Catalog Intelligence · Google UCP · Conversational Commerce
Google Changed How Products Get Found.
Is Your Catalog Ready?
Google now uses AI to answer shopping questions — not just show links. Whether your products appear in those answers starts with one thing: the quality of your product data.
01 — What Changed
Google Now Answers Shopping Questions. It Doesn’t Just Return Links.
Earlier this year, Google announced a significant change to how shopping works. Instead of showing a list of links and letting you figure it out, Google’s AI now reads the shopper’s question, builds an answer, and recommends specific products — all without the shopper leaving the search page.
The announcement, made at the NRF retail conference in January, had two distinct parts. The first was the Universal Commerce Protocol — an open standard for the full end-to-end shopping journey: discovery, checkout, payment, post-purchase. It was co-developed with Shopify, Etsy, Wayfair, Target, and Walmart, and endorsed by more than twenty others including Macy’s, Zalando, Visa, Mastercard, and Stripe. Google has begun rolling out these capabilities across AI Mode, Gemini, and Business Agent, with broader expansion planned.
The second part is what matters most for fashion retailers right now: Google simultaneously added dozens of new product data fields to Merchant Center — the system retailers use to submit their product catalog to Google. These fields are built specifically for AI-driven search. They go beyond the basics like price and size. They ask: what occasion is this product for? What does it pair with? What questions do shoppers typically ask about it? This is where the catalog enrichment opportunity actually lives.
“If your product data is incomplete or messy, your products won’t show up in these AI answers — regardless of how good the products actually are.”
Product data is now the primary lever you control. Ads and SEO still matter — but they cannot compensate for a catalog that Google’s AI cannot read clearly enough to recommend with confidence.
02 — How Shopping Search Has Changed
Shoppers Used to Type Keywords. Now They Ask Questions.
Think about how people search for clothes today compared to five years ago. Someone used to type “blue midi dress.” Now they type — or say out loud — something like: “I need something elegant but not too formal for my cousin’s outdoor wedding in November. I’ll be on my feet all day.”
That is not a keyword. That is a question with several layers built in: the occasion (wedding), the dress code (elegant but not too formal), the setting (outdoor), the season (November), and a practical need (comfortable enough to stand in all day). Google’s AI is now trying to answer all of that at once — and it pulls its answer from product data fields, not from a product description paragraph.
The Key Difference
Old search ranked pages. New search constructs answers from structured product fields. Structured fields are far more reliable for AI matching than long product descriptions — because the AI can read a field value directly, rather than trying to interpret a paragraph. If “outdoor wedding” doesn’t exist as a dedicated field, your dress is much less likely to surface in that answer, even if the description mentions it.
This is why well-written product descriptions are no longer enough on their own. Google’s AI needs clean, structured data fields — not paragraphs of copy — to confidently say: “This dress matches what you’re looking for.” If the right fields don’t exist, the product is invisible to that search.
03 — What Fashion Catalogs Are Missing
The Product Data That Most Fashion Brands Don’t Have
Most fashion catalogs are not short on data. The average product in a well-managed catalog has 25 to 40 data fields — material, color, size, weight, care instructions. On paper, that sounds thorough.
But almost none of those fields are what Google’s AI is now looking for. Here is where the gaps actually are.
Occasion. When a shopper types “wedding guest dress” or “what to wear to a work presentation,” they need a product tagged specifically for that moment. Most catalogs use broad labels like “casual” or “formal.” Those labels are too vague to be useful. The difference between tagging a dress as “formal” versus “outdoor garden wedding” is often the difference between showing up in the search result or not.
Style and mood. Searches like “quiet luxury,” “old money,” “resort style,” and “effortless chic” are now some of the most common ways shoppers describe what they want — especially in AI or voice search. These are real searches with real buying intent. But they almost never exist as structured fields in a fashion catalog. At best, they are buried in a product description where AI cannot easily use them.
What it pairs with. Google’s new data standard includes a field for compatible or complementary items — what top goes with this skirt, what shoes suit this dress. Shoppers think in outfits, not individual products. Most catalogs treat every item in complete isolation. That mismatch means missed opportunities whenever a shopper searches for a complete look.
Color names shoppers actually use. Many fashion brands use internal color codes — “Victoria Blue,” “Shadow 4,” “Amazon.” Those names mean nothing to a shopper searching for “navy” or “forest green.” Google’s AI may not reliably map “Shadow 4” to “charcoal” or “dark grey,” especially across a large catalog. Products become much harder to find for any color search that uses plain, everyday language.
04 — A Concrete Example
Same Dress. Two Catalogs. Completely Different Results.
Here is what this looks like in practice. A floral midi dress — same product, two different ways of describing it. One is how most catalogs look today. The other is what Google’s AI actually needs.
Typical Catalog Today
What Google's AI Needs
- Color: Floral Multi
- Length: Midi
- Fabric: 100% Viscose
- Fit: Regular
- Occasion: Casual, Smart Casual
- Description: “A beautiful floral midi dress with a relaxed fit, perfect for any occasion.”
- Color: Coral, Blush Pink, Sage Green
- Length: Midi (falls below the knee)
- Fabric: Lightweight viscose — breathable, good for warm weather
- Occasion: Garden party, outdoor wedding guest, summer lunch
- Style: Romantic, feminine, vintage-inspired
- Pairs with: Strappy sandals, straw bag, block-heel mule
The first version will not show up when someone searches “outfit for an outdoor summer wedding.” The second version will. Same dress. Completely different outcome — because of how the data is structured, not because of the product itself.
05 — Why Generic Tools Fall Short
Off-the-Shelf AI Tools Won’t Solve This for Fashion
The obvious response is: “We’ll run our catalog through an AI tool.” There are many available, and more are launching every month following Google’s announcement. Most of them will not work well for fashion — and understanding why matters before spending money on one.
Generic AI enrichment tools are built for broad product categories: electronics, furniture, household goods. When you feed them a fashion product, they try to fit it into those same categories. The results are often wrong in ways that are easy to miss.
A Real Example
Take a Western-style fringe suede jacket — a staple in country, rodeo, and festival fashion with its own specific vocabulary around silhouette, styling context, and occasion. A generic AI tool has no real category for this. It sees the image and returns something like “casual outerwear” or “women’s jacket.”
That output is nearly useless. The attributes that actually matter — the styling occasion (country concert, Coachella, Nashville bachelorette), the aesthetic (Western chic, bohemian), the items it pairs with — are simply not in the tool’s vocabulary. The same problem shows up across any fashion category where language moves fast or carries cultural specificity: streetwear, preppy, coastal grandmother, quiet luxury. Generic tools cannot keep up.
There is a second problem: generic tools over-tag. They apply broad, vague labels to everything — “versatile,” “can be dressed up or down,” “suits multiple occasions.” At catalog scale, this destroys the precision Google’s AI needs. When everything is tagged as versatile, nothing is easy to find. Vague tagging creates noise, not results.
06 — What to Do About It
Five Practical Steps to Get Your Catalog AI-Ready
You do not need to rebuild your entire product system. Google’s new data fields are supplementary — they sit alongside what you already have. The work is about adding the right fields in the right places, not starting over.
01
Find out what you are actually missing
Start with a simple audit. Check which of these fields exist in your catalog today: occasion, style mood, what the item pairs with, and plain-language color names. Most teams find at least two of these are completely absent — not just sparse, but not there at all.
02
Fix your color names first
This is the fastest, most impactful change most retailers can make right now. Map your internal codes to plain language: “Victoria Blue” becomes “navy,” “Shadow 4” becomes “charcoal.” It takes one enrichment pass and immediately improves performance across all search — not just AI search.
03
Start with your newest collection, not the whole catalog
New collections get the most traffic and marketing attention — but they are also the least enriched, because there is never enough time at launch. Fix these first. That is where the revenue impact is most immediate and most visible.
04
Use product images, not just written descriptions
A lot of the missing information — the occasion a product suits, the styling context, the aesthetic feel — is visible in the product photography. AI tools that analyse images often extract occasion and mood more accurately than tools that read copy, especially when the copy was written for SEO rather than accuracy.
05
Submit the new fields to Google Merchant Center
Once the enriched data exists, adding it to your Google product feed is straightforward. The new fields go alongside your existing product data — nothing needs to be replaced. Getting the data right is the hard part. The submission is just a feed update.
07 — The Bigger Picture
Search Has Changed. Product Data Is Now the Strategy.
For years, fashion retailers competed on ads, on SEO, on website experience. Those things still matter. But they all assume a shopper who arrives at your site. AI-powered search changes where the competition begins — it starts before the click, inside the answer Google builds for the shopper’s question.
Whether your product is in that answer depends more on your product data than on almost anything else you can control. Not your ad spend. Not your page speed.
The retailers who act on this now — before AI-powered shopping becomes the dominant way people discover products — will build an advantage that compounds over time. Every product enriched today is better positioned not just on Google’s current AI Mode, but on every AI-driven shopping surface that follows.
Visibility is not guaranteed. It is earned — one data field at a time.
Topics
conversational attributes
Google AI Mode product data
Merchant Center fashion 2026
AI product discovery fashion
fashion catalog enrichment
Google agentic commerce
fashion search 2026
Free Report
The Fashion Ecommerce Discovery Gap Report 2026
What 5M+ SKUs reveal about the structural gap between how retailers describe products and how shoppers search for them – and what to do about it.
Next step
Find Out Where Your
Catalog Stands Right Now
We’ll run a free check on a sample of your products — showing you exactly which fields are missing, which are too vague, and what that means for your visibility on Google’s AI search surfaces. No commitment required.
Author
Vinay Adapa
CTO & Co-Founder, Perspiq.ai
AI-powered catalog enrichment with expert oversight—delivering shopper-ready data that feeds search and SEO.
Product
Trust
Company
© 2026 Perspiq.ai. All rights reserved.