Perplexity’s decision last week to forego advertising in its chatbot — while criticizing its much larger competitor, OpenAI, for introducing ads to ChatGPT — is bringing more scrutiny.
Based on website traffic data aggregated by Wired, Perplexity appears to have “reach issues” which have undermined progress with ads.
Wired’s Maxwell Zeff reported:
“Data from the third-party analytics firm Similarweb suggests Perplexity had just over 60 million monthly active users across its website and mobile app in January. That’s more than double the users Perplexity had last year, according to Similarweb. People also now access Perplexity via its AI-powered browser, Comet, which Similarweb doesn’t track.
A source close to Perplexity says the agent in its Comet browser reached 2.8 million weekly active users (which were also Perplexity subscribers) in December 2025, down from a peak of 7.8 million WAUs earlier in the year.
Without accounting for Comet, Perplexity’s user base on web and mobile is less than 10 percent of OpenAI’s ChatGPT and Google’s Gemini, which have 800 million weekly active users and 750 million monthly active users, respectively.”
Unnamed Perplexity executives admitted in a press conference last week that its chatbot “isn’t for everyone.” Instead, the company execs said they hope to ride chatbot subscription revenue across consumers, developers and the enterprise (by differentiating with a higher level of accuracy) and position the company as a B2B orchestration layer for AI models, according to Wired.
Read: “Perplexity’s Retreat From Ads Signals a Bigger Strategic Shift” (February 19) – Wired (subscription)
Related: “Google’s best products (Search, YouTube, Maps, Chrome, Android) are all direct funnels for advertising revenue. Ads fund products and create a virtuous cycle that makes products better, all while enabling free access to the products so everyone (not just those who can afford a subscription) can enjoy them. This is why [TBPN co-host] Jordi is bearish on Perplexity dropping ads, and why I think consumer AI apps need to figure out ad monetization. Otherwise, your product might become another Google Reader.” (February 19) – John Coogan, TBPN on LinkedIn
LLMs & CHATBOTS
Developments
- Google Is Exploring Ways to Use Its Financial Might to Take On Nvidia (February 20) – The Wall Street Journal (subscription)
- Measuring AI agent autonomy in practice (February 18) – Anthropic
- Apple’s Next Big Thing Is A Push Into Visual Artificial Intelligence (February 22) – Bloomberg (subscription)
BRANDS
Meme: Meta’s Manus and the media buyer
“I’ve come to realize that Meta’s Manus acquisition marks the beginning of the end of the media buyer as we’ve known it for the past decade and a half….Media buyers will be doing work that really has nothing to do with buying media and will fully evolve to become full stack growth strategists.
In some ways it’s been a slow burn with more automation than ever in the last 36 months, but I truly believe this is the inflection where we see an exponential removal of the human touch on traditional media buying work.
This is a good thing, humans should be doing higher leverage tasks than toggling switches and budgets all day.”
Peter Quadrel is a growth marketer at Odylic Media, a D2C performance marketing agency he founded.
“We wanted to test the new Manus AI integration (Zuckerberg’s media buyer killer), and if you’re planning or buying social, it’s worth spending time with it ASAP.
It plugs straight into your live Ad Manager account and gives access to everything. Campaigns. Audiences. Creative. Structure. It was fast. Very fast.
Nothing it suggested was dramatically better than what a good operator would already do. But that’s not the point. (…)
Tools like this don’t replace the role of media. They shift where the value sits.”
Luke Maher is founder of an Australian performance marketing agency called Superhuman.
From tipsheet: The “end of the media buyer” meme is picking up speed.
With the Manus acquisition in December, it is increasingly clear that Meta will be aiming at advertising workflow with the new tech. SMBs (like the small performance marketing shops represented in these quotes) appear to be buying what Manus is selling.
For Meta, its AI is enabling “the little guy”, which in turn leads to more buying on Meta platforms and through its automated, AI-enabled buying platform Advantage+.
LLMs & CHATBOTS
Funding the chatbot app experience
Startups and the early evolution of creative formats in AI chatbots was explored by AdExchanger’s Joanna Gerber on Friday.
The product team supporting an AI assistant known as Luzia — popular in Latin America — found that using Google Ads weren’t enough. Luzia’s head of product, Natalia Solano Gutierrez, said her company turned to SF-based startup Koah, which positions its product as “seamless, context-aware ads that fund your app without breaking the experience.”
Koah Co-Founder Mike Choi explained how his startup’s product works to AdExchanger:
“Instead of just advertising contextually based on a specific query, he said, Koah targets users with ads based on personal data, like previous chat conversations within the platform and the user’s location.
For instance, said Choi, if a student is using Luzia for help with math homework, it might seem obvious to serve them an ad for Quizlet, or a math tutoring program. But a student is going to be solving math problems ‘all year long,’ said Choi. After encountering the same ad dozens of times, he said, the user will say, ‘I’ve seen enough of that.’”
Koah CEO Nic Baird added that “native, text-based ads that are ‘clearly demarcated’…” appear to be the best format.
Read more in AdExchanger. (February 20)
Koah’s $5 million seed round announced in September was led by Forerunner and included additional investment from South Park Commons (where Baird worked previously) and AppLovin co-founder Andrew Karam.
Read: Koah raises $5M to bring ads into AI apps (September 7) – TechCrunch
More: Spanish AI company Luzia closes a new $13.5 million funding round led by Prosus Ventures (May 2025) – Luzia blog
From tipsheet: It’s notable that AppLovin’s Karam is an investor.
AppLovin’s future influence on — and participation in — chatbot advertising seems inevitable given the company’s mobile in-app advertising leadership. The chatbot is an app, after all. Rumors are already flying (see last week on X).
With a seat at Koah’s investor table, perhaps AppLovin’s acquisition of the startup is a possibility at some point — unless AppLovin builds its own “ads in a chatbot” solution first.
BRANDS
RampUp conference next week
Data collaboration platform LiveRamp’s annual RampUp conference takes place next week, March 3-5, in San Francisco.
AI’s impact on the advertising industry and the opportunity ahead is the headline for nearly all of the agenda.
But, the subtext is LiveRamp’s future and several sessions which include LiveRamp’s CEO Scott Howe may shed more light:
- AI: Howe and product leader Matt Karasick appear on stage with a “vision” keynote on March 4 titled “History’s Guide to Winning (in the AI Era)”.
- CTV: Mr. Howe appears again later the same day with executive Nicolle Pangis of Netflix in a discussion titled, “Netflix, AI, and the Reinvention of CTV Advertising.”
- Programmatic: The Trade Desk’s Samantha Jacobson and consultant Paul Roetzer join Howe for “The Collaborative Advantage: How AI Lights Up the Network”
- Meta: Meta VP Simon Whitcombe joins Mr. Howe for “A Look Inside Meta’s AI Playbook.”
From tipsheet: A key longer-term bet for the company which Howe first signaled last June — and perhaps part of the “vision” keynote on March 4 — may be the success of the “Agentic Audiences” protocol (was “User Context Protocol”) now under the stewardship of IAB Tech Lab.
Themes such as “open web versus walled gardens”, agentic protocol development and the timeline for adoption all intersect in “Agentic Audiences” and understanding the AI opportunity ahead for LiveRamp.
PLATFORMS
Feeding the AI ad machine, Kokai
There continues to be controversy over The Trade Desk’s OpenPath (OP) supply framework which ultimately provides signal to the artificial intelligence that underpins The Trade Desk’s demand-side platform known as Kokai. Better signal means better performance — that’s part of the pitch with OpenPath — which is critically important to The Trade Desk (TTD) as it competes against the walled gardens of Google, Meta and Amazon.
On Friday, Adweek reported that, according to sources, Dentsu and WPP would no longer participate in OpenPath and were “wary of what they say is a lack of transparency in where exactly their ads are running alongside what they describe as hidden fees on the platform.” Read more. (February 20 – subscription)
On Saturday on X, Bayer paid media executive Ryan Verklin weighed in with his support for OpenPath (OP). Thereafter, a thoughtful debate ensued in the comments with contributions from ChaliceAI’s Adam Heimlich and former TTD engineer Jud Spencer:
Mr. Verklin began:
“Lots of discussion about OpenPath. I am pro OP. My thoughts below.
Someone tell me where I’m wrong. OpenPath has value to both the buy and sell side.
It has value to the buyer because TTD claims (and I’ll take their word for it) they run OP at cost. They claim their ~5% take rate is just to run it at break even. TTD is making all their money on DSP fees (different discussion) where the other SSPs have higher yields and dynamic pricing so they can make a profit which OP doesn’t need to.
The pubs don’t have the same margin rate across all SSPs (it’s dynamic and essentially the SSPs are trying to take the highest margin while also trying to win the auction in the Pub’s adserver/header). Hypothetically OP would have the lowest margin (because they are running it at break even) but the bid would still have to be the highest to win the auction in the pub’s adserver.
The one benefit for pubs using OP should be that they get more volume from TTD they wouldn’t otherwise get if they didn’t use OP…”
Read the entire thread on X. (February 21)
Meanwhile, in a separate thread on the Adweek article on LinkedIn, Nobuyuki (Nobu) Onodera of Dentsu Digital said:
“What this situation evokes for many is a broader concern: the gradual ‘walled-gardening’ of the open web, renewed arbitrage dynamics, and potential disruption of the existing supply-chain balance…”
Read his comment on LinkedIn. (February 20)
MEASUREMENT
Using LLMs and recommending ads
Over the weekend, analyst Eric Seufert discovered a new white paper released by Meta engineers earlier this month titled, “Generative Reasoning Re-ranker”.
He said on LinkedIn:
“Fascinating new paper from Meta on applying LLMs to recommendation systems. The domain discussed in the paper is content re-ranking, but I don’t see why this couldn’t be applied to ads. Re-ranking takes a ranked list of candidate items (following retrieval and, sometimes, pre-ranking) and updates the ordering to better optimize some objective function (eg., purchase).”
Read more from Seufert on LinkedIn. (February 21)
The white paper’s excerpt begins:
“Recent studies increasingly explore Large Language Models (LLMs) as a new paradigm for recommendation systems due to their scalability and world knowledge. However, existing work has three key limitations: (1) most efforts focus on retrieval and ranking, while the reranking phase, critical for refining final recommendations, is largely overlooked; (2) LLMs are typically used in zero-shot or supervised fine-tuning settings, leaving their reasoning abilities, especially those enhanced through reinforcement learning (RL) and high-quality reasoning data, underexploited; (3) items are commonly represented by non-semantic IDs, creating major scalability challenges in industrial systems with billions of identifiers.”
Enter the Generative Reasoning Reranker (GR2).
See it on Cornell U’s arXiv. (Originally published February 8)
From tipsheet: There are 21 authors of this paper — likely a substantial portion of Meta’s AI and ads talent. For all of Wall Street’s consternation about what Meta is using all of its capex for, Meta is focused on the future of advertising at the very least. And that’s very big.
CAREERS
Ads are shaping finance at OpenAI
OpenAI continues to hire in support of its ads strategy.
Two of the latest roles listed in the Careers section of OpenAI’s website are in the company’s Finance group.
When prompted with links to the job descriptions, OpenAI’s ChatGPT offered a quick summary on the two roles:
- The Senior Manager, Ads Revenue is a higher rank with wider influence, not just limited to accounting execution but also systems, controls governance, and strategic shaping of Ads monetization.
- The Ads Revenue Accounting Lead is a critical execution role that ensures the heart of Ads revenue accounting is accurate, compliant, and reliable.
Prior to the finance roles, the company has also listed ads-related roles for “Software Engineer, Ads Integrity” and “Monetization Product Marketing Manager, Advertising.”
Also notable is a job listing for “Head of Data Science and Machine Learning, Global Forecasting”. This role would appear to have ads strategy implications but there is no explicit mention of advertising.
MORE
- Pitch deck: Why Amazon believes its premium streaming inventory is worth the money (February 20) – Digiday (subscription)
- Operative and GraySwan Partner on AI-based Data Visualization for Immediate, Easy-to-Use Business Intelligence (February 17) – press release
- The creator economy’s ad revenue problem and India’s AI ambitions (February 20) – TechCrunch
- LiveRamp Partners With Scowtt to Unlock AI-powered Optimization Using Predictive First-party Data Signals (February 20) – press release
- Why are brands no longer afraid of producing AI slop? (February 21) – Creative Bloq