Podcast
AdTech AdTalk
Hosted by Adam and Gareth · EN
Buy Side Meets Sell Side.
Live Q&A with CEOs telling the truth about where Ad Tech is headed.
15episodes
Episodes
Newest firstAll episodes
I'll Take Any Test You're Willing To Take
2w ago01:01:45Tap to summarizeAdam and Gareth reunite to unpack an explosive Q2 in ad tech, kicking things off with the "triple hockey stick" growth driving Meta’s 33% revenue surge. The guys discuss how Meta’s ruthless AI shift and integration of massive server compute is crushing Google in the enterprise ad market, and why Google's strategy of over-servicing holding companies is backfiring.They also dive into the backlash against Trade Desk CEO Jeff Green and his "open internet" messaging, debating why retail media networks (RMNs) are growing so fast despite the inherent conflict of interest between retailers and CPG brands. Plus, Gareth explores the terrifying new "split brain" architecture between DSPs and SSPs, Adam makes the case for why "barter" and "trade" agencies still exist to liquidate bad media buys, and they examine how Swivel's new agentic tools are finally bringing automated, human-like operations to Connected TV (CTV) yield management.Key Topics:Meta's AI Dominance: Why Meta’s massive AI efficiency is crushing Google in the enterprise ad market, despite Google's bloated service teams.The Trade Desk Backlash: Unpacking the industry's frustration with Jeff Green's "open internet" pitch and why the growth numbers don't match the story.Retail Media Networks (RMNs): Why companies like Kroger and Albertsons won't share their data, and how it hurts CPG advertisers.The End of "Getting Lost": A philosophical take on how AI tools are stripping younger generations of resilience and problem-solving skills.Agentic Ad Ops (Swivel): How new AI agents are outperforming human traders in the CTV market by running 100 manual interactions per hour.00:00 Intro: Hangovers & Detroit Style Pizza04:55 Defending Jeff Green’s "Open Internet" Pitch11:15 Q1 Earnings: Meta Crushes the Competition13:12 The Conflict Within Retail Media Networks (RMNs)20:49 Why Amazon Built Ad Tech from Scratch22:50 The Risk of a Google Display Network Collapse25:27 Swivel's Agentic CTV Case Study34:04 The Problem with Barter and Trade Agencies41:19 The "Split Brain" Architecture of DSPs vs. SSPs48:39 Postmortem: The IAB Tech Lab's New DTE Standard57:30 Cash Patel and The Ad Tech Investigations01:00:23 Closing Thoughts: Why We Need ACRONYM
Transcribe →The Grand Bargain (ft. Matt Sattel, OpenX)
4w ago01:01:30Tap to summarizeAdam and Gareth return to the virtual studio to debate the definition of Supply Path Optimization (SPO) and why the sell-side and buy-side can never seem to agree on what it actually means. They are joined by OpenX CEO Matt Sattel to discuss his transition from the agency world to the SSP side, and why OpenX made the expensive, grueling leap to cloud infrastructure back in 2019 to prepare for the AI era.The trio unpacks the industry's so-called "Grand Bargain," debating whether a multilateral committee can truly fix the adversarial relationship between DSPs and SSPs. Gareth argues that ID Bridging is secretly a brilliant conversion-optimization engine in disguise, while Adam pitches his "Premium Coffee" theory to explain how publishers can finally escape the MFA (Made For Advertising) race to the bottom. Plus, a look at why 50% of programmatic ad requests are filtered out, and what OpenX really means by their new "Intelligent SSP" rebrand.00:00 Intro: The "Bright Lights" of Live Shows03:03 The Arc of the Sophisticated CMO06:56 Introducing Matt Sattel (CEO, OpenX)10:04 The Two Definitions of SPO13:16 The Premium Coffee Theory (Escaping MFA)14:52 Filtering Out 50 Billion Requests a Day16:41 The "Grand Bargain" Between Buyers and Sellers19:39 Gareth's Take: ID Bridging is Actually Optimization24:14 The True Cost of Eliminating MFA27:28 Google's Role in the MFA Epidemic32:41 The "Lego Baseplate" Strategy38:42 Why OpenX Moved to the Cloud in 201944:07 Direct Sales & Agentic Buying (ADCP)51:30 Differentiated Demand for Publishers57:40 The Intelligent SSP Rebrand
Transcribe →Claude the Gaude (ft. Adam Markey)
May 101:03:49Tap to summarizeIt's an all-Adam episode this week! With Gareth out, Adam Heimlich is joined by stealth startup founder and product leader Adam Markey to discuss the reality of building an ad tech company from scratch in the AI era.The guys dive deep into the philosophy of AI adoption, contrasting Anthropic's enterprise-focused "Constitutional AI" against OpenAI's consumer-driven, entertainment-style models. Markey shares how using Claude as a "full-stack" co-founder has drastically reduced product development cycles from six months to just three days. Plus, they explore why treating AI as a "replacement human" is the wrong starting point, how to avoid AI-generated "slop" in both creative and media planning, and why Anthropic's $20 billion valuation might actually be justified if they can maintain their compounding context moat.Finally, they touch on the IAB’s Agentic Realtime Framework (ARTF), the return of "frenemy" collaboration in ad tech, and why Meta's insane 33% revenue growth is tied directly to their massive investments in server compute.00:00 Intro: Welcome to Planet Ad Tech03:45 Bootstrapping a Startup in the AI Era06:40 Using AI for Complex Rule Prioritization08:13 Anthropic vs. OpenAI: Enterprise vs. Consumer12:03 Claude’s "Compounding Context" Moat14:28 Building a RevOps Dashboard in 2 Hours16:45 Change Management and Full-Stack Founders19:59 AI as a Force Multiplier, Not a Replacement25:35 The Value of Markdown Files for Context28:06 AI as the Ultimate Information Condenser32:28 Anthropic’s Path to Profitability37:25 Working with Google’s "Stitch" Design Agent39:07 The Edge of AI: Humor, Differentiation, and "Slop"44:09 Agentic Standards (ADCP & ARTF)51:03 Advice for Product Managers at Legacy DSPs53:50 The Enterprise Rush to Adopt AI59:49 Meta's 33% Growth and the Server Compute Race
Transcribe →Triple Hockey Stick
Apr 2401:00:35Tap to summarizeAdam and Gareth dive into a chaotic start to Q2 in ad tech, leading off with Meta's jaw-dropping "triple hockey stick" growth in revenue per employee. They discuss how Meta’s ruthless AI shift—including tracking employee keystrokes for training data—is driving massive margins, even as hands-on buyers complain about constant bugs and platform outages.The guys also cover the rise of "Index Cloud" and Bedrock (a DSP living natively inside an SSP), explaining why the traditional DSP/SSP divide is collapsing and how this new architecture enables ultra-low-latency bidding for live sports. Plus, Gareth goes on a rant about the rampant QPS duplication problem in programmatic, they debate the return of Vickrey (Victory) auctions, and they react to a spicy $18M legal battle between TV measurement giants EDO and iSpot. Finally, Adam shares his frustration with the IAB's new Programmatic Governance Council and pitches his own "ACRONYM" coalition to guarantee 80% working media for advertisers.00:00 Intro: An Explosive Start to Q201:36 We Are in the Third Inning of AI in Ad Tech04:23 Meta's "Triple Hockey Stick" Revenue per Head05:15 Tracking Employee Keystrokes for AI Training09:08 The Broken Meta Ad Experience & Bugs16:19 Meta's Internal AI Models (Andromeda & Muse Spark)20:14 Can the Open Web Compete with Walled Gardens?30:37 News: The Trade Desk Launches AI Agents31:39 Index Cloud & Bedrock (A DSP Inside an SSP)37:59 Data Scientists vs. DSP Engineers43:21 The 3 Million QPS Myth & Bid Duplication45:31 Vickrey Auctions and Bid Shading48:22 EDO vs. iSpot: The $18M Lawsuit & Spicy Quotes50:45 The IAB's New Programmatic Governance Council51:54 ACRONYM & The Push for 80% Working Media57:56 The Lutnick / Epstein Ad Tech Connection
Transcribe →Dynamic Bulldinky
Apr 1701:08:10Tap to summarizeAdam and Gareth reunite in the virtual studio after a week apart, and they have a lot of philosophical ground to cover. Gareth shares a bizarre realization about startup growth—comparing a mature company to a "disembodied brain"—and how AI is acting as the ultimate force multiplier for lean teams. They debate the value of the "founder struggle," questioning if the current generation of entrepreneurs will lose resilience now that AI can write complex code and solve technical crises in 30 seconds.On the industry front, they tackle the latest IAB Tech Lab standard (DTE) and whether "traffic shaping" is just a polite word for throttling. Gareth explains why the "split brain" architecture of DSPs and SSPs makes optimal decisioning impossible, and why the Agentic Realtime Framework (ARTF) is the only real fix. Plus, a look at Allbirds pivoting from sneakers to AI data centers, the rise of AI-generated "slop" on LinkedIn, and why Salesforce's new Agentforce might be a desperate defense against Claude disintermediating their entire business model.Key Topics:The Disembodied Brain: How a startup evolves as you hire, and what it means when AI becomes your hands and feet.The Death of Getting Lost: Does the convenience of AI tools strip founders of the resilience needed to survive crises?Traffic Shaping vs. Throttling: Breaking down the IAB's new DTE standard and the architectural impossibility of the "split brain."Allbirds Pivots to AI: Why the market rewarded a failing DTC sneaker company for becoming a GPU rental service.Salesforce vs. Claude: Is Salesforce's Agentforce a true innovation, or a defensive moat against demand aggregation?00:00 Intro: Adam Returns from Vacation01:05 Startup Burn Rates & The "Disembodied Brain"03:08 AI as a Force Multiplier04:24 The Value of the Founder Struggle06:22 The 90s: When You Could Actually Get "Stood Up"11:32 AI "Slop" and the Premium on Human Content16:49 Can AI Replace the Enterprise Sales Playbook?19:51 The 15 Commitments of Conscious Leadership24:39 The OpenAI "Molotov" Backlash30:12 What Knowledge is Playbookable?35:54 News: Allbirds Pivots from Sneakers to AI39:31 IAB Tech Lab's New DTE Standard41:19 The "Split Brain" Problem in Ad Tech48:39 Postmortem: ADCP vs. ARTF (The IAB Standard Wins)01:01:46 Salesforce's Agentforce vs. Claude
Transcribe →The GAMpire Strikes Back
Apr 1000:35:15Tap to summarizeWith Adam out this week, Gareth flies solo to drop four massive predictions about the next five years of ad tech. Spoiler alert: The current system is unsustainable.Gareth argues that the traditional DSP/SSP paradigm is dying , crushed under the weight of Deal IDs and conflicting decisioning engines. He explains why the future of ad tech architecture lies in open, modular bidding frameworks (like ARTF) where custom logic can be executed natively without "middleman" friction.But the spiciest prediction? Google Ad Manager (GAM) is about to have a massive resurgence. Gareth theorizes that if ADCP (Automated Direct Sales) succeeds, GAM will emerge as the ultimate direct sales engine, bypassing DSPs entirely for enterprise buyers who want to set up cross-publisher campaigns with zero tech tax. Plus, he tackles the inevitable rise of "orchestration layers," why the mid-market is about to see a DSP renaissance , and why the shift away from ID-based buying will finally force publishers to reduce ad density and increase quality.Key Topics:The Death of DSPs & SSPs: Why splitting campaign logic between two platforms using Deal IDs is no longer architecturally viable.The Return of GAM: How ADCP could turn Google Ad Manager into a direct competitor to traditional DSPs.The Orchestration Era: Why modern bidding systems will look less like DSPs and more like open hubs for modular valuation and data vendors.The End of Ad Density: Why the shift to predictive modeling (and away from IDs) will finally reward publishers for running fewer, higher-quality ads.The LiveRamp EU Struggle: Why strict European privacy laws (GDPR) are making PII-based identity graphs incredibly difficult to operate abroad.00:00 Intro: Gareth Flying Solo01:57 Prediction 1: The End of the DSP/SSP Paradigm04:08 The Architectural Flaw of Deal IDs08:34 Prediction 2: The GAMpire Strikes Back (Google Ad Manager)10:14 How ADCP Turns GAM into a DSP Competitor12:27 Prediction 3: A Proliferation of Mid-Market DSPs15:01 The Approaching SSP Bloodbath18:22 The Rise of the "Orchestration Layer"22:50 Prediction 4: Publisher CPMs Will Go Up24:55 Why User IDs and Cookie Syncing are Commodities27:09 The Frequency Capping Problem with ADCP30:14 The LiveRamp EU/UK Privacy Struggle33:49 The Bull Case for Fingerprinting & Synthetic Audiences34:21 Outro
Transcribe →Hunting a White Whale
Apr 300:52:04Tap to summarizeAdam and Gareth are joined in the virtual studio by Laura Manning, SVP of Data and Measurement at SINT. They break down the fundamentals of Brand Lift 101, explaining how SINT acts as the infrastructure matching survey-takers across the internet with relevant ad exposure data. Gareth details exactly why traditional attribution models fail massive enterprise brands like Procter & Gamble. The trio also discusses the agonizing eight-year waiting list to get into Google's Ads Data Hub (ADH), and why allowing walled gardens to report their own brand lift metrics is essentially letting them grade their own homework. Finally, Adam explains why the shift from linear broadcast to digital resulted in an "epidemic of waste" due to algorithms driving frequency instead of true reach.Key Topics:- What is SINT: Acting as a DSP for online research that matches respondents with surveys.- Attribution vs. Brand Lift: Why enterprise CPG brands don't rely on clicks and e-commerce tracking.- The "O" Word: Using survey response data via API to actually optimize campaigns in real-time.- The White Whale: SINT's years-long quest to integrate with Google's Ads Data Hub.- The Reach Problem: Why digital algorithms inherently favor frequency over reach, unlike traditional linear TV.00:00 Intro00:57 Introducing Laura Manning & SINT 06:40 Attribution vs. Brand Lift 08:28 How SINT Measures Brand Lift 12:10 Standard Survey Questions 14:12 The "O" Word (Optimization) 25:16 Google's Ads Data Hub 28:50 Linear TV vs. Digital Reach 39:10 Post-Mortem: Is Brand Lift a "Cope"?
Transcribe →Gold in a Mine
Mar 2701:03:25Tap to summarizeGareth powers through a blown-out back to join Adam for a loaded episode exploring the true value of data and the hidden mechanics of media buying. They kick things off by breaking down Gamera's new viewability segments in the Index Exchange marketplace, exposing how antiquated, manual ad slot labeling in Google Ad Manager is suppressing publisher yield, and how AI is fixing it.The duo dives deep into the philosophy that raw signal is just "gold in a mine" without the right prediction models to extract its value. They also unpack the ANA's latest study on Principal Media buying, shedding light on the clandestine world of agency secondary liquidation and "barter" agencies. Plus, a look at Microsoft’s baffling apathy toward its own ad tech stack, the dirty Facebook arbitrage secret that fueled Made For Advertising (MFA) sites, and the landmark negligence ruling against Meta for social media addiction.Key Topics:Automating Viewability: Why manual ad ops labeling is costing publishers money, and how Gamera's on-page AI creates predictive viewability segments.The "Gold in a Mine" Philosophy: Why data is commoditized and the real competitive moat is modeling.Microsoft's Missed Opportunity: Why a tech giant with Azure, LinkedIn, Xander, and OpenAI is ignoring the ad tech wars.Principal Media & Barter Agencies: Breaking down how holding companies liquidate bulk media buys (and occasionally trade ads for pizza sauce).The MFA Arbitrage Secret: How MFA publishers used Facebook remarketing to poach high-value programmatic users.Social Media Liability: Meta's landmark loss in a social media addiction case and the rise of "digital citizenship" in elementary schools.00:00 Intro & Gareth's Blown Back02:44 Gamera Viewability Segments in Index04:49 The Manual Nightmare of Google Ad Manager10:32 Data is Gold in a Mine14:51 Google's MMM & The Enterprise Fight18:18 The "Affinity for Yogurt" Trap20:41 How Big Tech Captured All Marginal Value24:15 Microsoft's Baffling Ad Tech Apathy33:48 ANA Study: Demystifying Principal Media38:36 Barter Agencies & Pizza Sauce Trades42:41 Why Google & Facebook Broke the Agency Model46:01 Value vs. Cost & Vinnie Rinaldi's Post49:25 The Dirty Secret of MFA Arbitrage51:53 Meta Loses Social Media Addiction Case58:21 The Guardian & Publisher Direct Sales01:01:36 Outro & The Ibuprofen Overdose Story
Transcribe →We Think Ads Are Starting To Work Better
Mar 2001:02:06Tap to summarizeAdam and Gareth log in to the virtual studio to unpack a week of industry backstabbing, gossip, and the true meaning of "transparency." While the industry argues over who is to blame for the lack of clarity in programmatic, the duo highlights a quote from Hershey's Vinnie Rinaldi, arguing that it's ultimately up to advertisers to demand proof of value rather than just hunting for low CPMs.The guys break down the evolution of the enterprise market, explaining why CFOs don't care about vanity metrics like Video Completion Rates when evaluating a $100 million digital investment. They discuss how brands are starting to apply the same rigorous causal inference models to ad tech that they currently use for their physical supply chains.Plus, a deep dive into the CTV ad server wars (Freewheel vs. Publica vs. SpringServe), a look at OpenAds' rapid pivot from LLM networks to a creative-first DSP, and a spicy prediction on what Netflix should (and probably will) do with its burgeoning ad business.Key Topics:The Transparency Debate: Why agencies and ad tech nerds define "transparency" completely differently.The Value of Digital Media: Why the enterprise market must move past vanity metrics and demand real ROI proof.Supply Chain Logic: How brands are treating ad tech data like logistics data.CTV Ad Serving: The rise of Publica and SpringServe, and why legacy players missed the boat.Netflix's Ad Future: Will they open up to biddable programmatic, or build a walled garden for SMBs?00:00 Intro: Virtual Sets & Industry Gossip03:09 Jay Friedman & Vinnie Rinaldi on Transparency08:37 Why Advertisers Must Demand Value11:38 The Enterprise Wake-Up Call (CFO Scrutiny)16:14 Applying Supply Chain Logic to Ad Tech22:02 The "Agentic" Layer Debate24:31 Why Publishers Still Love Direct Sales28:14 SSPs Becoming DSPs (The Blurring Lines)30:30 The CTV Ad Server Wars (Freewheel vs. SpringServe)32:16 The Rise of Retail Media Ad Servers38:09 Vinnie Rinaldi's Media Playbook42:49 OpenAds Pivots to Creative AI46:16 Why Publishers Feel Betrayed by Programmatic49:06 Speculating on Netflix's Ad Strategy55:05 The Danger of Hiding Publisher Data01:00:45 Closing Thoughts: Models & World Peace
Transcribe →User IDs and the Powers That Be
Mar 1300:57:21Tap to summarizeAdam and Gareth return from the spring conference circuit (RampUp and Beeler Basecamp) feeling reflective. Adam unpacks the repressed trauma of the startup grind, while Gareth shares a brutal story of getting stiffed for $350k by an industry "partner" and why selling your company doesn't automatically mean a happy ending.On the industry front, the duo tears into the "weird garbageness" of the Connected TV (CTV) market. They explain how 12-hour ad caches create shady "dark pools" for arbitrage, and why premium publishers are fleeing biddable CTV for guaranteed deals. Adam revisits his classic pitch on why relying on Facebook’s algorithm forces DTC brands to subsidize their own competitors, and Gareth drops his spiciest take of 2026: Yield management no longer exists. Plus, a deep dive into how User IDs are artificially inflating programmatic auction prices, why SSPs are effectively becoming DSPs via the Agentic Realtime Framework (ARTF), and a hilarious look at a new AI app for couples founded by a former mobile ad network CEO.Key Topics:Founder Trauma: The emotional toll of bootstrapping, getting scammed, and surviving corporate acquisitions.The CTV Garbage Fire: Why Connected TV pricing is broken and how long cache windows enable arbitrage.The DTC Trap: How Meta captures the value of direct-to-consumer brands by giving away their audience data.The Death of Yield Management: Gareth argues that modern SSPs don't manage yield; they just manage prebid competition.The ID Illusion: Why bidding based on inventory signals is vastly cheaper than bidding on DSP audience IDs.00:29 Intro & Conference Exhaustion02:08 Repressing Founder Trauma06:01 Life After the Company Sale08:46 The CTV Garbage Fire11:44 CTV Arbitrage & Dark Pools13:45 How Facebook Eats DTC Brands16:18 The Unhinged Agency Hiring Process21:30 The Problem With User IDs28:33 Bid Decoration & ARTF30:14 Postmortem: Yield Management is Dead38:05 Why Open Path Isn't an SSP42:11 SSPs are the New DSPs48:49 AI vs Principal Trading52:32 Ad Tech Meets AI Sex Apps
Transcribe →