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.

Highlights:

• 𝗟𝗲𝘁 𝗶𝘁 𝗿𝗶𝗽 𝗼𝗿 𝗹𝗼𝗰𝗸 𝗶𝘁 𝗱𝗼𝘄𝗻: Ekta says companies tend to pick one of two AI philosophies, giving it to everyone or locking it down. e.l.f. has tried to do both so it doesn't kill innovation.
• 𝗧𝗵𝗿𝗲𝗲 𝗽𝗼𝗱𝘀: e.l.f. runs three AI pods covering agentic commerce/growth, intelligent operations, and the human-agent workforce. Each pod owns data, change management, integrations, and training. The agentic growth roles are now being hired as dedicated full-time positions across e.l.f.'s brands.
• 𝗡𝗼 𝗼𝗻𝗲 𝗵𝗮𝘀 𝗺𝗲𝗮𝘀𝘂𝗿𝗲𝗺𝗲𝗻𝘁 𝗳𝗶𝗴𝘂𝗿𝗲𝗱 𝗼𝘂𝘁: "Anybody that tells you they have figured it out haven't figured it out." Ekta measures readiness instead, and focuses on the products where e.l.f. has the right to win.
• 𝗥𝗲𝗮𝗹 𝗽𝗿𝗼𝗺𝗽𝘁𝘀, 𝗳𝗶𝗻𝗮𝗹𝗹𝘆: Jason points out that most visibility tools rely on synthetic prompts. Google's new Agentic Insights tab in Merchant Center is the first look at what real people actually type, at least on Google surfaces.
• 𝗕𝗮𝗱 𝗱𝗮𝘁𝗮 = 𝗯𝗮𝗱 𝗔𝗜: Ekta's case for fixing the foundation first. If you don't solve the plumbing, you can't scale.
• 𝗘𝘅𝘁𝗲𝗿𝗻𝗮𝗹 𝘃𝗼𝗶𝗰𝗲𝘀 𝗹𝗶𝗴𝗵𝘁 𝘁𝗵𝗲 𝗳𝗶𝗿𝗲: About 14 months ago e.l.f. brought in an outside search expert, and that put "fire in the belly."
• 𝗦𝘂𝗽𝗽𝗹𝘆 𝗰𝗵𝗮𝗶𝗻 𝗶𝘀 𝘁𝗵𝗲 𝘀𝗹𝗲𝗲𝗽𝗲𝗿: e.l.f. has invested in Lyric for forecasting and demand planning. Its planners "are actually now engineers," and AP, AR, payroll and inventory teams are building their own AI workflows.
• 𝗔𝗴𝗲𝗻𝘁 𝗿𝗲𝗹𝗮𝘁𝗶𝗼𝗻𝘀𝗵𝗶𝗽 𝗺𝗮𝗻𝗮𝗴𝗲𝗺𝗲𝗻𝘁: Ekta's prediction is that Meta wins the agent shopping race. She says brands need to put agent identity in place now, so they know which agent is arriving and which customer it represents. It's the next CDP/CRM-style plumbing project.

Huge thanks to Ekta for running over from the main stage, and to our live audience at RetailClub for the great questions. Enjoy the conversation.

Timestamps:
01:17 Live from Retail Club: meet Ekta Chopra & co-host Jason Goldberg
02:04 One-year reunion and a new title
02:38 e.l.f. Beauty: a house of brands, 30 straight growth quarters
03:19 Eyes, lips, face… and the Jeopardy! moment
03:43 Ekta's role as Chief Technology & AI Officer
04:02 From the main stage: let it rip or lock it down
05:04 e.l.f.'s three AI pods
06:14 Making agentic growth a full-time job
07:00 Who do you hire for agentic commerce?
08:33 Jason's "you've reached your quota" hiring hack
08:48 Selling AI investment to a public company board
10:06 Measuring readiness when no one has measurement figured out
11:31 "My board member couldn't find our product"
13:06 Synthetic prompts vs. real prompts: Google's Agentic Insights
14:00 Audience Q: build vs. buy
14:52 Where to start: external voices and fire in the belly
16:23 Foundation first: bad data equals bad AI
17:47 Back office, supply chain, and Lyric
19:52 Personal shopping assistants and agents
20:59 Serving a house of brands and working with retail partners
22:55 Audience Q: is governance keeping pace?
24:50 Next 12 months: Meta, agent identity & agent relationship management
26:20 Wrap-up

👉 Connect with Ekta: https://www.linkedin.com/in/ektachopra/
👉 Learn more about e.l.f. Beauty: https://www.linkedin.com/company/e-l-f-beauty/

🧠 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 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

This week’s episode is a platform deep dive with one of the longest-tenured companies in e-commerce. Elastic Path has been building commerce software for 26 years and Bryan is now steering the company straight into the agentic era.

It’s part founder-journey (craft brewing to CEO is not a typical path), part product deep dive, and part industry-shift conversation about what AI does to the economics of running a commerce platform.

Highlights:

• 𝗙𝗿𝗼𝗺 𝗯𝗿𝗲𝘄𝗲𝗿 𝘁𝗼 𝗖𝗘𝗢: Bryan spent eight years as a professional brewer, with an actual brewing degree, at a brewpub across from Fenway Park. "Everything I learned in my career that really moved the needle for me, I learned as a brewer."
• 𝟮𝟲 𝘆𝗲𝗮𝗿𝘀 𝗼𝗳 𝗰𝗼𝗺𝗺𝗲𝗿𝗰𝗲: Elastic Path's claim to fame was building the store for the Vancouver 2010 Winter Olympics, the birth of its self-managed commerce product.
• 𝗧𝗵𝗲 $𝟱𝟬𝟬𝗞 𝘂𝗽𝗴𝗿𝗮𝗱𝗲, 𝗸𝗶𝗹𝗹𝗲𝗱: Elastic Path built an AI updater so one developer can upgrade a self-managed platform in about two weeks, replacing what used to be a half-million-dollar annual project.
• 𝟭𝟬,𝟬𝟬𝟬 𝘃𝗮𝗿𝗶𝗮𝘁𝗶𝗼𝗻𝘀: where typical platforms cap out at 250 variations per item, Elastic Path supports 10,000, plus bundles of up to 1,500 items.
• 𝗦𝗶𝘅 𝘄𝗲𝗲𝗸𝘀 𝗮𝗻𝗱 𝟭𝟬𝟬 𝗴𝗿𝗮𝗻𝗱: Bryan's prediction… AI takes re-platforming from a 6-month-to-2-year, $500K–$3M digital transformation down to a setup fee. "That's the place we're gonna see the biggest impact of AI in commerce."
• 𝗧𝗵𝗲 𝟴𝟬-𝗽𝗮𝗴𝗲 𝗣𝗗𝗙: in a world of AI discovery, every product needs the equivalent of an 80-page PDF of data to train the LLM on when to recommend it. (Scot's line, which Bryan says he "steals with credit" as often as he can.)
• 𝗣𝗿𝗼𝗼𝗳 𝗶𝗻 𝗻𝘂𝗺𝗯𝗲𝗿𝘀: a UK custom greeting-card seller saw conversion jump 40% almost immediately after adopting Elastic Path's checkout
• 𝗡𝗼𝗯𝗼𝗱𝘆 𝘄𝗮𝗻𝘁𝘀 𝗰𝗵𝗮𝘁𝗯𝗼𝘁𝘀: customers are asking about UCP and ACP, discovery, and the long tail, not chatbots.
• 𝗪𝗵𝗲𝗿𝗲 𝘁𝗵𝗲 𝗿𝗼𝗮𝗱𝘀 𝗮𝗿𝗲 𝘄𝗲𝗹𝗹-𝗽𝗮𝘃𝗲𝗱: Bryan's framework for where AI lands first… well-understood patterns like checkout, front ends, and reorders, not judgment-heavy edge cases.

Bryan is a product guy after our own heart, and this one gets satisfyingly deep, from CPQ to protocols to the future of re-platforming. Enjoy the conversation.

Timestamps:
03:20 Welcome Bryan + podcasting since 2009
04:05 Internet Retailer booth memories
04:55 From craft brewing to software
07:35 The Acquia and Drupal years
08:20 Moltin, Neural Magic, joining Elastic Path
11:05 Elastic Path history: the Olympics store
13:20 Two products: SaaS + self-managed
13:45 The AI updater: killing the $500K upgrade
14:50 80 people, profitable, constraints breed innovation
17:00 Customer base and the B2B focus
19:15 CPQ, PXM, and 10,000 variations
22:45 Serena & Lily case study
24:10 The MACH Alliance backstory (spicy)
29:55 Vibe coding the front end + Shopify as fintech
33:20 AI re-platforming: six weeks and $100K
39:00 The ReFiBuy partnership + the 80-page PDF
40:55 AI roadmap: workflows, promotions, analytics
43:20 B2B agentic commerce and EDI
47:10 UCP, ACP, and the long tail
49:05 Predictions: where the roads are well-paved
51:20 What's next for Elastic Path

👉 Connect with Bryan: https://www.linkedin.com/in/bhouse/
👉 Learn more about Elastic Path: https://www.elasticpath.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

This week’s episode is a little different. We’ve has been noticing vibe-code platforms quietly showing up all over digital retail… at trade shows, in conversations with retailers and brands, and on the platforms’ own commerce pages. So we went to the source.

This is a platform deep dive with Vercel, the enterprise company behind v0 and Next.js, on what a “vibe coding” company is actually doing in our world of commerce, and it turns out the answer is: a lot more than you’d think.

Highlights

• 𝗧𝗵𝗲 𝟭𝟬𝟬-𝗺𝗶𝗹𝗹𝗶𝘀𝗲𝗰𝗼𝗻𝗱 𝗺𝗮𝘁𝗵: A Deloitte Digital + Google study cites clients found 100ms of load-time reduction drives a 7–8% lift in e-commerce conversion. “If the thing is faster, you’re gonna convert more. It’s just math.”
• 𝗛𝗲𝗹𝗹𝘆 𝗛𝗮𝗻𝘀𝗲𝗻’𝘀 𝗻𝘂𝗺𝗯𝗲𝗿𝘀: The Kontoor Brands-owned retailer cut load times from 2.5 to 1.5 seconds on Vercel, grew conversion 20% year over year, doubled its A/B test velocity, and blew away its Black Friday/Cyber Monday plan.
• 𝗧𝗵𝗲 𝗰𝗼𝗰𝗸𝘁𝗮𝗶𝗹-𝗽𝗮𝗿𝘁𝘆 𝗽𝗶𝘁𝗰𝗵: Vercel in three phrases… a developer tools company, an infrastructure provider, and an open-source software factory.
• 𝗬𝗼𝘂’𝘃𝗲 𝗮𝗹𝗿𝗲𝗮𝗱𝘆 𝘂𝘀𝗲𝗱 𝗩𝗲𝗿𝗰𝗲𝗹: walmart.com and target.com run on Next.js, and Under Armour, Crocs, Buck Mason, PetSmart, McAlister’s, DeWalt, Clorox, and even Meta.ai are Vercel-delivered applications.
• 𝗡𝗼𝘁 𝗮 𝗦𝗵𝗼𝗽𝗶𝗳𝘆 𝗸𝗶𝗹𝗹𝗲𝗿: Vercel doesn’t do product catalogs, inventory, or order management. It’s the storefront layer on top of your commerce platform. “We help you own your storefront and own your destiny.”
• 𝘃𝟬 𝗺𝗲𝗲𝘁𝘀 𝗦𝗵𝗼𝗽𝗶𝗳𝘆: v0 is now deeply integrated with Shopify, so you can customize your storefront in plain English
• 𝗧𝗶𝗺𝗲𝗹𝗶𝗻𝗲𝘀 𝗵𝗮𝗹𝘃𝗲𝗱: Custom storefront projects that were planned for a year are landing in six months, six-month plans in three.
• 𝗧𝗵𝗲 𝗮𝗴𝗲𝗻𝘁 𝗰𝘂𝘀𝘁𝗼𝗺𝗲𝗿: Next.js 16 was built to be readable and manipulated by Cursor, Claude, and Codex, “now we have this new customer, this agent customer, who wants to interact with your code base.”
• 𝗧𝗵𝗲 𝗹𝗶𝗴𝗵𝘁-𝗯𝘂𝗹𝗯 𝗺𝗼𝗺𝗲𝗻𝘁: Jaime’s conversion story: feeding 50-page client 10-Ks into ChatGPT in early 2023 and realizing everything about go-to-market, and the factories that make software, was about to change.

This one really lifted the veil on what the vibe-code platforms are actually doing in retail, and I think we’re just seeing the tip of the iceberg. Enjoy the conversation.

Timestamps:
05:29 Welcome Jaime Sorgente
06:08 Jaime's path to Vercel: ChatGPT and 10-Ks
12:29 Utilities vs. builders
15:00 What is Vercel? The cocktail-party pitch
17:10 Why retail as a vertical
21:20 The Helly Hansen story
24:05 Vercel's other verticals: digital natives and marketing
26:45 Enabler, not competitor: where Vercel fits vs. Shopify & Salesforce
29:35 Beyond the storefront: clienteling and staff apps
30:45 Agents everywhere: Crocs, Amazon image annotation
36:05 v0 + Shopify: customize your storefront in English
37:45 Next.js 16: built for the agent customer
39:20 Custom storefront timelines, halved
40:35 Does owning your storefront de-risk replatforming?
42:20 The 100-millisecond study
44:30 The compounding effect: 4X feature velocity
46:45 Next.js explained: the stage manager analogy
52:20 Framework-defined infrastructure
54:30 Vercel vs. Lovable, Replit, and Base44
58:15 Where to learn more + wrap-up

👉 Connect with Jaime: https://www.linkedin.com/in/jsorgente/
👉 Learn more about Vercel: https://vercel.com/for/retail

🧠 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 week’s episode is a deep dive into personalization, a huge category in e-commerce that’s getting rebuilt from the ground up for the AI era.

FERMÀT is an AI-native personalization platform that sits on top of your e-commerce platform and customizes the site experience based on a visitor’s behavior, not their identity. That distinction sounds subtle, but it turns out to be the whole ballgame and it connects directly to agentic commerce because agents show up with intent, not an identity. We also pick up Rishabh’s founder journey and some very candid company-building advice along the way.

Highlights

• 𝗧𝗵𝗲 𝟮𝟬% 𝗰𝗼𝗻𝘃𝗲𝗿𝘀𝗶𝗼𝗻 𝗹𝗶𝗳𝘁: FERMÀT's data shows personalizing on behavior alone, even for fully incognito visitors, converts roughly 20% better than personalizing on identity and past purchases.
• 𝗠𝗮𝗻𝘆 𝘁𝗼 𝗼𝗻𝗲, 𝗻𝗼𝘁 𝗼𝗻𝗲 𝘁𝗼 𝗼𝗻𝗲: the marketer's dream of one-to-one was "a lack of imagination", the same person is a different shopper on Instagram, Google, and ChatGPT.
• 𝗗𝗲𝘁𝗲𝗰𝘁, 𝗿𝗲𝗰𝗼𝗺𝗺𝗲𝗻𝗱, 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗲: instead of humans watching a sample of session recordings and shipping site changes months later, agents trained on commerce watch every session, recommend changes, and generate the new experience.
• 𝗧𝗵𝗲 𝗕𝗹𝗼𝗼𝗺𝗿𝗲𝗮𝗰𝗵 𝗽𝗿𝗼𝘃𝗼𝗰𝗮𝘁𝗶𝗼𝗻: identity-based personalization was state-of-the-art 10 years ago. Rishabh's take on the incumbents: if nothing changes, "their 3 to 400 million of revenue will become 2 to 300."
• 𝗠𝗶𝗻𝗲 𝘁𝗵𝗲 𝗮𝗱, 𝗻𝗼𝘁 𝘁𝗵𝗲 𝗜𝗗: even when an Instagram visitor is anonymous, the ad they clicked carries demographics, performance history, and frequency data… context most brands never use.
• 𝗧𝗵𝗲 𝗿𝗲𝗮𝗹 𝗲𝘀𝘁𝗮𝘁𝗲 𝗮𝗴𝗲𝗻𝘁 𝘁𝗲𝘀𝘁: an agent is only as good as the feedback, context, and iteration speed you give it and every "actually, I also care about the collar" update is behavior data.
• 𝗦𝗮𝘆 𝘄𝗵𝗮𝘁'𝘀 𝘁𝗿𝘂𝗲: Platforms are using AI to verify your pages and truth gets rewarded, not dinged.
• 𝗦𝘂𝗽𝗽𝗼𝗿𝘁 𝗮𝗴𝗲𝗻𝘁 ≠ 𝘀𝗵𝗼𝗽𝗽𝗲𝗿 𝗮𝗴𝗲𝗻𝘁: support agents should stick to facts and policies; shopper agents should infer.

Rishabh is one of the clearest thinkers I’ve talked to on where personalization goes in an agent-first world, and his founder advice at the end is worth the listen on its own. Enjoy the conversation.

Timestamps:

02:44 Welcome Rishabh
03:37 Rishabh's backstory: physics, Wall Street, PhD
06:33 Building three businesses at LiveRamp
07:56 Founding story: Apple ATT and iOS 14.5
10:33 Detect, recommend, generate
12:27 Where FERMÀT sits in the stack
13:57 Many-to-one personalization
15:22 The Bloomreach question
19:00 Does current behavior beat past behavior?
21:18 The data: a 20% conversion lift
22:34 The agentic commerce connection
23:44 The real estate agent analogy
25:41 Reading intent from incognito traffic
26:28 Into the weeds: mining a Meta ad for context
30:37 Side quest: the SEO cloaking myth
33:47 Writing for agents, bots, and humans
34:49 On-site shopper agents
39:40 Company building: know your why
42:38 Where to find Rishabh + the 1 to 100 podcast

👉 Connect with Rishabh: https://www.linkedin.com/in/rishabhmjain/
👉 Learn more about FERMÀT: https://www.fermatcommerce.com/
👉 1 to 100 podcast: https://open.spotify.com/show/6WAB5oOzco9oWrQmYmuFZ6

🧠 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 week’s episode is a data deep dive… two of them, actually. Salesforce just dropped a back-to-back data release: the latest edition of its State of Commerce report and its Holiday 2026 predictions. Caila Schwartz, Salesforce’s newly minted Head of Agentic Commerce Shopper Insights, walks us through both. Think of it as a rearview mirror on the first half of 2026 and headlights into the holiday season, with agentic commerce showing up in nearly every number.

Highlights:

• 𝗧𝗵𝗲 𝟮𝟬𝟬% 𝗷𝘂𝗺𝗽: 15% of shoppers now start their buying journey with an LLM or agent, up from 5% last spring.
• 𝗛𝗮𝗹𝗳 𝗼𝗳 𝘀𝗵𝗼𝗽𝗽𝗲𝗿𝘀: About 50% have used a general-purpose LLM, social media agent, or branded agent somewhere in their buying journey in the past 12 months, a 67% increase over last year.
• 𝗖𝗼𝗻𝘃𝗲𝗿𝘀𝗶𝗼𝗻 𝗸𝗶𝗻𝗴: Agentic search traffic converts at roughly 2.5–3.5%, about 2–3x overall conversion and far above social’s sub-1%.
• 𝗧𝗵𝗲 𝘁𝗿𝘂𝘀𝘁 𝗻𝘂𝗺𝗯𝗲𝗿𝘀: 74% of shoppers trust agent recommendations, and over 80% feel more confident in purchases made after interacting with a brand-owned agent like Rufus or Sparky.
• 𝗧𝗵𝗲 𝗿𝗲𝗳𝗲𝗿𝗿𝗮𝗹 𝗯𝗹𝗶𝗻𝗱 𝘀𝗽𝗼𝘁: Discovery is happening inside conversations that never send a click back to the brand site.
• 𝗧𝗵𝗲 𝗖𝗮𝗶𝗹𝗮 𝗙𝗶𝘃𝗲: Salesforce’s Holiday 2026 predictions: non-human traffic, branded agents everywhere, social commerce acceleration, the store’s new hybrid role, and the free-shipping squeeze.
• 𝟮𝟬% 𝗻𝗼𝗻-𝗵𝘂𝗺𝗮𝗻 𝘁𝗿𝗮𝗳𝗳𝗶𝗰: This holiday, Salesforce predicts about one in five e-commerce visits will come from crawlers and agents, not people.
• 𝗕𝗿𝗮𝗻𝗱𝗲𝗱 𝗮𝗴𝗲𝗻𝘁 𝗿𝗮𝗰𝗲: About a third of B2C orgs have agents deployed today, over half plan to deploy within six months, and by Cyber Week one in three e-commerce sites will have a branded agent on site.
• 𝗦𝗼𝗰𝗶𝗮𝗹 𝗴𝗿𝗼𝘄𝘀 𝟵𝗫: Social commerce is predicted to grow nine times faster than traditional e-commerce this year, and it’s high-income shoppers, even more than Gen Z, saying they’ll buy in-app this holiday.
• 𝗛𝘆𝗯𝗿𝗶𝗱 𝗵𝗼𝗹𝗶𝗱𝗮𝘆: 37% of holiday dollars will be “hybrid sales” that rely on both digital and physical to close the deal, 80% of shoppers use their phone while standing in your store.
• 𝗗𝗮𝘁𝗮 𝗶𝘀 𝘁𝗵𝗲 𝗯𝗲𝗱𝗿𝗼𝗰𝗸: Four in five commerce pros say their tech stack grew over the last two years, making data unification the #1 barrier to good agents.

Caila brings the receipts, real numbers on a shift most of us are still trying to eyeball. We’ll get her back in the fall when the actual holiday forecast drops. Enjoy the conversation.

Timestamps:

06:30 Interview begins, Caila's new title
07:20 Caila's role and 12 years at Salesforce
08:30 Two reports: State of Commerce + Holiday predictions
09:30 State of Commerce methodology
12:50 What's new in this edition
14:45 The agentic discovery shift: 50%, 200%, 15%
17:25 Who's losing share in discovery
19:00 Faster than mobile or social ever moved
22:20 The measurability problem
25:20 Conversion: 2–3x better than average
27:10 Why shoppers lean on agents, by income
29:30 Trust: 74% and the 80% confidence stat
32:00 The branded-agent race
34:30 The #1 barrier: data unification
40:20 Caila Five #1: 20% non-human traffic
42:30 #2: One in three sites get a branded agent
43:15 #3: Social commerce grows 9x
44:25 #4: The store's new role 37% hybrid sales
46:20 #5: Free shipping and the K-shaped economy
47:30 Decoupled transactions deep dive
53:10 Top recos: unify data, turn on ACP

👉 Connect with Caila: https://www.linkedin.com/in/caila-schwartz/
👉 Learn more about Salesforce: https://www.salesforce.com/
👉 Check out the State of Ecommerce: https://www.salesforce.com/commerce/resources/state-of-commerce/
👉 Look at the Holiday 2026 Predictions: https://www.salesforce.com/blog/holiday-retail-predictions-2026/

🧠 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