Sephora loves ChatGPT - and vice versa

Does your favorite retailer have their app inside OpenAI’s ChatGPT yet?

Anca Marola, Chief Digital Officer of Sephora, said on LinkedIn Wednesday (March 24):

“We’re excited to announce today the Sephora App inside ChatGPT! Our Beauty addicts will have the possibility to get personalized advice from Sephora directly where they are, in ChatGPT. What I love the most about this is our clients will get the best of both worlds: the AI power of ChatGPT boosting the beauty expertise and benefits of Sephora.” – Anca Marola, Chief Digital Officer, Sephora on LinkedIn (March 24)

OpenAI CEO of Applications Fidji Simo “loved” Ms. Marola’s LinkedIn post.

More: “Sephora App in ChatGPT Brings a New Personalized Beauty Experience” (March 24) – Sephora

From tipsheet #1: Now that the app is in ChatGPT for the retailer, what’s next:

Ads: Are Sephora and its competitors thinking about ads, and offers, in ChatGPT? You bet.
RMN: Could Sephora or any retailer offer ad placements for its RMN within their app which is embedded in the ChatGPT app? I’m still waiting to find out. OpenAI can take a cut, too, if it wants.
Shopping assistants as a “Trojan Horse”: Building on the RMN-within-retailer-app-within-ChatGPT idea, could shopping assistants like Walmart’s Sparky or Amazon’s Rufus be the “Trojan Horse” for the RMN and offer not only a robust shopping assistant to a retailer like Sephora, but an easy way to create and sell RMN inventory through something like Rufus’ Sponsored Prompts.I ‘smell’ a new kind of chatbot ad network led by ChatGPT shopping assistants.

From tipsheet #2: The race is ON for retailers who need their own app, i.e. their own conversational website, within ChatGPT now that OpenAI has mothballed its Instant Checkout initiative.

Consequently, is the ads-in-a-chatbot opportunity about to get a lot bigger courtesy of retail media networks inside of ChatGPT’s growing retailer app ecosystem?

LLMs & CHATBOTS

Developments

“China reviews $2bn Manus sale to Meta as founders barred from leaving country” (March 25) – Financial Times (subscription)
“Sam Altman says OpenAI’s next model finished pretraining; Fidji Simo becomes CEO of ‘AGI Deployment’” (March 25) – Alex Heath’s Sources
TurboQuant: Redefining AI efficiency with extreme compression (March 24) – Google Research

PLATFORMS

Meta targets RMNs with AI in-app

At eCommerce event ShopTalk in Las Vegas, Meta rolled out a new AI toolkit for creators focused on product discovery. Retail media networks received a boost from the company’s AI capabilities, too.

Meta’s Director of Product, Commerce and Signal Growth, Divya Amarnath, explained on LinkedIn (my summary in bold):

“The way people discover and buy products has fundamentally changed. Creators have become the most trusted voices in shopping. (…)

Here’s a snapshot of what we announced:”

Affiliate marketing opportunity expands: “Putting creators at the center of commerce — We’re giving creators new ways to tag and recommend products directly in their content, and earn from it. We’re expanding affiliate partnerships across Facebook and beginning to test them on Instagram, with partners spanning the U.S., Latin America, and Asia. This enables businesses to harness the influence and creativity of creators to drive sales while helping communities discover products from the voices they trust most.”

AI-powered content recommendations: “Using AI to remove friction from the shopping journey — We’re bringing AI into post-click experiences to surface the product details people need to buy with confidence: reviews, recommendations, and brand info right when it matters.”

One-click checkout: “And we’re testing a new one-tap checkout experience with partners like PayPal and Stripe (with Adyen and Shopify to come) so people can go from discovery to purchase without leaving the app.”

Optimization for retail media networks (RMNs): “Giving retail media advertisers more control — We’re introducing product-level optimizations so brands and retail media networks can be more precise about which products get promoted, with clearer reporting to match.”

RESEARCH

OpenAds CEO Liss on AI and ads

New Street Research analyst Dan Salmon interviewed Steven Liss, co-founder and CEO of AI-native ad network and platform OpenAds, last Friday.

In a note to investors yesterday, Salmon shared a summary of compelling “key topics” covered by Mr. Liss including:

“AI reallocates engagement to Google/Meta AI surfaces without expanding total time”
“Contextual advertising improves, benefiting Meta, open-web, The Trade Desk & PubMatic”
“Frontier models commoditize, Big Tech wins on data, distribution, and infrastructure”
“ChatGPT monetization remains constrained. Criteo, Shopify (likely), The Trade Desk supporting”
“Amazon advantaged as AI improves discovery faster than transactions”
“Reddit, Pinterest increasingly act as inputs into AI-driven discovery”
“AI lowers barriers to entry in ad tech, pressuring incumbents like The Trade Desk, Magnite, PubMatic”

EVENTS

NewFronts: Yahoo and chatbot signals

As part of its appearance at IAB’s NewFronts, Yahoo announced that it is taking data generated by its Scout chatbot and using it for ad targeting.

Adweek reported:

“The centerpiece of Yahoo’s presentation was Yahoo Scout, an AI answer engine and intelligence platform now in beta in the U.S. The company is embedding Scout across its entire product portfolio, including mail, finance, and sports, to deepen audience signals and, in theory, give advertisers a richer view of consumer intent than traditional targeting allows.”

Read: “Introducing Yahoo Scout, a New AI Answer Engine” (March 24) – Adweek

From tipsheet: So, potentially, conversational retargeting is beginning to connect with Yahoo’s formidable identity graph from Yahoo Mail. More to come, no doubt.

This is the sort of product that any publisher could benefit from – retargeting the conversation and informed by a huge identity dataset — as long as it’s privacy-safe, user-approved.

EVENTS

ShopTalk threads: Gap, Google and Reddit

From ShopTalk:

Gap CTO Sven Gerjets announced on LinkedIn (March 25):

“We’re integrating Bold Metrics’ Agentic Sizing Protocol to bring personalized fit guidance directly into AI-driven shopping flows. And we’re supporting Google’s new Universal Commerce Protocol (UCP), with the aim to enable more seamless conversational and agentic commerce experiences.”
Read: Gap Inc. Elevates Online Shopping with AI-Powered Fit and Conversational Checkout (March 24) – Gap

Google VP/GM, Advertising and Commerce, Vidhya Srinivasan commented on Gerjets’s LinkedIn post (March 25):

“Sven, thank you and the Gap Inc. team for collaborating with us to start integrating the Universal Commerce Protocol! Soon, people will be able to purchase your products directly within AI Mode and the Gemini App. Removing hurdles in the digital shopping journey one step at a time!”

Reddit Product Lead, Shopping and Commerce, Vinay Sridhar promoted new product listing ads and a Shopify integration on LinkedIn (March 25):

From Reddit on LinkedIn: “We’re expanding our shopping solutions with new tools that help retailers and ecommerce brands turn high-intent conversations into conversions, including a new Shopify integration that simplifies setup and scaling from day one. This also includes Collection Ads to connect discovery to purchase, and Community and Deal overlays that surface trusted signals from Reddit.”
Read: “Introducing More Ways to Tap into Shopping on Reddit” (March 24) – Reddit

LLMs & CHATBOTS

World models and Conversational DOOH

In new research for investors focused on private markets, Morgan Stanley analyst Adam Jonas took a look at AI’s move into the physical in a piece titled, “World Models: AI’s Journey from Digital to Physical.” (March 22)

Though the research is not ads and marketing-specific, it’s an AI trend worth watching.

Jonas’ thesis:

“AI’s Next Challenge: Modeling the Physical World. LLMs train on, process, and generate text in its many forms. While they have proven to be powerful tools for white-collar tasks such as coding, search, and writing, AI’s broader potential may lie in the physical world, which is governed by the laws of physics – substances, thermodynamics, fluid dynamics and the behavior of light – in a constantly changing 3-dimensional space.

‘World Models’ are AI systems designed to understand, simulate, and reason about environments, with applications spanning video games, design, robotics, autonomous vehicles, and more. Potential use cases range from generating video game content from text prompts, to robots simulating outcomes before acting, autonomous vehicles training on billions of rare edge cases, or architects modeling entire cities before construction begins.

He shared some of the following examples on progress with world models. Again, though they are not ads- and marketing-specific, it’s easy to see the intersection.

Jonas’ “anecdotes”:

“Recent research has suggested that robots trained using data generated by world models can perform comparably to those trained on real-world interaction data.”
“Waymo (Leveraging DeepMind’s Genie 3 World Model) has reported running billions of miles of virtual driving tests to train and validate its systems across rare edge cases that would be difficult or dangerous to encounter in the real world. Waymo is a subsidiary of Alphabet.“
“Last year, Microsoft unveiled a fully AI-rendered, playable version of 1997 shooter Quake II built on its Muse world model instead of a traditional game engine.“
“Earlier this year, Roblox announced a new research project with the goal of using its own world model to generative immersive environments and iterate on games through natural language prompts.“

From tipsheet: One prediction… given world models, digital-out-of-home (DOOH) advertising surfaces and spend explode — a good explosion.

Conversational DOOH, anyone?

PLATFORMS

Google on lead quality in the AI era

The latest episode of Google’s Ads Decoded podcast pitches video platform YouTube as a performance marketing channel.

Moderated by Google’s Ads Product Liaison Ginny Marvin, Global Head of Lead Gen Solutions Mimi Forsythe and Lead Gen Measurement PM Lydia Azaret discuss “Beyond the form fill: Mastering lead quality in the AI era.”

Ms. Marvin previewed the show on LinkedIn:

“Mapping the full journey: Why tracking from the first online click to the offline sale is foundational for lead gen performance.

Data Manager: How to use this hub to connect your CRM and other offline data sources for consistent, high-frequency data uploads.

The ‘Sweet Spot’: Identifying conversion actions that balance volume with short enough delay to fuel Smart Bidding.

YouTube for Lead Gen: The role YouTube can play in your performance-scaling engine.”

Read LinkedIn. (March 25)

In the podcast, Ms. Azaret, Google’s lead gen measurement product manager, emphasized measurement setup and the importance of mapping out the customer’s complete journey so that AI can work its magic to maximum benefit:

“I think lead gen advertisers more than even other segments understand that their users’ touch points are more fragmented, more cross-device, more cross-channel than ever before. Mapping out the most crucial touch points that turn a casual browser, casual lead into that final purchaser and converter is foundational to understanding what touch points you actually want to bring in to Google to help drive our AI.

You hear this from us a lot: Google AI is only as good as the data that you feed it. Mapping out that customer journey helps you figure out what you ultimately want to measure and optimize towards in Google Ads. (…) We’re trying to ensure all advertisers harness the power of measuring the full customer journey that we know bridges the online and offline world.”

More:

See the podcast on YouTube.
Hear it on the Apple Podcasts app. (March 25)

From tipsheet: Google and other video platforms want you to see ‘video as a performance marketing opportunity.’

The push-pull of video is that — from a TV perspective — video is better-suited for “lean back” brand advertising.

AGENCIES

Opinions

“Most of the advertising industry is still treating #AI as a strategy question. That is the wrong question. What matters now is pricing…” (March 25) – Analyst Ian Whitaker on LinkedIn
“Agent inflation is the new technical debt: why marketing needs smarter AI, not more of it” by Shirli Zelcer, Chief Data & Technology Officer, dentsu on The Drum
“Going Crazy About Principle Media”- Jason Krebs on his Substack

People Inc.’s Jon Roberts on the AI licensing boom – and the revenue lag (March 25) – Digiday (subscription)
ZeroClick partners with Dappier to monetize conversational AI platform (March 25) – ZeroClick
Keynes Secures $40M Minority Investment from Volition Capital to Accelerate Growth in Connected TV – press release
Vaudit Achieves $30M Ad Spend Recovery Rate, Setting a New Standard for Cross-Platform Advertising Accountability (March 24) – Vaudit
InfoSum partners with Dstillery to enable privacy-safe and scalable audience intelligence solutions (March 25) – press release

Sephora loves ChatGPT – and vice versa

LLMs & CHATBOTS

Developments

PLATFORMS

Meta targets RMNs with AI in-app

RESEARCH

OpenAds CEO Liss on AI and ads

EVENTS

NewFronts: Yahoo and chatbot signals

EVENTS

ShopTalk threads: Gap, Google and Reddit

From ShopTalk:

LLMs & CHATBOTS

World models and Conversational DOOH

PLATFORMS

Google on lead quality in the AI era

AGENCIES

Opinions

MORE