Ashish Gupta, VP/GM of Merchant Shopping at Google: UCP Deep Dive, Nerding out on Product Data Feeds, the importance of Product-Level Context PLUS Google's Vision for Agentic Commerce
Retailgentic | Consumer Behavior & Retail TrendShow Notes
This episode explores the shift from traditional search and ecommerce flows into a new agentic commerce environment where shoppers are using AI surfaces to discover, compare, and increasingly delegate parts of the buying journey. Ashish explains what Google was seeing internally that led to the development of UCP, why the protocol was designed to be open and ecosystem-wide, and why the ambition goes far beyond checkout. He also shares how Google is thinking about discovery, conversational product data, merchant readiness, and the long-term role of AI across the shopping journey. For retailers, brands, marketplaces, and commerce tech teams, this conversation is a helpful window into what Google believes is changing now, and what foundations businesses should be investing in before the market fully shifts. Highlights How Google’s large-scale ad and data systems shaped his approach to commerce Why UCP emerged now, and the two major trends that inspired it Why Google believes agentic commerce needs an open, ecosystem-wide standard Why the future of UCP extends far beyond checkout How discovery is starting to take shape within the protocol Why richer, more conversational product data matters in AI shopping experiences What retailers need to do now to improve visibility on AI surfaces How Google is thinking about the relationship between agentic commerce and advertising Why agentic commerce is still early, but already moving from concept to reality Agentic commerce is still early, but this conversation makes one thing clear: the infrastructure decisions being made now will shape who gets discovered, considered, and chosen in the next era of shopping.
Timestamps 2:32 — Welcome and introduction to Ashish Gupta 5:01 — Building Google’s data infrastructure and declarative querying at scale 9:43 — AI at Google before the current wave 11:32 — What it means to be an Engineering Fellow at Google 14:00 — Balancing technical leadership and operational leadership 16:10 — Why Ashish moved into merchant shopping 17:41 — The origin story behind UCP 24:58 — Why Google chose an open protocol approach 27:50 — Responding to skepticism around Google’s commerce efforts 32:30 — Discovery, catalog search, and where UCP expands next 35:39 — Multi-item and multi-merchant cart support 37:43 — New protocol capabilities and partner onboarding 38:42 — Why product data quality matters more in AI shopping 43:22 — How shoppers are using AI surfaces in more complex ways 47:47 — Agentic commerce and advertising 50:10 — Where agentic commerce may go from here 51:46 — Closing thoughts and where to follow Ashish
👉 Connect with Ashish: https://www.linkedin.com/in/ashishgupta98/ 👉 Check out UCP announcement: https://www.retailgentic.com/p/flash-googlenrf-announces-universal
🧠 Want to stay ahead in AI commerce? Subscribe and follow along: 📰 Subscribe to the free Substack: retailgentic.ai 📺 Watch episodes on YouTube & Subscribe for updates: https://www.youtube.com/@Retailgentic
More from Retailgentic | Consumer Behavior & Retail Trend
This week is a product deep dive with a fresh funding story attached. eComID just closed Europe’s largest fashion tech seed round, and the company is building something agentic commerce needs: shopper context that belongs to the shopper and goes wherever they shop. The ID was born inside H&M, it already runs on about 70 brand sites, and it’s now headed for the US. Highlights 𝗧𝗵𝗲 $𝟭𝟳.𝟰𝗠 𝘀𝗲𝗲𝗱 𝗿𝗼𝘂𝗻𝗱: Europe's largest fashion tech seed round. Oscar says it's the second-biggest in the world, led by London's Systemiq Capital, and Nordic sports retailer Stadium (one of eComID's biggest customers) joined as a strategic investor. Scot rang the gong 🔔 𝗕𝗼𝗿𝗻 𝗶𝗻𝘀𝗶𝗱𝗲 𝗛&𝗠: Oscar pitched eComID internally in a Dragon's Den–style process and H&M's startup arm invested from day one. 𝗔 𝘃𝗲𝗿𝘆 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝘁 𝘀𝘁𝗿𝗮𝗻𝗴𝗲𝗿: Models are very capable, catalogs are open to agents and payments are getting solved. But without shopper context, AI "still starts from zero." It doesn't know your size, your taste or what you sent back. 𝗧𝗵𝗲 𝘀𝗵𝗼𝗽𝗽𝗶𝗻𝗴 𝗽𝗮𝘀𝘀𝗽𝗼𝗿𝘁: One source of truth for your size, taste and preferences, tied to your email. It goes with you across eComID brands like H&M, Björn Borg, Peak Performance and Haglöfs. "We want to democratize personalization." 𝟵𝟭% 𝗯𝘂𝘆 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝘀𝗶𝘇𝗲𝘀: 91% of shoppers buy different sizes from different brands. Oscar says that shows how broken sizing is. 𝗧𝗵𝗲 𝘁𝗿𝗮𝗰𝘁𝗶𝗼𝗻: About 70 merchants, about 20 million shoppers reached each month, and 15% of users already in the US. Shoppers who use eComID have about 30% lower return rates and about 10% higher conversion. +𝟮𝟲% 𝗿𝗲𝘃𝗲𝗻𝘂𝗲 𝗽𝗲𝗿 𝘃𝗶𝘀𝗶𝘁𝗼𝗿: In a head-to-head test, eComID's AI search beat traditional on-site search by 26% on revenue per visitor. 𝗞𝗶𝗹𝗹 𝗮𝘃𝗲𝗿𝗮𝗴𝗲 𝗼𝗿𝗱𝗲𝗿 𝘃𝗮𝗹𝘂𝗲: "Average order value is a completely useless metric." What matters is average kept order value, because each handled return costs brands $21–46. 𝗧𝗿𝘂𝘀𝘁 𝗶𝘀 𝘁𝗵𝗲 𝗽𝗿𝗼𝗱𝘂𝗰𝘁: If an agent spends $1,000 for you, you need to know whose agent it is, what you authorized, what context it used and what happens if something goes wrong. Oscar argues Europe's GDPR-first approach suits agentic commerce. 𝗦𝗵𝗼𝘄 𝘀𝗵𝗼𝗽𝗽𝗲𝗿𝘀 𝘁𝗵𝗲𝗶𝗿 𝗿𝗲𝘁𝘂𝗿𝗻 𝗿𝗮𝘁𝗲: 92% of shoppers want to see their own return rate, and 75% say seeing it makes them want to return less. The live demo on Houdini Sportswear shows the search, sizing and nudge modules working together. Timestamps: 00:35 Welcome to Retailgentic 01:50 Announcement: Kiri Masters launches Adgentic 05:00 Welcome Oscar Rundqvist, CEO & founder of eComID 05:25 Big news: $17.4M, Europe's largest fashion tech seed round 06:15 Oscar's background at H&M 07:35 Lessons from H&M: why personalization never worked 08:55 Pitching eComID inside H&M's startup studio 09:40 Why we need eComID: AI without context starts from zero 10:55 The shopping passport: portable shopper context 13:20 Building the consumer side through brand modules 14:45 How the ID works across brands 16:30 Shopper control and a visible shopping algorithm 18:05 Beyond sizing: capturing taste and intent 20:35 The merchant suite: search, sizing, personalization, agents 23:00 Modular by design 23:50 Traction: 70 merchants, 20M shoppers, -30% returns 24:50 Why average order value is a useless metric 25:35 Use of proceeds and coming to the US 26:30 The fundraising journey and cap table 28:50 Fashion, shoes and future categories 30:15 Why AI can crack sizing when others couldn't 32:50 Product-led growth: no marketing, no sales 33:35 Agentic commerce adoption in Europe 37:15 AI platforms vs. brand sites and product context 38:55 Live demo: eComID on Houdini Sportswear 42:30 Wrap-up and how to reach Oscar 👉 Connect with Oscar: https://www.linkedin.com/in/oscarrundqvist/ 👉 Learn more about eComID: https://www.ecomid.com/ 🧠 Want to stay ahead in AI commerce? Subscribe and follow along: 📰 Subscribe to the free Substack: retailgentic.ai 📺 Watch episodes on YouTube & Subscribe for updates: https://www.youtube.com/@Retailgentic
Retailgentic listeners: this week we're sharing an episode of AdGentic, our sister show in the ReFiBuy Media Network. Retailgentic covers how AI agents are changing shopping. AdGentic covers what happens to advertising when your agent does the shopping. If you like this one, subscribe to AdGentic for more. Personal AI agents feel magical right now, and part of the magic is what's missing: ads. Kiri Masters walks through how Google, Facebook, WhatsApp and Amazon all started clean and ended up ad-funded, why AI is following the same script, and why that might not be the worst outcome. The real question isn't whether ads come to your agent. It's how bad they'll be. In this episode: → The broken no-ads promises of Google, Facebook and WhatsApp → Ensh*ttification, and why ads are still the least-worst way to pay for consumer tech → Subscriptions vs fees vs ads, and why fees quietly turn into ads → What Meta's Muse, OpenAI's Dots and Instinct have said about making money → The ad formats starting to appear in AI, and the ones agents might borrow Chapters 00:25 Welcome to AdGentic 00:50 The magic of an ad-free agent 01:55 Every platform starts like this: Google 02:30 Facebook and the scarf 03:05 WhatsApp's "No Ads!" note 03:25 Amazon: the store became the content 04:00 Ensh*ttification 04:40 Why I'm an advertising maximalist 05:05 Three ways to pay for an app 06:20 How Muse, Dots and Instinct plan to make money 07:05 The problem with subscriptions 07:30 Why fees turn into ads 08:25 How bad will it be? 09:15 AI ad formats so far 09:55 Formats agents could borrow 10:25 Off-site: ads without ads 11:20 Kiri's wish list for Ad Santa 11:50 The dark scenario 12:30 Ads are the least worst Subscribe to AdGentic Substack: https://www.adgentic.ai/ LinkedIn: https://www.linkedin.com/company/adgentic-podcast Podcast: https://adgentic.transistor.fm/ 🧠 Want to stay ahead in AI commerce? Subscribe and follow along: 📰 Subscribe to the free Substack: retailgentic.ai 📺 Watch episodes on YouTube & Subscribe for updates: https://www.youtube.com/@Retailgentic
This episode was recorded live at the RetailClub AI Festival in Huntington Beach (Sept 22–24), where Jason Goldberg and I set up shop to record with retail and e-commerce leaders between sessions. You’ll hear a few nods to the festival in the conversation, and if there’s a little extra crunch in the audio, blame the snacks. We start with a quick tour of every CPG brand Justin has worked for, and the very scientific method he uses to pick the next one. From there it turns into a crash course in how challenger brands actually grow: the brutal economics of the physical shelf, why big brand houses are reshuffling their portfolios, and what it really takes to get into Costco and stay there.
We pick up where we left off a year ago and get into what has changed since. Ekta walks us through how e.l.f. is structuring its AI work, what kind of person they’re hiring to do it, and how a public company makes the case for investing before the ROI is fully provable. The live audience jumps in with some sharp questions along the way. We close by looking ahead to what Ekta thinks brands need to get ready for in the next 12 months. Hit play.
This week is a research deep dive. ReFiBuy commissioned a behavioral study of how real people shop inside ChatGPT and Google's AI Mode, and Scot sat down with the researcher who ran it to walk through what was found. The full study is published today This is the conversation behind the numbers. Highlights: 𝟴𝟰% 𝗼𝗳 𝗰𝗵𝗼𝗶𝗰𝗲𝘀 𝘀𝘁𝗮𝗿𝘁 𝗼𝗻 𝗮 𝗰𝗮𝗿𝗱: About 84% of product choices in the study came from a product card. That's 83.7%, and Scot rounds it up "if you allow." 𝗣𝗼𝘀𝗶𝘁𝗶𝗼𝗻 𝘁𝗵𝗿𝗲𝗲 𝗯𝗲𝗮𝘁𝘀 𝗽𝗼𝘀𝗶𝘁𝗶𝗼𝗻 𝘁𝘄𝗼: The first card wins by a wide margin, with roughly 43% of picks. But card three outperformed card two in both ChatGPT and AI Mode. Eric's take is that "it's not really a ranking." Scot's guess is curiosity, because card three is only partly visible in the carousel. 𝗣𝗶𝗰𝘁𝘂𝗿𝗲𝘀 𝗯𝗲𝗳𝗼𝗿𝗲 𝗽𝗮𝗿𝗮𝗴𝗿𝗮𝗽𝗵𝘀: About 75% of shoppers skipped the AI's written answer and went straight to the product cards. They'd come back to the text later, but only as a second step. 𝟵𝟴% 𝗼𝗳 𝗾𝘂𝗲𝗿𝗶𝗲𝘀 𝘁𝗿𝗶𝗴𝗴𝗲𝗿𝗲𝗱 𝗰𝗮𝗿𝗱𝘀: Purchase-intent prompts like "I want to buy X today" brought up product cards almost every time. 𝗧𝗵𝗲 𝗳𝗶𝗿𝘀𝘁 𝗼𝗳𝗳𝗲𝗿 𝘄𝗶𝗻𝘀: When shoppers had to click out to a retailer, they leaned even harder toward the first offer than toward the first product card. The likeliest path is first card, first offer. 𝗔𝗱𝘀 𝘁𝗵𝗮𝘁 𝗽𝗲𝗼𝗽𝗹𝗲 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗹𝗶𝗸𝗲: About 6 in 10 free-plan users saw ads or sponsored content on a task. 71% rated them at least as useful as Google search ads, and about 35% said an ad influenced what they considered buying. 𝗧𝗵𝗲 𝗿𝗲𝘃𝗶𝗲𝘄𝘀 𝗽𝗮𝗿𝗮𝗱𝗼𝘅: Reviews were the main reason for a pick only 2% of the time. Afterward, about 75% said reviews mattered. Scot's explanation is that nobody reads them, but everyone takes the 4.8 over the 3.2. 𝗧𝗿𝘂𝘀𝘁, 𝘂𝗻𝘁𝗶𝗹 𝗶𝘁 𝗯𝗿𝗲𝗮𝗸𝘀: Eric says the biggest surprise after two years of these studies is "the extent to which people trust, even when they say they don't." A mismatched price or a wrong spec breaks that trust quickly. 𝗧𝗵𝗲 𝗸𝗲𝘆𝘄𝗼𝗿𝗱 𝗷𝗮𝗶𝗹 𝗲𝗿𝗮: Scot compares 20 years of chasing "men's running shoe" on Google with a world where shoppers let the AI choose for them. 𝟰𝟬 𝗽𝗲𝗼𝗽𝗹𝗲, 𝗺𝗼𝗻𝘁𝗵𝘀 𝗼𝗳 𝘄𝗼𝗿𝗸: The study had 40 remote participants doing 30-minute think-aloud sessions on screen recordings. The team logged small behaviors down to seconds spent and scroll percentages. Scot came out of the Amazon era believing the buy box was everything, and this study tells him ChatGPT has one too. Enjoy the conversation. Timestamps: 01:57 Meet Eric Van Buskirk, Clickstream Solutions 02:15 Why ReFiBuy commissioned this study 02:48 Eric's background: SEMrush, Backlinko, Kevin Indig 04:05 Stated preference vs. observed behavior 06:05 How a behavioral study works 08:30 Why 40 people is enough 10:20 Who took part, and who got cut 11:55 The Amazon buy box and agentic commerce optimization 14:25 Finding #1: Pictures beat paragraphs 16:40 Card position: #1 wins, #3 beats #2 18:55 84% of choices come from a product card 20:35 Google SEO vs. AI: the "keyword jail" era 23:40 When do shoppers click out? 25:30 75% skip the text and go straight to the cards 27:25 Offer cards: the first offer dominates 30:20 Ads in ChatGPT: what shoppers really think 35:50 Trust and the reviews paradox 40:25 When bad data breaks trust 41:35 Where to get the study and work with Eric 44:55 Wrap-up 👉 Connect with Eric: https://www.linkedin.com/in/vanbuskirk/ 👉 Learn more about Clickstream Solutions: https://clickstream.cc/ 🧠 Want to stay ahead in AI commerce? Subscribe and follow along: 📰 Subscribe to the free Substack: retailgentic.ai 📺 Watch episodes on YouTube & Subscribe for updates: https://www.youtube.com/@Retailgentic