Google AI Mode Shopping for Holiday 2026: Product Visibility, Merchant Center Feeds and Your Collection Data

Google AI Mode shopping for holiday 2026: what Google announced on 16 September, the new Merchant Center conversational attributes, and how your Shopify product and collection data feed AI Mode.

Google AI Mode shopping is becoming one of the places where holiday 2026 purchase decisions start, and on 16 September 2026 Google told retailers how it wants them to prepare. In a post on its official blog titled 'Retailers: Get ready for AI-driven holiday shopping', Ashish Gupta, Google's VP and GM for Merchant Shopping, announced new Merchant Center insights, new product data fields, an advertising beta and updates to the checkout protocol that powers buying inside AI Mode. This post explains what was announced, what it means for a Shopify store's visibility in AI Mode, and how the product and collection data you already maintain feeds into it.

We have written before about agentic commerce in general and about product feed hygiene for AI shopping. This post is narrower and more current. It covers what Google itself said in September 2026, what its Merchant Center documentation now specifies, and the practical work a store can do in October before the holiday peak. Where a figure comes from Google, we say so, and we do not extend it beyond what Google claimed.

What Google announced on 16 September 2026

Google's announcement, as published on its blog and reported the same day by PYMNTS, contained four main pieces.

• AI performance insights in Merchant Center, now generally available to businesses in Australia, Canada, India, New Zealand and the US. Google says they show how a brand and its products are discovered by comparing share of voice with other brands on surfaces such as AI Mode and AI Overviews.
• New product data options: a video link attribute that Google says can power highly visual, shoppable ad formats, conversational attributes that add richer context to listings, and loyalty data so member perks and pricing can show across Google.
• Updates to the Universal Commerce Protocol, or UCP, which powers checkout inside Google's AI surfaces: new cart transfer and checkout flow testing capabilities, rolling out gradually in the US, with Australia and Canada to follow early next year. Google says UCP already enables direct checkout for hundreds of thousands of brands.
• A Business Agent for YouTube ads, launching as a beta in the US, which lets viewers ask product questions and get guidance without leaving YouTube.

Two figures in the post are worth quoting accurately because they are Google's own. Google says that, on average, merchants who adopt core Merchant Center feed best practices see a 5% increase in conversions the following month. And in a test with lululemon, Google says conversational attributes submitted by the brand were incorporated 50% of the time in relevant product recommendations in AI Mode. The first is an average across merchants, as measured by Google. The second is one brand's test. Neither is a promise about any particular store.

Why this matters for Indian stores

For readers in India, the most immediately useful part of the announcement is that AI performance insights in Merchant Center are generally available to businesses in India. That means an Indian store with a Merchant Center account can, according to Google, see how its share of voice on AI Mode and AI Overviews compares with other brands.

Agentic checkout through UCP is a different matter. Google described the new UCP capabilities as rolling out in the US, with Australia and Canada following early next year. India was not named. So for an Indian store selling to Indian shoppers, the realistic goal this holiday season is visibility, being shown and recommended in AI Mode, rather than having AI Mode complete the purchase on Google's side. Stores that also sell to US customers through Shopify should read the agentic checkout section below.

How product visibility in Google AI Mode works

Google's documentation for site owners on AI features explains the basics. AI Overviews and AI Mode use a technique Google calls query fan-out, issuing multiple related searches across subtopics and data sources to build a response. That is why an AI Mode answer to a shopping question can draw on product listings, merchant websites, reviews and other sources at once.

Google is clear that there is no special trick. Its documentation says there are no additional requirements to appear in AI Overviews or AI Mode, no special optimisations necessary, and no new machine-readable files, AI text files or special schema.org structured data needed. Pages need to be indexed and eligible to appear in Search with a snippet. The same documentation does single out one thing for businesses: keep Merchant Center and Business Profile information up to date.

For a store, that gives two routes into AI Mode shopping answers:

• Your website, through normal Search indexing: product pages, collection pages and content that answer the shopper's question.
• Your product data, through Merchant Center: the structured feed of titles, descriptions, prices, availability, images, identifiers and attributes that Google uses for its shopping surfaces.

Most Shopify stores connect the second route through Shopify's Google & YouTube sales channel. Shopify says the channel automatically syncs the product catalogue to Google Merchant Center, making products eligible across Google's surfaces, including Google Shopping, Search, Images and Lens. The quality of what the channel sends depends entirely on the quality of the product data in Shopify.

Merchant Center conversational attributes, explained

The most practical new item in Google's announcement for a store owner is conversational attributes. Google's Merchant Center Help documentation lists six of them. Google says they are primarily intended for conversational experiences such as AI Mode in Search, that they are optional, and that including them will not affect the approval status of existing products.

• Question and answer [question_and_answer]: frequently asked questions about the product, with both a question and an answer of up to 1,000 characters each. It can be repeated up to 30 times, with a total limit of 10,000 characters per product.
• Document link [document_link]: links to related documents such as manuals or instructions.
• Related product [related_product]: complementary or cross-sell items.
• Item group title [item_group_title]: a title for a product that comes in several variants, describing the product family rather than one variant.
• Variant option [variant_option]: the properties that identify a variant, such as size or colour.
• Popularity rank [popularity_rank]: how well a product sells compared with the rest of your inventory, as a percentile.

Google recommends submitting them through a supplemental data source, though the primary data source and the Merchant API also work. It also asks merchants not to duplicate information already sent in the title, description, product detail or product highlight attributes.

popularity_rank: the attribute that comes from your sales data

One conversational attribute deserves its own section because it draws on the same data that should drive your collection pages. Google's Merchant Center Help defines popularity_rank as a number from 0.0 to 100.0, with at most one decimal place, that communicates a product's sales performance relative to your entire inventory as a percentile. Higher values mean better-performing products. Google's examples are 97.5 for a best seller and 12.5 for a lower performer.

Google's requirements are that the value reflects accurate ranking based on actual sales data and ranks the product correctly against the rest of your inventory. Its best practice is to update the values regularly, through supplemental data sources, when popularity changes significantly based on recent sales. Google says the attribute is primarily intended for conversational experiences such as AI Mode.

Think about what that asks of a store. To send an honest popularity_rank, you need a current, ranked view of which products are selling, refreshed as the season moves. That is the same thing a store needs to order its collection pages well. A store that already ranks its collections by recent orders or revenue has done most of the thinking; a store that still sorts by all-time best sellers will find that its idea of 'popular' is out of date for both purposes.

A note of caution: Google does not say how much weight popularity_rank carries, and we will not guess. It is an optional signal. The value lies in it being accurate, and in it reflecting recent sales rather than a ranking set once and forgotten.

A worked method for calculating popularity_rank

Google specifies the format and the principle but leaves the method to the merchant. Here is a simple, defensible way to calculate it from Shopify data. It is our suggestion, not a Google requirement.

1. 1Choose a window that reflects current demand. For festive season, the last 30 days is a reasonable default; a store with low order volume may need 60 or 90 days to get stable numbers.
2. 2Choose the measure. Net units sold is the closest match to 'popularity'. Net revenue favours expensive products. Pick one and use it consistently.
3. 3Include every active product in your feed, including those with zero sales in the window, so the ranking is against your whole inventory as Google asks.
4. 4Rank products from best to worst on the measure, then convert each position to a percentile from 0.0 to 100.0, rounded to one decimal place. The top product is close to 100.0, the bottom close to 0.0.
5. 5Give products with identical sales the same value, so ties do not create a false order.
6. 6Upload the values through a supplemental data source and refresh them on a schedule, weekly during the festive season and whenever a product's sales change sharply.

Two practical cautions. New products will start near the bottom simply because they have had little time to sell, which is honest but worth knowing. And products that sold well only because of a short promotion will look more popular than they are for the length of your window. Neither is a reason to adjust the numbers by hand, because Google asks for values based on actual sales.

Writing question and answer pairs that help

The question and answer attribute is the one most stores will get the most from, and also the easiest to fill with filler. The best source of good questions is your own customer service inbox, your WhatsApp chats and the questions in your product reviews. Some examples of the kind of pair that adds information a title and description usually do not:

• For jewellery: 'Is this ring resizable?' with an honest answer about whether and how far it can be resized, and 'Does the plating wear off?' with a straight answer about the metal and finish.
• For apparel: 'Does this kurta run true to size?' with what customers actually report, and 'Is the fabric see-through in light colours?' with a plain answer.
• For home products: 'Will this fit a standard Indian gas stove?' with the dimensions that matter, and 'Is it dishwasher safe?'
• For beauty: 'Is this suitable for sensitive skin?' with what the product contains and what it does not claim.

Keep answers factual, avoid prices and dates as Google asks, and do not repeat the description. If you would be uncomfortable reading the answer aloud to a customer in your store, it does not belong in the feed.

How collection data feeds AI Mode

Collections are not a field in a Merchant Center feed, but they influence AI Mode visibility in several indirect ways.

• Collection pages are landing pages. Category searches such as 'silver anklets under 2000' are often answered best by a collection page. A well-built collection page with a clear title, a short useful introduction, working filters and relevant, in-stock products at the top is a strong candidate when Google's systems look for pages that satisfy the query.
• Collections shape product data. Many stores organise product types, tags and metafields around their collection structure. Those same fields often become the product type, category and custom labels sent to Merchant Center.
• Collection order reveals priorities. The products a store puts first in a collection are, in effect, its statement of what it recommends in that category. When that order is based on real performance, it agrees with an honest popularity_rank. When it is based on upload date, it does not.
• Internal links come from collections. Product pages that are only reachable deep in a collection's tail get fewer internal links, which affects how easily they are discovered and how important they appear.

How product data feeds AI Mode

Most of the work is still in the product record itself. These are the fields we see most often missing or weak on Shopify stores, and how they reach Google.

• Titles: Shopify's product title usually becomes the Merchant Center title. Include the product type and the attributes people search for, such as material, colour and size range, in plain language.
• Descriptions: original descriptions that answer real questions, not copied supplier text or a list of keywords.
• Identifiers: GTINs or barcodes where products have them, and accurate brand fields. Shopify's own guidance on Google's AI shopping features lists completing product data and identifiers such as GTINs and MPNs as a preparation step.
• Variants: consistent option names such as Size and Colour, so variants group correctly. Item group title and variant option give Google extra help here.
• Availability and price: accurate, in real time. An AI answer that recommends an out-of-stock product is a bad experience, and feed errors for price or availability can cause disapprovals.
• Images: clear product images on a plain background, plus lifestyle images where they help. The new video link attribute lets you add product video.
• Reviews: genuine customer reviews, which Shopify also lists as a preparation step.

Agentic checkout and UCP: what applies to Shopify stores

Agentic checkout means an AI system starts and completes a purchase for a shopper after the shopper confirms the details. Google began rolling out agentic checkout in AI Mode for US merchants in November 2025, as TechCrunch reported, with select Shopify merchants among the first. Shopify says UCP is an open standard co-developed by Shopify and Google, and that Shopify stores are UCP-enabled by default, so no extra setup is required. Shopify also notes the feature is rolling out in the US and may not be available for all merchants.

The September 2026 announcement added cart transfer and checkout flow testing to UCP, rolling out gradually in the US first. For an Indian store selling domestically, none of this is live yet, based on Google's stated rollout markets. For a store that sells to US customers, the practical step is to make sure the product data, shipping rates, return policy and checkout are clean, because an agent completing a purchase relies on them being accurate.

AI performance insights: how to use them

Because AI performance insights are available in India, they are worth setting up before the festive peak. Google describes them as a way to see how a brand and its products are discovered by comparing share of voice with other brands on AI surfaces. We suggest using them in three ways.

1. 1Record a baseline now, in early October, before the peak and before you change anything in your feed.
2. 2Make one set of changes at a time, such as adding question and answer to your top 50 products, and note the date.
3. 3Review share of voice weekly through the season, alongside your Search Console clicks and your Shopify sales, rather than reacting to a single day's number.

Remember what share of voice measures: how often you appear compared with others. It is not sales. A rise in share of voice that does not turn into sessions and orders tells you something about your product pages or prices, not about AI Mode.

An October checklist for Google AI Mode shopping

1. 1Check the Google & YouTube channel in Shopify is connected and syncing, and clear any product disapprovals in Merchant Center.
2. 2Fix titles on your 50 best-selling products first so they state the product type and key attributes plainly.
3. 3Fill in GTINs or barcodes and brand fields where products have them.
4. 4Set up a supplemental data source in Merchant Center for conversational attributes, so you can add them without changing your main feed.
5. 5Add question and answer pairs to your top products, drawn from real customer questions, without prices or dates.
6. 6Decide how you will calculate popularity_rank from recent sales, and how often you will refresh it. Weekly during the festive season is a reasonable start.
7. 7Make sure your top collection pages open with relevant, in-stock products and have a short, useful introduction.
8. 8Record a baseline in AI performance insights, Search Console and Shopify analytics, then review weekly.

What not to do

• Do not stuff conversational attributes with keywords. Google asks for real product information without duplication.
• Do not invent a popularity_rank. Google requires values based on actual sales data.
• Do not create AI text files or special markup in the hope of appearing in AI Mode. Google says none is needed.
• Do not judge the work on a single week. Google's own 5% figure refers to the month after adopting best practices, on average.
• Do not neglect your website for the feed. AI Mode draws on web pages as well as product data, so your collection and product pages still need to answer questions well.

Where SellRank fits

SellRank, our Shopify collection-sorting app, does not send data to Merchant Center, and we will not pretend it does. Where it helps is upstream. It ranks the products in a collection by real performance, including views, add-to-carts, orders, revenue, conversion rate and profit, and writes that order into Shopify's own manual sort, so the collection pages that act as landing pages for category searches open with what is actually selling. Daily or weekly re-sorts on the Grow plan and up keep that order current as festive demand moves, and conditions can keep out-of-stock products off the first rows.

That same discipline, deciding from recent sales what deserves to be first, is what Google is asking for with popularity_rank. A store that does it for its collection pages will find it far easier to do honestly for its feed.

Get your store ready for AI-driven holiday shopping

If you want help auditing your product data, setting up a supplemental feed or getting your collection pages in shape before the peak, tell us about your store on our contact page. If you want to see your collection pages ordered by what sells right now, SellRank has a free plan for one collection.