Google's own AI Mode data gives D2C brands an unusually direct brief. According to Google's mid-2026 report on how people use AI Mode in the US, shoppers start on traditional Search, click into AI Mode, and ask follow-up questions — and the attributes they ask about most are price, availability, colour, size, and material. The playbook below turns that data into the work: expose those exact facts as clean, liftable, structured statements so AI Mode can cite and recommend your products.
What does Google's data say shoppers ask AI Mode?
Google reports that people often begin a shopping journey on Search and then move into AI Mode to go deeper. The top retail attributes they ask about, in Google's stated order:
| Priority | Attribute | What to expose on the page |
|---|---|---|
| 1 | Price | Current price, currency, any range or variant pricing |
| 2 | Location | In-store/near-me availability, shipping regions |
| 3 | Colour | Named colours per variant, not just swatches |
| 4 | Brand | Consistent brand name and descriptor |
| 5 | Availability | In-stock status, restock, lead time |
| 6 | Size | Size range, fit notes, a real size guide |
| 7 | Material | Fabric/components, care, origin |
| 8 | Style | Occasion, aesthetic, cut |
| 9 | Type | Category and sub-type in plain words |
| 10 | Quality | Durability, warranty, review-backed claims |
The pattern: shoppers ask for concrete, factual attributes, not marketing copy. A page that only shows colour as a swatch image or hides material in a spec graphic gives AI Mode nothing to lift. The fix is the same discipline as writing product pages AI can quote — turn every attribute into a sentence and a structured field.
The step-by-step playbook
- Stay indexable for the opening query. AI Mode grounds in Google's index, so confirm Googlebot and Google's AI crawlers reach your product and category pages — check you aren't blocking them at the CDN. If Search can't retrieve the page, AI Mode can't cite it.
- Expose the top-10 attributes as plain sentences. For each product, state price, availability, colour, size, material, and the rest as self-contained facts in text — "This trail shoe weighs 248g, comes in five colours, and ships free in 2 days" — not locked in images.
- Mirror them in Product/Offer structured data. Use
Product+Offermarkup so machines read price, availability, and attributes without guessing — see preparing your catalogue for AI agents. - Answer the "which one" follow-up. Google reports "which" decision queries grew 40% faster than AI Mode queries overall. Add comparison and use-case context on the page ("best for wide feet", "vs the trail version") so you win the fan-out follow-up, not just the first query.
- Make quality verifiable. Quality is a top-10 attribute, so back it with specifics — warranty terms, durability tests, genuine review counts — because AI engines pull verifiable, corroborated claims over adjectives.
- Keep it fresh. Price and availability are the top two attributes and they change; stale facts get a brand described wrongly or dropped, so keep the live fields accurate.
AI Mode shoppers ask for price, availability, size, and material in plain words — so the brands that win the follow-up are the ones that wrote those facts as sentences, not buried them in a spec image.
Why does this matter more in AI Mode than classic search?
Because AI Mode is conversational, the shopping decision plays out over several turns, each one fanning out again. A classic product listing only has to win the click; an AI Mode product has to answer the follow-up question on the page. Google's data shows those follow-ups cluster on a predictable set of attributes — which is good news, because it tells you exactly what to make extractable.
This is the same lesson the corpus reaches across categories: what makes AI recommend a product is reachability, specificity, corroboration, and freshness, not adjectives. For category-specific nuance, see the fashion and apparel playbook and the electronics playbook, and for how the engines differ, the ChatGPT vs Google vs Perplexity comparison.
How do you know it's working?
Don't guess — measure. Take the high-intent shopping questions your customers ask (start from the attributes above), pose them in AI Mode, and track over time whether your products are named and your pages are cited, by attribute and by category. That cross-engine, multi-prompt read — presence, citations, and sentiment tracked daily rather than spot-checked — is exactly what Buffy Intel is built for.
