The CPG GEO & AEO Guide: Winning AI Search and Agentic Shopping (2026 Update)

How Consumer Brands Win Visibility, Recommendations, and Sales in AI Search and Agentic Shopping

Originally published September 2025. Updated October 15, 2026 with a year of changes in AI shopping: Google’s Universal Commerce Protocol and Universal Cart, ChatGPT’s shift from in-chat checkout to discovery and product-feed ads, Amazon Rufus at scale, new click-through and conversion data, and a revised playbook for CPG teams.

Reading time: ~15 minutes · Audience: U.S. consumer packaged goods (CPG) brands, retail and e-commerce teams, and their agencies

Why this matters: AI assistants are now a real shopping channel. Adobe measured AI-referred traffic to U.S. retail sites up roughly 693% in the 2025 holiday season, and Salesforce attributes about 20% of global holiday online sales to AI and agents. If your product data and content aren’t ready for answer engines and shopping agents, you lose the moment when the assistant decides which brand to name.

The TL;DR

GEO and AEO are about being named in the answer. Generative Engine Optimization (GEO) and Answer Engine Optimization (AEO) mean earning citations and recommendations inside AI answers and shopping agents, not just ranking blue links. For CPG, the “shelf” is increasingly the short list an assistant returns.
Discovery happens in AI; checkout mostly still happens with the retailer. OpenAI retired ChatGPT’s Instant Checkout in March 2026 and refocused on product discovery, while Google built agentic checkout into AI Mode and Gemini through its Universal Commerce Protocol. The durable lesson: win the recommendation, then make the buy path easy wherever it happens.
Product feeds are now the front door to AI shopping. ChatGPT shopping, Google AI Mode and Gemini all draw on structured product feeds. Google added “conversational attributes” for AI surfaces, and OpenAI now uses the same product feed for both organic shopping results and ads.
Retailer agents matter as much as chatbots. Amazon says more than 300 million customers used Rufus in 2025, which it credits with nearly $12 billion in annualized incremental sales. Walmart’s Sparky now runs inside ChatGPT and Gemini. Your retail listings, reviews and Q&A are AI content.
Clicks are scarcer, but AI visitors are worth more. Seer found organic click-through rates fall about 61% when Google shows an AI Overview, but brands cited in the Overview get 35% more organic clicks than those that aren’t. Adobe found AI referrals converted 31% better than other traffic last holiday season.
Classic search still feeds AI. Our own analysis found non-AI-Overview keywords still carry far more traffic potential than AI Overview keywords. Evergreen search content is still the raw material assistants retrieve and cite.

What Changed Since Our First Edition

When we published this guide in September 2025, agentic shopping was mostly announcements. A year later, it has been launched, tested, partly rolled back, and rebuilt. Here is what CPG teams need to know:

September 2025: OpenAI launched Instant Checkout and the Agentic Commerce Protocol (ACP) with Stripe, letting shoppers buy inside ChatGPT.
November 2025: Perplexity made its agentic shopping free for U.S. users and added PayPal checkout across more than 5,000 merchants. Amazon sued Perplexity over its Comet shopping agent.
December 2025: Instacart launched a ChatGPT app that turns a recipe into a grocery cart and checks out in the conversation, the first grocery example of full agentic checkout.
January 2026: Google announced the Universal Commerce Protocol (UCP), co-developed with Shopify, Etsy, Wayfair, Target and Walmart, to power checkout in AI Mode and Gemini. Google also added dozens of new Merchant Center attributes for conversational shopping and launched Business Agent, a brand-voiced shopping assistant on Search.
February 2026: Amazon reported Rufus results at scale. Ahrefs and Seer published updated studies showing sharply lower click-through rates when AI Overviews appear.
March 2026: OpenAI retired Instant Checkout, refocusing ChatGPT shopping on discovery with merchants handling checkout. Walmart reportedly moved from ChatGPT’s checkout to running its own Sparky agent inside ChatGPT and Gemini. A federal judge temporarily barred Perplexity’s agent from shopping on Amazon.
May 2026: OpenAI began turning product feeds into ads in ChatGPT. At Google I/O, Google announced Universal Cart, a cross-merchant cart for Search and Gemini, and said AI Mode had passed 1 billion monthly users.
August 2026: A federal appeals court lifted the Amazon injunction, finding that Perplexity’s users, not Perplexity itself, accessed Amazon through the agent. The rules for shopping agents are still being written.

What it means for CPG: The protocols and checkout buttons will keep changing. What has held up across every platform is the same set of inputs: complete, consistent product data; content that answers shopper questions clearly; credible third-party proof; and availability information that tells the assistant where to buy.

What GEO and AEO Mean for CPG

Traditional SEO was built around ranking pages for keywords. GEO and AEO shift the goal to appearing as a cited source or recommended brand in Google AI Overviews and AI Mode, Gemini, ChatGPT, Claude, Perplexity, and retailer assistants like Amazon Rufus and Walmart Sparky. iPullRank’s AI Search Manual describes the same shift: the blue-link era is giving way to answer engines, and relevance engineering is how brands get selected.

For CPG, this has concrete implications:

Assistants compare products on attributes. They need structured, comparable facts: size, count, flavor, ingredients, allergens, dietary labels, certifications, price, and retailer availability.
Assistants lean on sources you don’t control. BCG notes that LLMs evaluate brands using reviews and discussions across the web, including negative ones. If a product doesn’t deliver on its claims, those sources can undercut your marketing.
Assistants answer for specific shoppers. The same model may recommend different brands to a “health-conscious” shopper than to a general one. Visibility has to be earned persona by persona and use case by use case.
Answers can draw on live data and training data. Search-connected assistants pull your pages, feeds and retailer listings in real time. Model knowledge reflects brand mentions in the broader web when the model was trained. You need both.

The Numbers: How AI Is Changing CPG Discovery

Signal	Latest data	Source
Shoppers using AI in their purchase journey	Nearly 1 in 3, about triple the level of 18 months earlier; 13% ultimately buy products AI recommends	BCG, survey of 13,000 consumers in 12 markets (Sept. 2026)
AI-referred traffic to U.S. retail sites	Up about 693% year over year in the 2025 holiday season	Adobe Analytics (Jan. 2026)
Conversion of AI-referred visitors	31% higher than other traffic sources; 45% more time on site	Adobe Analytics (Jan. 2026)
Holiday sales influenced by AI and agents	About 20% of global online holiday sales, or $262 billion	Salesforce (2025 holiday season)
Amazon Rufus	300+ million users in 2025; nearly $12 billion in annualized incremental sales; users convert 60%+ higher	Amazon Q4 2025 results (Feb. 2026)
Google AI Mode	More than 1 billion monthly users	Google I/O (May 2026)
Organic CTR when AI Overviews appear	About 61% lower; cited brands get 35% more organic clicks than uncited ones	Seer Interactive (Feb. 2026)
Top-ranking page CTR with AI Overviews	About 58% lower	Ahrefs, 300,000 keywords (Feb. 2026)

These numbers come from different methodologies and shouldn’t be added together, but they point the same way: fewer clicks overall, more influence concentrated in AI answers, and higher-value visits when the AI does send someone.

Where CPG Shoppers Meet AI, and How Your Products Get There

Surface	How products get in	What CPG teams should do
Google AI Overviews, AI Mode and Gemini	Google Merchant Center feeds, the open web, structured data; checkout through UCP for eligible retailers	Keep Merchant Center complete and current, add the new conversational attributes as they open up, and make PDPs answer common questions
ChatGPT shopping	Merchant product feeds (direct feeds for approved partners, or Shopify’s agentic channel), web search results, and retailer apps such as Instacart and Walmart	Get your products into feeds directly or through retail partners; keep price and stock current; consider feed-based ads where they fit
Amazon Rufus	Your Amazon listings, A+ content, reviews, and customer Q&A	Treat listings and Q&A as answer content; fill attribute gaps; respond to recurring questions
Walmart Sparky (in the Walmart app, ChatGPT and Gemini)	Walmart item content, reviews and availability	Audit Walmart item content for completeness and consistency with your own site
Instacart in ChatGPT	Instacart retailer catalogs	Make sure product names, sizes and images are accurate across Instacart retailers; recipe-friendly content helps
Perplexity	Web search, merchant data, PayPal-connected merchants	Earn citations from reviewers and publications; keep PDPs crawlable and specific
Claude and other assistants	Web search and training data	Build a consistent, well-cited brand footprint across trusted sources

Why You Still Need Classic Search Demand to Fuel GEO

Our research found that non-AI-Overview keywords still carry far more traffic potential than keywords that trigger AI Overviews. The 2026 click-through data makes AI Overview queries even less valuable as traffic sources on their own. That’s not a reason to ignore AI surfaces; it’s a reason to keep building the evergreen pages, guides and product content that assistants retrieve and cite.

Google’s own guidance for AI features in Search says the same thing in plainer terms: the fundamentals of helpful content, crawlable pages, and structured data that matches what’s on the page still apply.

The CPG GEO and AEO Playbook (2026 Edition)

1) Build a reusable “AI-Ready Product Profile” (ARPP)

Create a single source of truth for every SKU as JSON (for APIs and your PIM) plus CSV or Parquet (for feeds). Include every attribute an assistant might use to compare products, and add the answer-oriented fields that AI shopping surfaces increasingly ask for: common questions, substitutes, compatible items, and use cases.

Starter JSON (adapt to your PIM or headless CMS):

{
  "sku": "ABC-123",
  "gtin": "00012345678905",
  "brand": "YourBrand",
  "product_name": "YourBrand Sparkling Water, Lime, 12 fl oz, 12-pack",
  "category": ["Beverages", "Sparkling Water"],
  "short_desc": "Lime sparkling water made with carbonated water and natural flavors.",
  "long_desc": "Zero sugar, zero calories. 12 cans x 12 fl oz. Non-GMO Project Verified.",
  "bullets": ["Naturally flavored lime", "0 sugar, 0 calories", "Non-GMO Project Verified", "12 x 12 fl oz"],
  "ingredients": ["Carbonated Water", "Natural Flavors"],
  "allergens": [],
  "dietary": ["Vegan", "Gluten-Free", "Kosher"],
  "certifications": ["Non-GMO Project Verified"],
  "pack_size": {"unit_count": 12, "unit_volume_fl_oz": 12},
  "use_cases": ["Mocktails", "Lunchbox", "Post-workout"],
  "common_questions": [
    {"q": "Does it contain sweeteners?", "a": "No. It has no sugar and no artificial or natural sweeteners."},
    {"q": "Is the can lining BPA-free?", "a": "Yes."}
  ],
  "substitutes": ["YourBrand Grapefruit Sparkling Water 12-pack"],
  "pairs_with": ["YourBrand Variety Pack"],
  "images": [
    {"url": "https://cdn.yourbrand.com/sku/abc-123/main.jpg", "alt": "YourBrand Lime sparkling water 12-pack"},
    {"url": "https://cdn.yourbrand.com/sku/abc-123/nutrition.jpg", "alt": "Nutrition facts panel"}
  ],
  "msrp_usd": 8.99,
  "availability": [
    {"retailer": "Walmart", "in_stock": true, "url": "https://www.walmart.com/ip/..."},
    {"retailer": "Target", "in_stock": true, "url": "https://www.target.com/p/..."},
    {"retailer": "Instacart", "in_stock": true}
  ],
  "brand_urls": {
    "pdp": "https://www.yourbrand.com/products/lime-sparkling-water-12x12oz",
    "faq": "https://www.yourbrand.com/help/sparkling-water-faq"
  },
  "country_of_origin": "USA",
  "updated_at": "2026-10-15T00:00:00Z"
}

Why this matters: Every AI shopping surface rewards complete, consistent, machine-readable facts. The same profile can feed your PDPs, Google Merchant Center, ChatGPT’s product feed, and retailer content, so the facts match everywhere an assistant looks. Use one GTIN per item, consistently, across every channel.

2) Publish AI-friendly PDPs with JSON-LD

Add Product markup to every product page so search engines and assistants can parse brand, size, price, ratings and availability. Keep the markup in sync with what’s visible on the page.

<script type="application/ld+json">
{
  "@context": "https://schema.org/",
  "@type": "Product",
  "name": "YourBrand Sparkling Water, Lime, 12 fl oz, 12-pack",
  "image": ["https://cdn.yourbrand.com/sku/abc-123/main.jpg"],
  "description": "Zero sugar lime sparkling water. 12 x 12 fl oz cans.",
  "sku": "ABC-123",
  "gtin14": "00012345678905",
  "brand": {"@type": "Brand", "name": "YourBrand"},
  "offers": {
    "@type": "Offer",
    "priceCurrency": "USD",
    "price": "8.99",
    "availability": "https://schema.org/InStock",
    "url": "https://www.yourbrand.com/products/lime-sparkling-water-12x12oz"
  },
  "aggregateRating": {"@type": "AggregateRating", "ratingValue": "4.6", "reviewCount": "1284"},
  "additionalProperty": [
    {"@type": "PropertyValue", "name": "Dietary", "value": "Vegan, Gluten-Free, Kosher"},
    {"@type": "PropertyValue", "name": "Allergens", "value": "None"}
  ]
}
</script>

Only include ratings you can substantiate on the page, and keep price and availability current.

3) Feed every AI shopping surface from one template

Use one canonical template, then map it to each channel. Keep field names stable and transform per destination.

CSV headers (starter):

sku,gtin,brand,product_name,short_desc,long_desc,ingredients,allergens,dietary,certifications,unit_count,unit_volume_fl_oz,use_cases,common_questions,image_url,msrp_usd,retailer,retailer_price,availability,product_url,updated_at

Channel mapping:

Google Merchant Center: Maintain GTIN, brand, google_product_category, price and availability. Add Google’s new conversational attributes (product Q&A, substitutes, compatible items) as they become available to you. This feed now powers AI Mode, Gemini, Universal Cart and Business Agent.
ChatGPT shopping: If you sell direct, apply for OpenAI’s merchant product feed (or use Shopify’s agentic channel). The core required fields are item ID, title, description, URL, brand, seller name, image URL, availability and price. OpenAI expects at least a daily full snapshot, with intraday price and stock updates. If you sell mainly through retailers, focus on accurate content at those retailers, since their apps and feeds are how your products reach ChatGPT.
Amazon: Split bullets cleanly, fill every relevant attribute, keep A+ content factual, and monitor customer Q&A; Rufus draws on all of it.
Walmart: Validate GTIN and category, complete item attributes, and keep content consistent with your PDP; it now reaches shoppers through Sparky in Walmart’s app, ChatGPT and Gemini.
Instacart, Target and others: Align pack sizes, units, images and dietary badges across retailers.

Push price and availability daily or faster, and content changes weekly.

4) Treat retailer listings, reviews and Q&A as answer content

Retailer assistants answer from what’s on the retailer’s page. For CPG brands that sell mostly through retail, this is often the most important AI surface you have.

Audit your top SKUs on Amazon, Walmart, Target and Instacart for missing or inconsistent attributes.
Answer recurring shopper questions in customer Q&A and fold the answers into your listing copy.
Watch review themes. If shoppers repeatedly say a product falls short of a claim, assistants will pick that up; fix the product or the claim.

5) Earn the third-party footprint assistants trust

Assistants recommend brands that credible sources talk about. Prioritize:

Reviews and “best of” coverage from publications in your category
Retailer editorial features and badges
Registered dietitian, expert or creator content that cites specific product facts
Accurate brand information in public reference sources
Consistent facts across your site, retailer pages, and press materials

This is also the long game for model knowledge: brands that are widely and consistently described are better represented when models are retrained.

6) Write content blocks assistants can quote, for each shopper persona

For each SKU and category page, maintain short, factual blocks:

Who it’s for: “Sparkling water fans who want zero sugar and no sweeteners”
What’s different: ingredients, sourcing, flavor profile, packaging
How to use it: recipes, pairings, occasions
Proof: certifications, verified ratings, awards
Persona answers: short sections for the shopper types that matter in your category, such as health-conscious, budget, family, or allergy-sensitive shoppers

Visibility can swing dramatically by persona. One 2026 tracking study of large CPG brands in ChatGPT found the overall leader dropped from first to seventh when the prompts came from a health-conscious shopper. Write for the questions each persona actually asks.

7) Measure AI visibility like a share-of-shelf metric

Track how often your brand is named, how it’s described, and whether it’s recommended over competitors, across a fixed set of prompts and personas, at least weekly. Results are volatile day to day, so look at trends.
Segment AI-referred sessions in your analytics (ChatGPT, Perplexity, Gemini, Copilot and others) and compare conversion to classic organic.
Track your presence in Google AI Overviews for your key category queries, and whether your pages are the ones cited.
Feed the gaps you find back into your product profiles and content.

8) Decide where paid fits

AI shopping surfaces are adding ads quickly. OpenAI now builds ChatGPT ads directly from product feeds, and Google is piloting Direct Offers, which shows exclusive discounts to shoppers in AI Mode who are ready to buy. Retailer assistants sit inside retail media networks. Paid placements can complement organic visibility, but they don’t replace it: the same product data powers both.

What not to over-invest in

llms.txt files: Adoption has grown, but a 2026 study of Ahrefs server logs found 97% of llms.txt files received zero requests in a month. It’s cheap to add, but it isn’t a substitute for good pages and feeds.
Repetition without substance: Repeating real product facts across relevant contexts helps; padding pages with keyword-stuffed copy does not.
Betting on one checkout button: In-chat checkout options have launched and been retired within months. Optimize for being recommended, and make sure every path to purchase works.

Tutorial: A 90-Day GEO and AEO Plan for a CPG Catalog

Days 1–30: Centralize product truth

Build the ARPP for your top 50 to 100 SKUs; store it in your PIM and version every change.
Fix GTIN, size and naming inconsistencies across your site and top retailers.
Run a baseline AI visibility audit: 30 to 50 category prompts, across two or three shopper personas, in ChatGPT, Google AI Mode, Gemini, Perplexity and Rufus.

Days 31–60: Publish and syndicate

Render Product JSON-LD on every PDP (server-side) with price, availability, ratings and dietary properties.
Add Q&A, comparison, ingredient and usage blocks to PDPs and category pages.
Clean up Google Merchant Center and enroll in new conversational attributes where available.
Apply for the ChatGPT merchant feed if you sell direct, or prioritize retailer content if you don’t.
Audit and update Amazon, Walmart, Target and Instacart listings for your top SKUs.

Days 61–90: Expand and measure

Pitch category publications and reviewers with specific, verifiable product facts.
Publish persona-specific explainer content (for example, “How to choose a low-sugar electrolyte drink”).
Set up AI referral segments in analytics and a weekly AI visibility report.
Re-run the baseline audit and compare. Feed every gap back into the ARPP.

GEO Content Patterns for CPG PDPs and Category Hubs

Use these templates so assistants have consistent blocks to cite:

A) “Why this product” (75–120 words): Who it’s for, the core benefit, the key differentiator, pack size, and a concrete use case. Avoid fluff; state facts the model can reuse.

B) “Compare alternatives” table: Brand, pack size, per-unit price, key claims, diet suitability, and where to buy.

C) “Ingredient transparency” (bulleted list): Sourcing, flavoring, sweeteners, and third-party certifications.

D) “Usage and storage” (bulleted list): Instructions, shelf life, storage temperature, and recycling guidance.

E) “Proof points” (short list): Awards, verified ratings, retailer badges, and test results.

F) “Substitutes and pairings”: What to buy if this item is out of stock, and what goes with it. AI shopping surfaces increasingly ask for exactly this.

G) “Questions shoppers ask”: Real questions from customer service, reviews and retailer Q&A, answered in one or two sentences each.

FAQ

Q1: How do we show up when shoppers ask ChatGPT, Gemini or Claude for product ideas?

Get your products into the feeds those assistants use: Google Merchant Center, ChatGPT’s merchant feed if you sell direct, and accurate content at the retailers whose apps and feeds reach those assistants.
Make sure credible third-party sources describe your products accurately.
Keep PDPs specific, structured and current, with clear answers to common questions.

Q2: Do AI Overviews help or hurt our traffic?

Both. Clicks fall sharply when an AI Overview appears, but brands cited in the Overview get meaningfully more clicks than those that aren’t. Aim to be cited, and measure AI-referred visits separately, since they tend to convert better.

Q3: Should we build for in-chat checkout?

Make your products eligible where it’s offered (for example, Google’s UCP through your retail partners or your own store), but don’t make it the strategy. ChatGPT retired its in-chat checkout after six months. Recommendation is the durable prize.

Q4: We sell almost entirely through retailers. Does GEO still matter?

Yes, and retailer content is where most of your work is. Rufus, Sparky and Instacart’s ChatGPT app all answer from retailer catalogs, reviews and Q&A. Your own site still matters as the source of truth that publications and assistants cite.

Q5: Is repeating facts across pages helpful?

Repeating verifiable product facts consistently across relevant pages, feeds and retailer listings helps assistants see a consistent signal. Repeating empty claims does not.

Q6: Do we need an llms.txt file?

It won’t hurt, but current evidence shows AI systems rarely request it. Put that effort into feeds, structured data and content first.

Implementation Checklist

[ ] ARPP built for 100% of active SKUs, versioned in your PIM
[ ] One consistent GTIN, name and size format per SKU across every channel
[ ] Product JSON-LD on every PDP with price, availability, ratings and dietary properties
[ ] Google Merchant Center complete and current; conversational attributes added where available
[ ] ChatGPT merchant feed set up (if selling direct) or retailer feeds verified
[ ] Amazon, Walmart, Target and Instacart listings audited for your top SKUs
[ ] Q&A, comparison, ingredient, usage, substitute and proof blocks on PDPs
[ ] Persona-specific category content published
[ ] Third-party review and PR program focused on verifiable product facts
[ ] Weekly AI visibility tracking across prompts, personas and platforms
[ ] AI-referred sessions segmented in analytics
[ ] Evergreen search content refreshed to keep classic search demand flowing

Why Natural State AI for GEO and AEO

We work where CPG actually sells. We’re based in Northwest Arkansas, alongside the retail and supplier teams that live this every day, and we focus on product data, retailer content and feeds, not just blog posts.
We make your catalog AI-readable. We design your product profile schema, wire structured data into your PDP templates, and build the channel mappings that keep your data consistent everywhere assistants look.
We measure what assistants say about you. We track your visibility across ChatGPT, Google AI Mode, Gemini, Perplexity and retailer assistants, by shopper persona, and turn the gaps into a prioritized fix list.

Ready to put this to work?

If you’re a CPG brand or agency and want your SKUs to show up in AI answers, not just search results, book a GEO working session with Natural State AI. We’ll audit your current AI visibility, design your AI-ready product profile, and build the plan to get your products recommended.

Sources

This guide is updated as AI shopping platforms change. Last updated October 15, 2026.