Key Takeaways
- Google's Content API for Shopping shuts down on August 18, 2026 — every merchant on it has to move to the new Merchant API before then.
- The Merchant API isn't a version bump; it's a modular, AI-ready rebuild of how Google reads product data.
- Its headline feature is conversational attributes — optional fields that feed AI Mode, Gemini, and Shopping the structured facts they need to recommend products.
- Because those fields are optional and don't affect approval, early adopters gain an AI-visibility edge before they become table stakes.
- Hand-writing them across thousands of SKUs is impractical — which is the gap UCP Radar automates.
What the Google Merchant API Actually Changes
If you run a product feed, you’ve probably had the migration email. Google is retiring the legacy Content API for Shopping and replacing it with the Google Merchant API, and the deadline is real: the old API stops operating on August 18, 2026. Miss it and your integration simply stops working. So the migration isn’t optional, and it isn’t far off.
It’s tempting to treat this as a plumbing job — new endpoint, same feed, ship it and move on. That read is what will cost merchants. The Merchant API is a different shape, not just a different address. Where the Content API was one monolithic interface, the Merchant API is modular, split into sub-APIs for products, accounts, data sources, and reports. Product data itself is restructured: attributes are nested rather than flat, prices arrive in micros instead of a value-plus-currency string, and a single GTIN becomes an array of GTINs. Enums are normalized to clean uppercase values like IN_STOCK and NEW.
Those are the mechanical differences. The reason they exist is the part worth your attention.
The Real Story: It’s Google’s AI-Ready Pipe
Here’s the framing that makes sense of the whole change. The Content API was built to move product data so Google could show it — in ads and Shopping listings. The Merchant API is built so Google’s AI can understand it.
That’s not marketing language. Google introduced a new class of fields called conversational attributes alongside the API, and it’s explicit about where they go: AI Mode in Search, Gemini, the Shopping Graph, and the agentic-commerce surfaces where an assistant shops on a buyer’s behalf. Shoppers increasingly ask full-sentence questions — “waterproof hiking boots for wide feet under $150 that ship in two days” — instead of typing keywords. To answer those, Google’s AI needs structured, machine-readable product facts, not a marketing paragraph it has to guess from. The Merchant API is the pipe that carries those facts.
This is why the “same feed, new endpoint” mindset is a trap. You can migrate perfectly and still ship the exact thin data you shipped before, which passes review and gets you nothing on the AI surfaces where discovery is shifting. The migration is the moment to fix that, not just survive it.
The 8 Conversational Attributes That Feed Google’s AI
The new AI-focused fields are optional, and — importantly — they don’t affect whether your products get approved. They exist purely to give Google’s AI more to work with. There are eight commonly cited ones:
- product_highlight — short benefit fragments, the “why this product” bullets AI answer cards tend to surface directly.
- product_detail — structured technical specs as machine-readable rows (attribute name plus value), for detailed questions.
- question_and_answer — FAQ pairs in the feed itself, intended specifically for conversational experiences like AI Mode.
- variant_option — explicit differentiators so AI understands how your variants actually differ.
- item_group_title — a shared generic title across a variant family.
- related_product — declared relationships like accessories, required parts, or substitutes.
- document_link — HTTPS links to manuals or spec sheets the AI can read from.
- popularity_rank — a relative sales-performance signal for “best” and “most popular” queries.
Notice what these have in common. They convert a merchant’s product knowledge into the clean, factual, question-shaped form that language models match against well. A question_and_answer pair that reads “Is it waterproof? Yes, to 5m” mirrors the exact shape of a shopper’s query. That’s far more useful to an AI than the same fact buried in a description paragraph.
The old API listed your products. The new one lets Google’s AI understand them. The merchants who fill the AI fields get understood; the ones who migrate and stop get listed and skipped.
Why This Is Hard to Do by Hand
If conversational attributes are so valuable, why isn’t everyone filling them in? Because doing it manually doesn’t scale.
Authoring product_highlight, product_detail, and question_and_answer for a few dozen products is a pleasant afternoon. Doing it for hundreds or thousands of SKUs — in the right language, inside Google’s character limits, without drifting into policy-violating claims — is a project most teams never finish. And feeds decay: new products arrive thin, prices change, variants get added. The work is never actually done. This is the same reason supplemental feeds matter more as AI shopping grows — the enrichment layer is where AI visibility is won, and it’s exactly the layer that’s painful to maintain by hand.
So the practical question isn’t whether structured, AI-ready data helps. Google has answered that by building an entire API around it. The question is how you produce and maintain it without a full-time data team.
How UCP Radar Turns the Migration Into an AI Advantage
This is the gap UCP Radar was built for. It already migrated its own Google data pipeline to the new Merchant API ahead of the sunset, so it reads your live product data in the new shape — including Google’s own per-product verdict, the real approval status and the exact reasons a product was disapproved. That’s ground truth, not an estimate, and it’s the honest starting point for knowing what to fix.
From there, UCP Radar’s AI optimizer generates the conversational attributes for you and emits them as a supplemental feed in Google’s RSS format — product_highlight, product_detail, question_and_answer, and variant_option, alongside optimized titles, descriptions, and standard attributes. That supplemental feed merges onto your primary feed by product ID inside Merchant Center, so your live Google data gets AI-enriched without you hand-editing a single field. You connect once; it keeps the enrichment fresh as your catalog changes.
Two things make this more than a text generator. First, guardrails: brand names, identifiers, price, and availability are never altered, health and material claims are kept compliant, and demographic fields are only set when your data actually supports them. Second, measurement: UCP Radar scores each product’s AI-readiness on a 0-to-100 scale and records a before-and-after for every optimization run, so the uplift is something you can see rather than assume. It also works across the surfaces AI shopping is spreading to — the same enrichment that serves Google’s AI Mode also feeds Perplexity, with a separate complete feed for ChatGPT Shopping — in Turkish, English, German, French, and Arabic.
To be clear about the boundary: UCP Radar prepares and hosts the AI-ready feed; you register it in Merchant Center. It doesn’t submit on your behalf, and conversational attributes don’t buy you a ranking — they buy you the structured, machine-readable data the AI needs to consider you at all.
Conclusion
The Google Merchant API migration is going to happen whether you engage with it or not, because on August 18, 2026 the old API stops answering. The choice you actually have is what you do with the move. Treat it as plumbing and you’ll carry the same thin feed into a system Google explicitly rebuilt for AI understanding. Treat it as the upgrade it is, and you fill the conversational attributes that decide whether AI Mode, Gemini, and Shopping can understand and recommend your products — while they’re still optional and most of your competitors are ignoring them. UCP Radar reads your feed through the new Merchant API, generates those AI-ready fields automatically, and scores the improvement, so the migration you have to do anyway becomes the AI advantage you’d otherwise have to build by hand.
Frequently Asked Questions
Google has said the legacy Content API for Shopping stops operating on August 18, 2026. After that date, integrations still on the old API stop working, so every merchant managing a feed through it needs to move to the new Merchant API before then.
The Merchant API is Google's modular replacement for the Content API for Shopping. Instead of one monolithic API, it splits into sub-APIs (Products, Accounts, Data Sources, Reports, and more) and reshapes product data — nested attributes, prices in micros, an array of GTINs — into a form built for Google's AI systems to understand, not just list.
Conversational attributes are optional Merchant Center product fields introduced for Google's AI surfaces — product_highlight, product_detail, question_and_answer, variant_option, item_group_title, related_product, document_link, and popularity_rank. They don't affect product approval, but they feed AI Mode, Gemini, and Shopping the structured facts those surfaces use to understand and recommend products.
No. Google is explicit that conversational attributes don't affect whether a product is approved. They're optional. That's exactly why they're an early-mover opportunity — they influence AI visibility without being a compliance requirement yet.
UCP Radar reads your live Google data through the new Merchant API — including Google's own per-product approval status — and generates the AI-ready conversational attributes as a supplemental feed that merges onto your primary feed by product ID inside Merchant Center. You get the enriched, AI-focused fields without hand-writing them for every SKU.
