data is there to help the system recognize the product, compare it and match it to a need, not to “wow the algorithm.” Don’t overdo it in the other direction either: the feed isn’t the place for keyword stuffing. The system doesn’t need special descriptions written for AI. It needs data that’s specific, unambiguous and factual.

You’ll find more about product content for shopping agents in our guide on how to prepare e-commerce content for AI shopping agents.

The Product Variant Is Just as Important as the Product Itself

To continue the example, consumer electronics and home appliances are full of variants. The same laptop may come in 16 GB / 512 GB, 16 GB / 1 TB or 32 GB / 1 TB versions, and the price and availability may differ.

That’s why OpenAI recommends modeling variants at the row level: a stable parent product ID and a unique ID for each purchasable option. Fields such as the title, URL, description, media, availability and price should contain information specific to the given variant. Google also requires unique IDs for variants in structured data.

Why Will Product Identifiers Become Even More Important?

If the same product appears in a store, on Google, in ChatGPT, on the manufacturer’s site and on a marketplace, the system has to determine whether it’s the same product. Stable identifiers help with this:

GTIN MPN SKU item ID

Correct product identification makes it easier for product systems to recognize the product correctly.

The Feed, Product Page and Product Schema Must Say the Same Thing

We’re dealing with three key layers here, which must be fully consistent with one another:

  • Product page: what the user actually sees and reads.
  • Product Schema: structured data embedded in the page code.
  • Feed: structured information sent to external commerce systems.

Google reads product information both through structured data on the page (Product structured data) and through files in Merchant Center. Elements such as merchant listing schema cover the price, stock level, shipping costs and return policies, among other things. If these three layers give different values, AI systems and search engine algorithms lose trust in the offer, which can result in the product being left out of recommendations.

PIM / ERP / CMS
↓
Product page
↓
Schema / Feed
↓
Search / Shopping / AI / Ads

You’ll find more on implementing these elements technically in our article on schema markup for e-commerce, and we describe the principles for designing the offer view itself for the new standards in detail in our guide to product pages for consumer electronics and home appliances.

Why Does the Feed Matter for AI Search and Agentic Commerce?

Having a feed doesn’t guarantee that ChatGPT will automatically start citing your store. You need to draw a clear line between informational AI Search and AI Shopping. The feed plays a key role where the system works directly with commerce data, meaning when an AI assistant has to show the user a specific product and point to its variant, its current price and a specific seller.

OpenAI uses product feeds to power product discovery, and without complete data, AI has less hard information to classify your product correctly and include it in an up-to-date comparison.

Product Feeds and Agentic Commerce

In the world of Agentic Commerce, optimized product data becomes the foundation of transactional infrastructure. Ultimately, an AI assistant isn’t just supposed to advise on “which TV to choose,” but also to independently:

  • check the price and availability in real time,
  • choose the right variant,
  • find the best seller,
  • proceed to complete the purchase.

That’s exactly why Google is developing the Universal Commerce Protocol and Universal Cart, treating them as frameworks for agentic shopping. At the same time, OpenAI is building its own solution, ACP, as a layer that connects e-commerce systems with ChatGPT users.

You can read about how to prepare your business for this technically in our consumer electronics and home appliance report on Agentic Commerce, and in our guide showing how to adapt your store architecture to Agentic Commerce.

The Four Pillars of Product Data Quality

Pillar Question Example of a problem
Completeness Do we have all the data we need? Missing material, size or GTIN
Accuracy Is the data true? Wrong price or specification in the database
Consistency Does it match the page and the schema? Different price in the feed and on the product page
Freshness Is it up to date? A product marked “in stock” that isn’t actually in the warehouse

How Do You Measure Feed Quality?

KPI What it checks
Product Feed Coverage What percentage of your store’s active catalog is included in the feed you share?
Attribute Completeness How complete are the key fields (e.g. technical attributes, variants) that an AI assistant needs to filter offers?
Price Consistency Is the price shown on your store’s website, in the structured data (Schema) and in the feed itself exactly the same?
Availability Consistency Does the AI system receive a status that matches your actual stock level, or are you promoting an unavailable variant?
Feed Error Rate What percentage of your products drop out at the verification stage because of incorrect identifiers, broken links or gaps in their specifications?

The Most Common Problem: Companies Treat the Feed as an SEM Issue

In many organizations, the feed is handled by a Google Ads specialist, when e-commerce, IT, SEO, product management, pricing and logistics should also be involved.

Why?

Because PPC won’t fix an incorrect GTIN in the ERP, a poor variant architecture, a delayed inventory sync or a wrong price in the database.

Feed quality isn’t an advertising problem. It’s a problem of product data quality across the organization.

The Feed as the Meeting Point of SEM, SEO and AISO

That’s exactly why at Delante we take a comprehensive approach to e-commerce. Today, the product feed connects so many traffic acquisition channels that working on it requires a broader view, not optimization for a single goal. See how its role breaks down across the different areas:

Area Role of the feed
SEM Powers Google Ads, Shopping campaigns and dynamic formats
SEO Indirectly helps keep product data and Merchant Center consistent. The feed isn’t a direct SEO ranking factor
AISO / AI Commerce Gives AI systems the product’s structure, price and availability

The conclusion is simple: feed management is no longer a job for a single department. At Delante, we bring these perspectives together. We help you structure your information architecture so that the same well-optimized product data works for your sales both in traditional search engines and paid campaigns and in new AI-powered shopping experiences.

What About Your Existing Google Merchant Center Feed?

You don’t always have to rebuild your Google Merchant Center feed from scratch. A well-prepared GMC feed can be a starting point, because the basic fields are similar in both environments. However, you need to check:

  • whether the title is unambiguous,
  • whether the description makes sense outside of ads,
  • whether all variants are labeled correctly,
  • whether the GTINs and MPNs are correct,
  • whether availability is up to date,
  • whether the categories are correct,
  • whether the images are assigned to the right variant,
  • whether the feed contains old workarounds made only for Google Ads.

How Do You Prepare Your Product Feed for 2027?

Don’t wait for Agentic Commerce to fully dominate the market. You can start optimizing your product data today. These six steps are all it takes:

  1. Audit your feed. Check completeness, errors, variants and identifiers.
  2. Align the page and the feed. Compare price and availability.
  3. Clean up titles and descriptions. Don’t write them only for campaigns.
  4. Use a single data source. Map your catalog to a PIM, ERP or CMS.
  5. Speed up updates. Shorten the refresh time, especially for prices and stock levels.
  6. Measure data quality. Start tracking the KPIs described above.

Preparing your store for this shopping model means adapting its infrastructure, which is why it’s worth getting to know our AISO services.

Checklist: Check Your Product Feed

Quick test: could a system that has never seen your website tell, based on the feed alone, what the product is, which variant is being sold, how much it costs and whether it can be bought?

Your readiness
0 / 26 checked
Identification
Product data
Commercial data
Media
Data quality
Data freshness
Consistency

What Does the Delante Report Show?

In the Delante report, we examine how ready stores are for Agentic Commerce, looking not only at content but also at the infrastructure that lets systems read and use information about the offer.

Among other things, we check:

  • schema,
  • API,
  • semantic HTML,
  • freshness,
  • whether the price and availability can be found.

The average score of the 15 stores we examined was about 33/100!

The results show that the broader data infrastructure agents need still requires work, even among the big players.

Product Feeds in 2027: A Summary

The product feed is going through an evolution similar to many other elements of e-commerce: from a tool used by a specific channel, it’s becoming part of a shared data infrastructure. Today, a product can be shown in Google Search, Shopping, Google Ads, ChatGPT, AI Shopping and other agentic environments.

That’s why the question “do we have a feed for Google?” is now too narrow. The question now is: do we have well-organized, complete and up-to-date product data that we can safely share with other systems?

See How Ready Your E-commerce Store Is for Agentic Commerce

We'll analyze your online store's product pages, structured data and feed, and show you what's worth putting in order first.

Contact Delante

Sources

  • OpenAI Developers, „Get Started“.developers.openai.com/commerce/guides/get-started
  • OpenAI Developers, „Key concepts“.developers.openai.com/commerce/guides/key-concepts
  • OpenAI Developers, „Best practices“.developers.openai.com/commerce/guides/best-practices
  • OpenAI Developers, „Products“.developers.openai.com/commerce/specs/file-upload/products
  • Google Search Central, „Generate Structured Data with JavaScript“.developers.google.com/search/docs/appearance/structured-data/generate-structured-data-with-javascript
  • Google Search Central, „Product Variant Structured Data (ProductGroup, Product)“.developers.google.com/search/docs/appearance/structured-data/product-variants
  • Google, „New tech and tools for retailers to succeed in an agentic shopping era“.blog.google/products/ads-commerce/agentic-commerce-ai-tools-protocol-retailers-platforms/
  • Google, „How we’re helping retailers thrive with new Universal Commerce Protocol features and AI tools on Google“.blog.google/products-and-platforms/products/shopping/shopping-updates-google-marketing-live/
  • PPC Land, „Google’s Heiko Hotz pitches 60 billion listings as fuel for shopping agents“.ppc.land/googles-heiko-hotz-pitches-60-billion-listings-as-fuel-for-shopping-agents/
Author
Wiktoria Wójciak - SEO Specialist
Author
Wiktoria Wójciak

Senior SEO Specialist

She joined the Delante team in 2021, specializing in local SEO and managing Google My Business Profiles.

A graduate in Information Management from Jagiellonian University, she analyzes and optimizes service and e-commerce websites in her daily work. She feels comfortable working in any industry, as long as client communication is based on mutual understanding and cooperation. She believes that good relationships and shared goals are key to achieving satisfying results.

At Delante, she is responsible for processes such as Google My Business optimization and listing positioning. She believes that SEO success lies in understanding the client’s needs and taking a flexible approach to the specifics of each industry.

Author
Author
Martyna Zemła

Junior SEO Specialist

Author
Wiktoria Wójciak - SEO Specialist
Author
Wiktoria Wójciak

Senior SEO Specialist

She joined the Delante team in 2021, specializing in local SEO and managing Google My Business Profiles.

A graduate in Information Management from Jagiellonian University, she analyzes and optimizes service and e-commerce websites in her daily work. She feels comfortable working in any industry, as long as client communication is based on mutual understanding and cooperation. She believes that good relationships and shared goals are key to achieving satisfying results.

At Delante, she is responsible for processes such as Google My Business optimization and listing positioning. She believes that SEO success lies in understanding the client’s needs and taking a flexible approach to the specifics of each industry.

Author
Author
Martyna Zemła

Junior SEO Specialist

FAQ

Will a Product Feed Replace the Product Page?

No. The feed supplies the data, while the product page presents the product, explains its specifications, provides context, builds trust and is responsible for UX. Schema organizes the information on the page. These three layers complement each other.

Can I Simply Connect My Google Merchant Center Feed to ChatGPT?

No. Direct product feeds in ChatGPT are currently available to approved partners. A well-prepared GMC feed can be a starting point, but it requires reviewing the titles, descriptions, variants and identifiers, as well as proper onboarding.

Is Feed Quality Only an Issue for Large Stores?

No. The more systems use product data, the higher the cost of inconsistency, regardless of scale. A smaller store can start by auditing prices, availability and variants, because these are the easiest places for data to drift apart between the page, the feed and the schema.