Robots With Wallets, Futureproofing Product Feeds For Agentic Shopping

Transcript

1. Robots With Wallets, Futureproofing Product Feeds For Agentic Shopping Henry Price Dentsu @Linkedin
2. How do users discover your products? Free Listings 2
3. They aren’t just found in the SERP
4. Image Search
5. LLM’s now aid the discovery 5
6. Agentic shopping is the next step They match preferences to products 6
7. Robots will begin to shop for us 7
8. All results powered by shopping (feed) attributes 8
9. UCP will provide instant purchase Google UCP 9
10. Feed Attribute Breakdown 10
11. What is a feed? 11
12. The feed divide 12
13. Google feed now compatible with ChatGPT Source: https://developers.openai.com/commerce/specs/file-upload/products#google-compatible-productdata-feeds 13
14. + attributes are optional MANDATORY id, title, price, link, image_link. OPTIONAL Enriched product information 14
15. Requirements that can be optimised Product Title Product Description Product Details 15
16. REQUIRED Translating product data into structured titles TITLE FORMULA Brand · Product Name · Category · Attributes What should you do? Include colour, size, material, model. Nike Air Force 1 Shadow Women's Trainers – Front-load terms in the first 70 characters. White - Size: 4.5 16
17. REQUIRED Deep context with product descriptions Lead with function and material in sentence one. Embed details like “waterproof” or “BPA-free”. Powers organic ranking & LLM visibility 80% of the organic weight Is carried by the first 160 to 500 words. 17
18. REQUIRED Product details matter LLMs and search engines will match products based on personal preferences for colour, size, material Ensure main features are in your feed. Size [Required] Color [Optional] Material [Optional] Dimensions [Optional] 18
19. Optional that really are required Product Type Product Highlight Q&A Conversational 19
20. OPTIONAL Categorisation with product types “product_type” index's products for relevant and long tail search queries Mirror your site’s breadcrumb taxonomy exactly. Go 3 to 5 levels deep to match specific queries. Keep paths stable so their value compounds. 1 2 3 4 5 6 7 Mens › Footwear › Trainers › Asics › Gel-Contend > Twilight Blue > Size 9 20
21. OPTIONAL Highlighting through quick scan value “product_highlight” powers quick-scan bullets on shopping detail pages and feeds AI Overviews and AI-driven search. Format Specs Quantity Submit 4-6 Minium (Max 100) Length 1 – 150 characts per bullet point Format Quick fragments focused on specs 21
22. Introduction to conversational attributes Attributes designed specifically for AI interactions 1 2 3 Answers & Documentation Variant & Family Matrix Relationships & Popularity Question_and_answer Item_group_title Related_product Document_link Variant_option Popularity_rank 22
23. OPTIONAL Questions and Answers Answer in one or two plain sentences & maximum 1,000 characters Cover fit, compatibility, care and delivery. Do not repeat what is found in the description EXAMPLE “Is this compatible with iPhone 16?” IMPACTS ORGANIC FREE LISTINGS AND AI MODE/AIO VISIBILITY “Yes, it works with every USB-C iPhone model.” 23
24. Document_link URLs to PDF manuals, care guides and spec sheets. A clean product family name that groups variants Machine-comparable options such as Variant_option colour and size Maps accessories, required parts and Related_product substitutes. Item_group_title 24
25. LLMs & crawlers struggle to find critical links between related products due to the API requirements 25
26. How can a LLM know to serve me relevant products? 26
27. Related Product advises LLms what to serve next “related_product” provides retailers the ability to tell LLMs what the next best product would be for a user's preference Four key variations to consider Part of set [part_of_set] Required part [required_part] Often bought with [often_bought_with] Accessory [accessory] 27
28. Prioritise optimising attributes that influence AIO & organic shopping Attribute Product Highlight Product Detail Question and Answer Product title & description Product Type Related Products Variant Option Document Link Impact/Priority High High High Medium Medium Medium Low Low 28
29. Technical Best Practices 29
30. Aligning feeds and on-page data Feeds match PDP content, word for word on key attributes. No conflicting availability, price or attribute values across sources. 30
31. Crawl before you think A clearly displayed, machine-readable product information is required for product visibility. Price via client-side JavaScript This is the first data point LLMs extract when generating product comparisons. Google Bot GMC Bot Price cannot be shown only with client-side JavaScript rendered. GMC bot or LLMs won't render this content. 21
32. Reporting on Performance Optimisation only matters if you can prove it. 32
33. Shopping performance matrix Main Search Image Search Other Placements Totals Google Merchant Center Google Search Console Google Analytics 4 Third party Providers Data available Setup required Unavailable 33
34. Google Merchant Center The product data hub behind every Google shopping surface Clicks and impressions per item ID. A single source of truth for every item. Live alerts before errors cost visibility. 34
35. Search Console vs Merchant Center Both platforms are required for organic shopping strategy Google Search Console Google Merchant Center 1. Query-level data. 1. Product-level performance. 2. Merchant schema report. 2. Feed diagnostics. 3. Breakdown by merchant data. 3. Competitor reviews. 35
36. Google Merchant Center AI Report AI Search Intent AI Search queries AI Attributes US ONLY 36
37. Key Takeaways 1 Feed the machine, not just the marketplace. 2 Speak the AI's language. 3 Keep every source in sync. 4 Measure across the whole matrix. 37 37
38. Thank you!