Google AI Overviews: The Product Feed Attributes You Need to Add (Most Merchants Are Missing These)

Shopify Discussion
,
1

Hey everyone,

I’m sharing something I have been working on with clients lately that has made a noticeable difference in Google Shopping performance, specifically around AI Overviews and AI Mode in Google Search.

Google has quietly called out specific product feed attributes that directly feed into its AI-driven results. Most merchants I speak to have never heard of them, let alone added them to their feed.

I ran a case study with one client where we improved nothing but the feed data, no bid changes, no creative updates, and saw an 80% improvement in Google Ads performance. CPC dropped, too. That was the moment I realized most merchants are leaving a lot on the table just by having incomplete or poorly structured product data.

The 8 Attributes Google Explicitly Connects to AI Overviews

Here is a quick breakdown of each one:

  1. Product Highlight: Your product’s key selling points. Think Amazon-style bullet points, but inside your Google feed. Focus on benefits, not specs. And if you are using a CSV file to submit these, watch out for commas inside your highlight text; they will break your data. Use TSV or escape the commas with a backslash in Google Sheets.
    Product highlight [product_highlight] - Google Merchant Center Help

  2. Product Detail: This is where your technical specs go: materials, battery life, dimensions, display type, and so on. It has three sub-attributes: Section Name, Attribute Name, and Attribute Value. So, for a watch, it might look like Battery: Life: 10 Years. Structured data like this is exactly what AI systems need to accurately answer specific shopper questions.
    Product detail [product_detail] - Google Merchant Center Help

  3. Variant Option: For products where your variants go beyond color, size, or material. If you sell laptops with different graphics card options, for example, there is no standard Google attribute for that. Variant Option lets you submit something like Graphic Card: GeForce 4070 for one variant and Graphic Card: GeForce 5070 for another. Needs to be used alongside item_group_id.
    Variant option [variant_option] - Google Merchant Center Help

  4. Item Group Title: A shared group-level title for all your variants. Different from the individual variant titles, which carry the specific details. Helps Google understand your product range as a structured group rather than a loose collection of items.
    Item group title [item_group_title] - Google Merchant Center Help

  5. Related Products: Tell Google what accessories, required parts, substitutes, and companion products exist in your catalog. Six relationship types are supported: part of a set, required part, often bought with, substitute, different brand, and accessory. When someone asks AI Mode, “What else do I need for this product?”, this is the data that drives the answer.
    Related product [related_product] - Google Merchant Center Help

  6. Question and Answer: Pre-written FAQ pairs you submit directly in your feed. Up to 30 pairs per product. Base them on the questions your customers actually ask, check your support tickets and reviews. Google uses these to answer detailed shopper questions in conversational AI responses.
    Question and answer [question_and_answer] - Google Merchant Center Help

  7. Document Link: Links to PDF documents about your product, user manuals, assembly guides, and product specs. Google crawls these and uses them to answer detailed questions in AI Mode. This one is seriously underused and a big opportunity right now.
    Document link [document_link] - Google Merchant Center Help

  8. Popularity Rank: A number between 0 and 100 that tells Google how popular this product is relative to the rest of your inventory. Your best sellers get a high number; your slow movers, a lower one. Google uses this to help shoppers make more informed buying decisions in AI-driven results.
    Popularity rank [popularity_rank] - Google Merchant Center Help

The Bigger Point

These eight are the ones Google has explicitly called out. But honestly, every attribute you improve in your feed contributes. Titles, descriptions, GTINs, categories, images, all of it matters. AI systems need complete, well-structured data across the board.

Start with these eight. Then improve everything else.

I created a full breakdown with detailed examples for each attribute on YouTube here: https://www.youtube.com/watch?v=p3Ay8iy8v0A

Happy to answer any questions, feed optimization is what I do every day, so ask away.

2

Hi @EmmanuelFlossie

Thank you for sharing a valuble insights. I think every merchant should check this.

3

Hi @EmmanuelFlossie

This is a genuinely useful breakdown, the Document Link and Question and Answer attributes especially are massively underused and you’re right that most merchants haven’t touched them.

Couple things I’m curious about from your case study. Did you see the lift concentrated in specific query types (more conversational/long tail) or across the board including generic product searches? And when you added Popularity Rank, did you base the numbers purely on your own sales velocity, or factor in margin too?

Either way, thanks for taking the time to write this up properly with the help doc links for each attribute. Posts like this are way more useful than the usual surface level feed advice. Appreciate you sharing it.

Cheers,
Moeed

4

Hello Moeed, that’s a good question. I have not looked into which intent of search terms has increased in that case study. I will write this down to look into this as well. As that is a valuable insight.

Popularity rank recommendations I made, are currently based on Google’s policy details. Not on actual data, because it was only released a few days ago.

And you might be confused why I mention a case study when the new attributes are only a few days old. The case study was done half a year ago, where I improved every single attribute, demonstrating the importance of product improvements for Google Shopping.

Here is the case study: https://www.youtube.com/watch?v=pqFCRer5K4Y

Thank you for your great questions.

5

Hi @EmmanuelFlossie,

Thank you for sharing such a very valuable insights,

I have one quick question regarding the implementation. For merchants using the standard Shopify Google and YouTube app, do these extra attributes sync over automatically if we have them in our product data, or is this something we definitely need to handle through supplemental feeds in Merchant Center?

Thanks a lot,

6

The google and youtube app, currently does not support this, the only one that is currently added, but very very badly is the product detail.

For my clients, I use my own proprietary solutions for advanced data feed optimization. If you want to do this yourself, supplemental feeds, or primary feed apps with custom attributes is the way forward.

7

Solid breakdown, the Document Link attribute is criminally underused, need to look into it. We’ve seen improvements in AI visibility across the other attributes.

One thing worth adding for those reading this Google AI overviews is one layer of the AI search problem, but it’s not the only one. ChatGPT, Perplexity and Grok are now making product recommendations though conversational search.

We’ve found that LLM’s favour semantically rich data. Schema markups, review aggregations, FAQ schema, structured product descriptions - all feed into show well a model can parse and surface a brand. When a user types a query into ChatGPT, the prompt get normalised to match intent, and the brands that get cited are the ones who’s data is the cleanest and most consistently validated across sources. Same underlying principle you’re describing - structured data wins - just expressed differently depending on the platform.

Seeing merchants optimising for one layer and have little to no visibility into the others. Happy to share what we’re seeing if useful.

8

Great breakdown, Product Detail and Related Products are especially under-used. This is what I wanted to focus on

One thing worth flagging though - the 8 attributes you list are all Google Merchant Center fields, so they only help where Google is the retrieval layer.

They don’t help when ChatGPT Shopping, Perplexity Shop, or Claude with web access go look at your storefront as those routes are pulling more traffic every month.

For that path, what matters is on-page structured data (Product + Offer + AggregateRating JSON-LD), GTIN coverage in the markup (not just the feed), and clean Q&A markup on the product page itself. Same underlying data, different surface.

Shopify’s Winter '26 Catalog MCP adds a third surface, agents can query the catalogue without going through a feed at all, and the MCP exposes whatever the product detail schema already has. Worth getting that schema right now if you haven’t.
On your follow-up about query intent for the 80% lift: in data I’ve worked through, the wins concentrate heavily on long-tail and modifier-rich queries (“women’s waterproof hiking boots size 9 with vibram sole” type), because that’s where missing attributes cause Google to drop you from consideration entirely. The head-term queries usually move less. Worth segmenting your search-terms report by query length to confirm.

Curious whether anyone’s seeing Catalog MCP show up in their referral logs yet as would be useful insights to know about .

9

Questions:

  1. What are the most important ranking signals Google AI Overviews uses when selecting ecommerce products and brands to recommend?

  2. Which Google Merchant Center attributes have the highest impact on AI Shopping visibility, and which are most frequently overlooked by merchants?

  3. What product page elements (titles, descriptions, specifications, FAQs, reviews, images, videos, etc.) contribute the most to AI Overview visibility?

  4. Do buying guides, comparison articles, and product-related informational content improve the likelihood of product pages being featured in Google AI Overviews?

  5. Does Google AI favor original, experience-based product descriptions over manufacturer or AI-generated content?

  6. How important are original product photography and lifestyle images in improving AI Shopping visibility?

  7. How do product reviews, merchant reviews, and customer-generated content influence AI Shopping recommendations?

  8. What are the most effective ways for ecommerce brands to demonstrate E-E-A-T and increase trust signals for AI Overviews?

  9. Beyond Product Schema, which structured data types provide the greatest benefit for AI visibility?

  10. How important is internal linking between blogs, collections, categories, and product pages for helping Google AI understand an ecommerce website?

  11. Do Core Web Vitals, page speed, and overall user experience directly influence visibility in Google AI Overviews?

  12. What are Google’s best practices for businesses operating multiple Shopify stores in different countries (e.g., UK and US) to maximize AI visibility while avoiding content duplication and authority dilution?

  13. How can ecommerce businesses measure traffic, impressions, and conversions generated specifically from Google AI Overviews?

  14. What are the most common optimization mistakes that prevent well-optimized Shopify stores from appearing in Google AI Overviews?

  15. Looking ahead 12–24 months, what capabilities or optimizations should ecommerce brands invest in today to remain competitive in us identify actionable opportunities that we

10

I will be answering this in your other thread you asked on the AMA here: Google AI Overview

11

Great questions - I’ll answered the ones I have genuine conviction on and highlighted where the data is still evolving.

1. Most important ranking signals for Google AI Overviews

Structured feed data is the foundation - the attributes already covered in this thread. But above that, topical authority matters. Google needs to trust your brand as a credible source in your category before it surfaces you in AI answers. That means consistent E-E-A-T signals across your site, not just individual product pages.

2. Highest impact GMC attributes most frequently overlooked

Document Link and Question & Answer are the most underused by a distance. Most merchants focus on titles and GTINs and leave everything else incomplete. Product Highlight is also widely ignored despite being one of the clearest signals for conversational queries.

3. Product page elements that matter most

Semantically rich descriptions written for intent, not just keywords. FAQ schema that mirrors how customers actually phrase questions. Review aggregations with specific language - not just star ratings. Videos are underrated; Google is increasingly pulling video content into AI responses for demonstrative queries.

4. Do buying guides and comparison content help?

Yes, significantly. Informational content that earns citations from other sources creates topical authority that bleeds into your product pages. A well-cited buying guide in your niche will lift the products it references.

5. Original vs manufacturer descriptions

Original experience-based descriptions win. Manufacturer copy is duplicated across hundreds of retailers - Google can identify it and it adds zero authority. Your own language, based on real product knowledge, is what differentiates you.

6. Original photography and lifestyle images

More important than most people realise. Google’s image understanding has improved significantly. Unique lifestyle images signal authenticity and original content. Stock or manufacturer images shared across multiple sites weaken your trust signals.

7. Reviews and customer-generated content

Review velocity and specificity matter more than volume. A hundred reviews that use product-specific language (mentioning features, use cases, outcomes) are worth more than a thousand generic “great product” reviews. Merchant reviews through Google’s verified program also directly feed AI shopping recommendations.

8. E-E-A-T for ecommerce

Experience signals are the hardest to fake and the most valuable. Author bylines with demonstrated expertise, real customer outcomes, press mentions, and third-party citations all contribute. For ecommerce specifically, being cited in editorial content in your niche is one of the highest-leverage things you can do.

9. Structured data types beyond Product Schema

FAQ schema, Review schema, BreadcrumbList, and Organisation schema are the highest impact for ecommerce. HowTo schema is worth adding for products with a demonstrable use case.

10. Internal linking

Genuinely important and consistently underestimated. Google needs to understand how your content relates to your products. Blog to collection to product linking creates a semantic map that helps AI understand your inventory in context.

11. Core Web Vitals and page speed

Direct ranking factor for traditional search. For AI Overviews specifically the evidence is less clear, but a slow site signals poor quality regardless of the mechanism. Treat LCP under 2.5s as a baseline hygiene requirement.

12. Multi-country Shopify stores

Hreflang implementation is non-negotiable. Separate domains or subdomains per market outperform subfolders in most cases. Avoid duplicating content across markets - localise properly or you’ll dilute authority in both.

13. Measuring AI Overview traffic

Google Search Console now shows AI Overview impressions as a filter in the performance report. It’s limited but it’s the only first-party data available right now. Third-party tools are catching up but the measurement layer for AI search is still immature.

14. Most common mistakes preventing visibility

Incomplete feed data is number one. Duplicate product descriptions number two. No FAQ schema number three. Beyond that - thin category pages, no editorial citations, and treating AI search as identical to traditional SEO when the signal weighting is different.

15. What to invest in for the next 12-24 months

This is where I’d add something the thread hasn’t covered yet.

Google AI Overviews is one layer. But ChatGPT, Perplexity, Grok and Gemini are making product recommendations through conversational search independently of Google. The signals overlap - structured data, semantic richness, citation density - but the platforms are different and the visibility is separate.

Merchants optimising only for Google AI Overviews are solving half the problem.

The brands that will own their categories in 24 months are the ones building citation authority and structured data signals that work across all of them - not just Google’s ecosystem.

That’s exactly what we track at DaitaFix - visibility across every major AI platform, not just Google. Happy to share what we’re seeing if it’s useful for anyone here.