Dynamic paywalls for humans and agents

In Adweek, MonetizationOS CEO James Henderson discussed the launch of his startup which aims to “personalize paywalls for both humans and AI agents.”

Adweek’s Mark Stenberg reported on the human angle:

“’For humans, this means not just tailoring when the paywall arrives, but also the product that it sells. A returning reader that engages exclusively with sports content might receive a subscription offer that offers unlimited access only to sports content for a fraction of the price of a standard subscription—same with someone who only reads a certain author or during a certain period of time. For every specific visitor, a specific experience.‘

There is a distribution demand curve for humans, where certain customers will be worth a lot of money, then a long tail of people who will not convert at the same rate,’ Henderson said. ‘The way you deal with that is by having a spectrum of products that match that demand curve.’”

Stenberg also notes that “The 24-person team has raised north of $6 million, and it counts backers including Google, Cloudflare, and Mather.”

Read the “bot angle” and more in Adweek. (February 19)

More: “MonetizationOS Launches Free Paywall Infrastructure to Fix Broken Economics of the Internet” (February 18) – press release

Related: Publishers explore selling AI visibility know-how to brands (February 19) – Digiday (subscription)

From tipsheet #1: MonetizationOS is another entrant in the race to help publishers monetize in the age of AI. Having Google and Cloudflare as investors is a fascinating “cocktail” given Google’s transition to AI search and Cloudflare’s defense of publishers in the face of AI bot crawling — including Googlebot.

From tipsheet #2: What a name!

LLMs & CHATBOTS

Developments

• “Gemini 3.1 Pro: A smarter model for your most complex tasks” (February 19) – Google’s The Keyword
• “OpenAI reportedly finalizing $100B deal at more than $850B valuation” (February 19) – TechCrunch
• “How Amazon Tracks Employee AI Usage—and Measures Results” (February 19) – The Information (subscription)

MARKETING

Marketing mix model scenarios — a tool

In a post on Google’s Ads & Commerce blog, Google engineer and data scientist Harikesh Nair explained how a new tool connected to his company’s open-source Meridian marketing mix modeling (MMM) framework can help marketers make real-world decisions emerging from the arcane world of marketing analytics.

Mr. Nair said:

“Scenario Planner is a user-friendly interface that allows marketers to experiment with different budget scenarios and see real-time ROI estimates — no coding required. It transforms the conversation from a look back at what happened to a collaborative plan for what’s next, regardless of technical expertise.”

More here. (February 19)

With incrementality increasingly used to understand the causal impact of marketing spend (especially in AI-enabled advertising environments), tipsheet asked how “bottom up” incrementality measurement intersects with the “top down” Scenario Planner.

A Google spokesperson responded:

“Re: incrementality, a key innovation Meridian brings is calibrating the MMM with results from real-world incrementality experiments, or ‘priors.’ This is important because it grounds the model in proven, cause-and-effect data rather than just historical correlation, delivering a much higher degree of accuracy.”

More: “Modern measurement needs to move from just reporting on outcomes to driving more of them. That’s why we’re launching the Meridian Scenario Planner.” (February 19) – Dan Taylor, VP of Global Ads, Google on LinkedIn

Related: “[See graphic] Agencies do the hard part of bridging solutions/data/activation/measurement across platforms and channels, we are moving upstream!…..also you cannot have a single ID across this all. (…) No one AI system can stitch these [silos] together yet.” (February 12) – Alan Morrissey, Data Strategy Partner, WPP Media’s Choreograph on LinkedIn

LLMs & CHATBOTS

Optimizing for Amazon’s Rufus chatbot

There’s plenty of talk these days about generative engine optimization (GEO), but what about the “shopping” subset of LLMs — can you optimize for the shopping chatbot?

Touting his company’s latest feature, Ross Caveille of Acorn Intelligence, a UK-based ecommerce tech and services firm, said his company’s agentic architecture is taking a step in the direction of unlocking the shopping bot by predicting how Rufus will respond.

He explained on LinkedIn yesterday:

“Our Texana Agent Framework allows brands to proactively discover how shopping agents respond to queries such as ‘what is the best smart TV’ or ‘what features should I look for in a smart TV’ …. and, now we can better understand Amazon Rufus.

Below is our Agent response to the second question above and right is Rufus’ response … seems a match made in shopper heaven. (…)

Super-useful as brands now need to understand how to optimise and work within the environment of people and agents as consumers.”

See the agent’s response — and read more from Mr. Caveille — on LinkedIn. (February 19)

From tipsheet: How much a product’s marketer can influence the shopping chatbot’s (Rufus) citations within the walled garden of the retailer — Amazon, in this case — appears to be a different GEO “game.”

The LLMs powering OpenAI’s ChatGPT or Google’s Gemini, for example, are informed by the entire web.

AGENCIES

Omnicom’s ‘data-led AI transformation’

Agency holding company Omnicom released its first earnings report since merging with IPG in November.

Revenue came in at $5.5 billion for Q4 2025 with a net loss of $0.9 billion. The new combined entity’s earnings missed — but revenues exceeded — expectations, according to Investing[dot]com.

Omnicom’s stock was up over 15% yesterday (a $25 billion market cap) as investors rejoiced over the company doubling cost-savings projections.

AdExchanger’s Allison Schiff covered the earnings call with Wall Street where the effects of the merger and the company’s AI-enabled “Omni” platform were front-and-center:

“Executives framed the deal as a ‘data-led AI transformation,’ highlighting plans to use AI for targeting, measurement and creative testing through Omni, Omnicom’s centralized identity and analytics service.

Omni is now integrated with Acxiom’s Real ID, Flywheel’s Commerce Cloud and Omnicom’s first‑party data, which strengthens Omnicom’s overall position in data, identity and AI, the holdco’s CEO and Chairman John Wren told investors.”

Read more in AdExchanger. (February 19)

Omnicom unveiled the latest “Omni” update at CES in January.

Omnicom CTO Paulo Yuvienco, who was also on the earnings call, said AI wasn’t replacing headcount at his newly-combined holding company: “I think the bulk of the labor-related synergies really relates to a number of things. AI is not necessarily the primary driver of how we looked at this.”

Nevertheless, Yuvienco later discussed how AI is helping his agency’s creative teams:

“Look, I know the narrative is always about do we do the same with less, but the reality is that what AI and generative AI is allowing us to do is to do more than we have ever been able to do. And more importantly, it is allowing us to do things that we have not been able to do in the past.

So just to give you some specifics around this, historically, creative teams would typically put two to three different concepts in front of our clients for a specific campaign. The reason is because it takes time. It takes a lot of effort to actually bring those concepts to life.

Today, with the use of the tools that [CEO John Wren] is talking about, with the agentic capabilities that we put in place, our teams can now test 20 concepts, can test 50 concepts. More importantly is that they can test them synthetically so that we can understand what the impact and value of that work could be.”

More:

• Omnicom Reports Fourth Quarter and Full Year 2025 Results (February 18) – Omnicom
• Omnicom (OMC) Q4 2025 Earnings Call Transcript (February 18) – Motley Fool
• Analyst Ian Whitaker breaks down Omnicom’s results (February 19) – Ian Whitaker on LinkedIn

Related: “Ex-WPP Chief Martin Sorrell warns agency consolidation reflects structural bloat” (February 18) – PRmoment India

TECH

AppLovin and the “Meta playbook”

A job listing at mobile ad technology company AppLovin is drawing attention as the company appears to be interested in building a “next-generation social platform.”

The new role is for a “Software Engineer” in Singapore and reads, in part:

“At AppLovin, we’re pioneering the future of social connectivity by building a next-generation platform that redefines how people create, share and engage with content. This new venture represents a unique opportunity to architect the foundational systems that will power immersive social experiences for millions of users. Combining the innovation of a startup with the scale of an industry leader, we’re assembling an exceptional technical team to bring this vision to life from the ground up.

A Day in the Life

As a founding Backend Engineer for this venture, you will architect the digital backbone of our next-generation social platform. (…) You won’t just write code – you’ll make critical architectural decisions that form the foundation of our product, working directly with product managers and AI engineers to transform ambitious ideas into technical reality.”

See the listing.

RBC Capital Markets analyst Matthew Swanson read the tea leaves and saw an expansion of AppLovin’s “Meta playbook” (but in reverse… monetization first then social). He said in a note to investors:

“At a high level, this becomes more compelling as we feel management continues to seek ways to expand their competitive moat, particularly with rising investor questions around competition and AI. Overall, we think this would eventually be another step towards creating a holistic end-to-end platform more similar to walled-garden peers such as Meta, TikTok and Snap.”

More in Seeking Alpha. (February 19)

LLMs & CHATBOTS

“AI Mode” adds 53 languages

According to Nick Fox, Google’s SVP of Knowledge and Information, “AI Mode” in Google Search has expanded.

Urging people to give it a try, Mr. Fox said on X (February 18), “Shipping AI Mode to 53 new languages (spoken by more than a billion people globally!).”

Read: “Get AI-powered responses with AI Mode in Google Search” – Google

From tipsheet: That’s a lot of new global audience — and data — for the AI machine.

PLATFORMS

Morgan Stanley: ‘Unlock’ coming for ads

Morgan Stanley analysts gathered for a discussion for investors this past week and discussed the prospects for the larger players in digital advertising given “AI disruption.”

Apparently, disruption isn’t a bad thing as Morgan Stanley analyst Brian Nowak was clearly optimistic:

“I think it’s very encouraging with what we’re seeing with GPU-enabled machine learning at AppLovin and Meta. I think for Meta the next unlock — and maybe it comes with AppLovin eventually as well — is you’re going to have more large language model (LLM) data analytics of really analyzing what people are doing on a real-time basis to even get more context.

… When people don’t click on an ad… when they don’t stop… when they do stop, when they scroll slower, scroll faster… how does that vary by daypart?

I feel like it’s still much earlier than appreciated in the digital advertising innovation we’re going to see. Meta and Alphabet are both leading that. But I think AppLovin is one of these that is showing the benefits of GPU-enabled machine learning.”

MEASUREMENT

Use case: Tracking outcomes for pharma

In a Yahoo blog post yesterday, Kurt Anderson, Director, Business Strategy at Yahoo DSP, discussed his demand-side platform’s (DSP) outcome-focused solution in the context of programmatic media partner, MiQ.

Mr. Anderson outlined the use case:

“MiQ led a campaign for a pharmaceutical advertiser, targeting audiences with a specific medical diagnosis and driving prescription growth. MiQ activated Yahoo In-Flight Outcomes alongside IQVIA Digital‘s healthcare data and measurement solutions to efficiently drive optimizations.

By accessing in-flight signals such as audience quality scores, office visits, and new-to-brand prescription (NBRx) conversions, the campaign moved beyond static targeting, continuously optimizing for quality and downstream impact to deliver strong results.

The results…”

Driving outcome discovery is Yahoo DSP’s AI-enabled Blueprint optimization engine.

Read: “Yahoo In-Flight Outcomes: Turning Measurement Into Real-Time Optimization” (February 19) – Yahoo

From tipsheet: “Outcomes” are the name(s) of the game these days.

LLMs & CHATBOTS

ChatGPT ads in the wild

Seizing the moment, Adthena CMO Ashley Fletcher announced on Wednesday that his team had seen ads in OpenAI’s ChatGPT for “signed-in desktop users in the US.”

He provided details on LinkedIn including:

Key Takeaway: Contrary to early predictions that ads would only appear after a back-and-forth dialogue, these triggered immediately on the first prompt response.

When asked, ‘What’s the best way to book a weekend away?’, the results delivered instant sponsored placements. Notably, the UI features a prominent brand favicon and a ‘Sponsored’ label, slightly different to OpenAI’s original concepts.”

Read more and see a screenshot. (February 18)

More: Best Buy, Expedia, Enterprise Mobility Are Among the First Brands Spotted Running Ads on ChatGPT (February 19) – Adweek

From tipsheet: Yup. Looks like shopping ads to me. This is only the beginning — a pilot.

FINANCIALS

Earnings reports next week

• The Trade Desk – Wednesday, February 25
• Magnite – Wednesday, February 25
• PubMatic – Thursday, February 26
• WPP – Thursday, February 26

MORE

• Walmart’s Ad Revenue Totaled $6.4 Billion In 2025 As The Ecommerce Flywheel Started To Spin (February 19) – AdExchanger
• Job opening: “Monetization Product Marketing Manager, Advertising” – OpenAI
• AIMSOO launches to provide SMEs with free visibility in AI Search (February 19) – AIMSOO blog
• “Distribution and the human attention premium” (February 19) – Analyst Eric Seufert on Mobile Dev Memo
• JWX Launches Vertical Video Product To Help Publishers Increase Engagement and Time-on-site (February 19) – press release