Programmatic Advertising Insights

AI is rapidly moving from passive text generators to active decision-makers. To understand where things are headed, it’s important to trace the stages of this evolution. 1. 𝗟𝗟𝗠𝘀: 𝗧𝗵𝗲 𝗘𝗿𝗮 𝗼𝗳 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗙𝗹𝘂𝗲𝗻𝗰𝘆 Large Language Models (LLMs) like GPT-3 and GPT-4 excel at generating human-like text by predicting the next word in a sequence. They can produce coherent and contextually appropriate responses—but their capabilities end there. They don’t retain memory, they don’t take actions, and they don’t understand goals. They are reactive, not proactive. 2. 𝗥𝗔𝗚: 𝗧𝗵𝗲 𝗔𝗴𝗲 𝗼𝗳 𝗖𝗼𝗻𝘁𝗲𝘅𝘁-𝗔𝘄𝗮𝗿𝗲 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝗼𝗻 Retrieval-Augmented Generation (RAG) brought a major upgrade by integrating LLMs with external knowledge sources like vector databases or document stores. Now the model could retrieve relevant context and generate more accurate and personalized responses based on that information. This stage introduced the idea of 𝗱𝘆𝗻𝗮𝗺𝗶𝗰 𝗸𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 𝗮𝗰𝗰𝗲𝘀𝘀, but still required orchestration. The system didn’t plan or act—it responded with more relevance. 3. 𝗔𝗴𝗲𝗻𝘁𝗶𝗰 𝗔𝗜: 𝗧𝗼𝘄𝗮𝗿𝗱 𝗔𝘂𝘁𝗼𝗻𝗼𝗺𝗼𝘂𝘀 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 Agentic AI is a fundamentally different paradigm. Here, systems are built to perceive, reason, and act toward goals—often without constant human prompting. An Agentic system includes: • 𝗠𝗲𝗺𝗼𝗿𝘆: to retain and recall information over time. • 𝗣𝗹𝗮𝗻𝗻𝗶𝗻𝗴: to decide what actions to take and in what order. • 𝗧𝗼𝗼𝗹 𝗨𝘀𝗲: to interact with APIs, databases, code, or software systems. • 𝗔𝘂𝘁𝗼𝗻𝗼𝗺𝘆: to loop through perception, decision, and action—iteratively improving performance.    Instead of a single model generating content, we now orchestrate 𝗺𝘂𝗹𝘁𝗶𝗽𝗹𝗲 𝗮𝗴𝗲𝗻𝘁𝘀, each responsible for specific tasks, coordinated by a central controller or planner. This is the architecture behind emerging use cases like autonomous coding assistants, intelligent workflow bots, and AI co-pilots that can operate entire systems. 𝗧𝗵𝗲 𝗦𝗵𝗶𝗳𝘁 𝗶𝗻 𝗧𝗵𝗶𝗻𝗸𝗶𝗻𝗴 We’re no longer designing prompts. We’re designing 𝗺𝗼𝗱𝘂𝗹𝗮𝗿, 𝗴𝗼𝗮𝗹-𝗱𝗿𝗶𝘃𝗲𝗻 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 capable of interacting with the real world. This evolution—LLM → RAG → Agentic AI—marks the transition from 𝗹𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱𝗶𝗻𝗴 to 𝗴𝗼𝗮𝗹-𝗱𝗿𝗶𝘃𝗲𝗻 𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲.

Canva just bought a DOOH company for $30 million. Read that again. The world's biggest design platform didn't acquire another design tool. They acquired Doohly - a Melbourne-built programmatic DOOH platform that manages, schedules, and tracks ad performance on outdoor screens. That's not a design play. That's a media play. And it tells you everything about where this industry is heading. Canva already owns your brand kit. Your social templates. Your video creative. Now they want to own the screen at your local Rebel Sport, your Mobil servo, your KX Pilates studio. Design to deployment. One platform. No middlemen. Sean Law and Tom Sawkins built Doohly after working at Vicinity Centres and seeing first-hand how painfully manual outdoor advertising still was. Their pitch was simple - do for DOOH what the iPhone did for photography. Make it accessible to everyone, not just the big guys. Six years later, Canva came knocking. Here's what most people will miss about this deal. It's not about Canva getting into outdoor. It's about the entire marketing stack collapsing into fewer platforms. Creation, deployment, measurement, optimisation - all under one roof. That's the sixth acquisition in two years. Leonardo. MagicBrief. Affinity. Mango AI. Cavalry. Now Doohly. See the pattern? For anyone still treating DOOH as a niche channel or a "nice to have," the biggest design company on the planet just spent $30 million saying you're wrong. The channel isn't emerging anymore. It's arrived. How will your media planning change when creative and DOOH buying live on the same platform as your brand assets? #DOOH #ProgrammaticAdvertising #Canva

We're running programmatic ads for an 8-figure eCom brand And in just their first 7 days, the performance is matching Meta/Google. Here’s how: THE CHALLENGE: Programmatic advertising has traditionally been seen as a slow-burn strategy. Brands expect to wait months for campaigns to optimize and deliver results. For many, this delay has been a barrier to entry. Our goal? 👉Accelerate the optimization process 👉Deliver measurable wins faster THE SOLUTION: This was a 5 step process: 1️⃣ Data Collection: We gathered all relevant client data to understand their goals, target audience, and key objectives. 2️⃣ Publisher Selection: Using The Trade Desk, we identified a list of publishers to run ads through. Publishers include companies like Activision Blizzard (which owns apps like Candy Crush). By establishing Private Marketplace Contracts (PMPs) with these publishers, we’ve gained access to their ad networks at preferential CPMs... ...making the campaigns more cost-efficient. 3️⃣ Strategy Design: We launched a Go-to-Market Strategy that combined: 👉CTV and App Ads for top-of-funnel awareness. 👉Display and Native Ads for retargeting and tracking conversions. 4️⃣ Tracking and Retargeting: For CTV ads (e.g., Roku smart TVs), we tracked site visits after ad views and used retargeting to drive conversions from these viewers. 5️⃣ Ongoing Optimization: Over the next 4-6 week period, we fine-tuned campaigns through manual optimizations, such as: 👉Adjusting targeting criteria. 👉Excluding underperforming assets. 👉Expanding into new publisher networks. THE RESULTS: After just 7 days, the brand achieved a 1.2x holistic ROAS - A STRONG early indicator in a space where optimization typically takes months. Essentially what this means - It’s just a matter of time before programmatic is on the same time horizon as Meta/Google. Getting in and testing now will give early adopters a massive head start against their competition when the inevitable adoption of programmatic becomes mainstream.

The End of an Era: Xandr’s Depreciation & What It Means for AdTech  Big news in the #AdTech world—Microsoft is officially sunsetting Xandr, marking the end of a once-dominant force in programmatic advertising. As the industry digests this shift, here’s what it means for DSP dynamics, data strategies, and the broader ecosystem:  1. The DSP Power Shift Xandr’s exit leaves a gap in the *premium programmatic* space, where it was a key player alongside The Trade Desk, Google DV360, and Amazon DSP. Expect:  - More consolidation: Buyers may shift budgets toward remaining DSPs, strengthening TTD and DV360’s dominance.   - Emerging challengers: Smaller DSPs with unique differentiators (retail media, CTV, privacy-centric solutions) could gain traction.   - Microsoft’s next move: Will they double down on their own demand stack or forge new partnerships?  2. Data & Identity Implications  Xandr’s data marketplace and identity solutions were key assets. With its deprecation:  - Publishers & buyers lose a data source*, pushing them toward alternative clean room and ID solutions.   - Microsoft’s first-party data* (LinkedIn, Xbox, Bing) may get repurposed into a new buying platform.   - Contextual & privacy-safe targeting* could see even more adoption as buyers reassess their strategies.  3. The Future of “Premium” Programmatic   Xandr was known for high-quality inventory (AT&T’s WarnerMedia, AppNexus legacy deals). Its absence might:   - Drive more direct deals as advertisers seek trusted supply.   - Accelerate curation & PMP growth*, with publishers leaning into private marketplaces.  Final Thought Xandr’s sunset is another sign of AdTech’s relentless evolution—*scale matters, but differentiation matters more. The DSPs that thrive will be those that combine transparency, performance, and adaptability in a cookie less, fragmented world.  What’s your take? Will this shift benefit the remaining DSPs, or will it open doors for new players? #Programmatic #Advertising #Data #Microsoft  #AdTech #ProgrammaticAdvertising #DigitalMarketing #Innovation #Xandr #MarketImpact

If you’re an AI engineer building a full-stack GenAI application, this one’s for you. The open agentic stack has evolved. It’s no longer just about choosing the “best” foundation model. It’s about designing an interoperable pipeline, from serving to safety- that can scale, adapt, and ship. Let’s break it down 👇 🧠 1. Foundation Models Start with open, performant base models. → LLaMA 4 Maverick, Mistral‑Next‑22B, Qwen 3 Fusion, DeepSeek‑Coder 33B These models offer high capability-per-dollar and robust support for multi-turn reasoning, tool use, and fine-grained control. ⚙️ 2. Serving & Fine-Tuning You can’t scale without efficient inference. → vLLM, Text Generation Inference, BentoML for blazing-fast throughput → LoRA (PEFT) and Ollama for cost-effective fine-tuning If you’re not using adapter-based fine-tuning in 2025, you’re overpaying and underperforming. 🧩 3. Memory & Retrieval RAG isn’t enough, you need persistent agent memory. → Mem0, Weaviate, LanceDB, Qdrant support both vector retrieval and structured memory → Tools like Marqo and Qdrant simplify dense+metadata retrieval at scale → Model Context Protocol (MCP) is quickly becoming the new memory-sharing standard 🤖 4. Orchestration & Agent Frameworks Multi-agent systems are moving from research to production. → LangGraph = workflow-level control → AutoGen = goal-driven multi-agent conversations → CrewAI = role-based task delegation → Flowise + OpenDevin for visual, developer-friendly pipelines Pick based on agent complexity and latency budget, not popularity. 🛡️ 5. Evaluation & Safety Don’t ship without it. → AgentBench 2025, RAGAS, TruLens for benchmark-grade evals → PromptGuard 2, Zeno for dynamic prompt defense and human-in-the-loop observability → Safety-first isn’t optional, it’s operationally essential 👩💻 My Two Cents for AI Engineers: If you’re assembling your GenAI stack, here’s what I recommend: ✅ Start with open models like Qwen3 or DeepSeek R1, not just for cost, but because you’ll want to fine-tune and debug them freely ✅ Use vLLM or TGI for inference, and plug in LoRA adapters for rapid iteration ✅ Integrate Mem0 or Zep as your long-term memory layer and implement MCP to allow agents to share memory contextually ✅ Choose LangGraph for orchestration if you’re building structured flows; go with AutoGen or CrewAI for more autonomous agent behavior ✅ Evaluate everything, use AgentBench for capability, RAGAS for RAG quality, and PromptGuard2 for runtime security The stack is mature. The tools are open. The workflows are real. This is the best time to go from prototype to production. ----- Share this with your network ♻️ I write deep-dive blogs on Substack, follow along :) https://lnkd.in/dpBNr6Jg

✅🖥️ TODAY WAS ABSOLUTELY INSANE IN THE ADVERTISING & MEDIA BUSINESS! 1. The Trade Desk $TTD went ALL-IN on agentic AI with the launch of Kokai Zuma, bringing AI automation, simplified measurement and a dramatically easier interface to its programmatic platform. The bigger story: media buying is starting to move from “optimize what I bought” toward AI deciding what should happen next. (The Trade Desk⁠) 2. Nielsen’s Gracenote struck its first-ever DSP integration — with The Trade Desk. Advertisers can now use show-level content intelligence inside TTD, potentially solving one of CTV’s biggest problems: knowing what programming an impression actually appeared against. (Gracenote⁠) 3. CTV is getting much more content-aware. Instead of simply buying “sports,” “news” or “entertainment,” advertisers are moving toward granular signals around the actual show, scene, mood and context surrounding the ad. That could fundamentally change how premium video is packaged and sold. (Gracenote⁠) 4. AI-powered interactive CTV took another step forward. KERV expanded its partnership with LG Ad Solutions globally, using AI to identify products, objects, scenes and contextual signals inside video and turn them into interactive and shoppable advertising opportunities. (StreamTV Insider⁠) 5. The CTV measurement battle is getting more serious. AudienceProject launched Elevate 3.0 with scaled benchmarking and AI-powered insights, while FreeWheel is integrating Truthset data-quality ratings directly into its ad-buying environment. 6. CTV continues to pull advertising dollars away from traditional TV. New AdImpact data shows movie advertisers generated 17.7 billion CTV impressions this summer — up 16% year over year — while broadcast movie commercials fell roughly 50%. Hollywood is increasingly using streaming not just as a distribution channel, but as the primary advertising vehicle for its own content. (THE MEASURE⁠) 7. The programming underneath CTV advertising is becoming a bigger story. Pixalate says News, Reality and Documentary programming accounted for roughly 75% of large-screen CTV open-programmatic ad spend in July. That means the “long tail” of streaming content isn’t necessarily where the money is — advertisers are concentrating around recognizable, high-demand programming. (GlobeNewswire⁠) 8. Netflix is still aggressively building advertising — while reshuffling the leadership team behind it. More than 60% of new subscribers in ad-supported markets are choosing the ad tier, and Netflix is targeting roughly $3 billion in advertising revenue this year. At the same time, its advertising product organization is going through a leadership transition. (BI⁠) 9. The real battle in CTV may no longer be TV vs. streaming. It is increasingly CTV vs. search and social for the same incremental advertising dollar. Streaming platforms need to prove that premium video can deliver not just reach and brand lift, but measurable business outcomes. (MediaPost⁠)

Why cutting-edge AdServing Tech is key to unlocking the power of Retail Media Networks (RMN) 🛒⚙️📈 By now, marketers are familiar with why RMNs are revolutionizing how brands connect with consumers. But what can turn RMNs into truly powerful marketing engines? The answer lies in cutting-edge adserving tech that not only delivers ads seamlessly but also ensures data-backed decisions and hyper-targeted ad placements, to boost sales and overall performance metrics 🚀 In the last few weeks, I’ve been diving into technologies like Kevel and Moloco. Each offers unique selling propositions and stands as a powerful alternative to Google Ad Manager (GAM), which is the most established player in the field. Here are some critical dimensions to consider when comparing different adserving technologies for RMNs: 1️⃣ Performance Management: Beyond tracking metrics like impressions, clicks, and conversions, adservers should enable real-time performance optimization. This helps identify trends and make data-driven decisions to improve your campaigns. 2️⃣ Business Logics: Setting business rules for ad prioritization, pacing, and capping is vital for maximizing ad revenue. These rules can be based on various metrics such as CPM, CPC, and CAC, allowing for flexible and optimized adserving strategies. 3️⃣ Customization: Experiences are key to engaging your audience. Configuring custom ad units and using creative templates help tailor ad content to match brand and audience preferences, driving ad effectiveness.   4️⃣ Creative Management: Managing creative assets involves producing, uploading, and organizing the visual and text components of ads. Efficient endpoints ensure that you can handle different ad sizes and types that are visually appealing and meet technical specifications. 5️⃣ Inventory Management: Managing the onsite properties (websites, apps, etc.) where ads are displayed involves pre-created inventory options and tools to manage and optimize this inventory effectively. 6️⃣ Scalability: A scalable ad serving infrastructure helps handle high traffic volumes and complex ad operations efficiently, by optimizing server setup, ensuring load balancing, and employing caching strategies to manage large-scale ad requests without compromising performance 7️⃣ API Integrations: The Decision API is crucial for real-time ad decision-making, allowing you to request ads and receive responses based on predefined criteria. The Management API helps you manage campaigns and creatives (CRUD). 8️⃣ Testing and Debugging: Detailed instructions for testing ad requests and responses are often overlooked . This ensures that your ad serving setup works correctly and efficiently. Debugging tips help troubleshoot any issues that arise, ensuring smooth ad operations. What am I missing? Have experience with any of these adserving technologies? How do they support your RMN strategies? Curious to hear thoughts. #advertising #media #tech

𝐀𝐈’𝐬 𝐧𝐞𝐰 𝐫𝐨𝐥𝐞 𝐢𝐧 𝐢𝐦𝐩𝐫𝐨𝐯𝐢𝐧𝐠 𝐦𝐚𝐫𝐤𝐞𝐭𝐢𝐧𝐠 𝐞𝐟𝐟𝐞𝐜𝐭𝐢𝐯𝐞𝐧𝐞𝐬𝐬 𝐚𝐧𝐝 𝐑𝐎𝐈. AI has already proven to be a valuable tool in increasing #advertising efficiency, helping brands plan better, work smarter and save time. But can AI also drive measurable impact and increase #marketing effectiveness? Nielsen has recently measured over 50,000 #AI powered brand campaigns on YouTube and over 1 million performance campaigns on Google to uncover ROAS and sales effectiveness. The results from those MMMs showed that Google AI-powered advertising solutions consistently outperformed manual campaigns in both #ROAS and sales effectiveness: 🚀 Google AI-powered video campaigns on YouTube deliver 17% higher ROAS than manual campaigns. 💪 Google AI-powered VRC for Efficient Reach + VVC delivers 23% higher sales effectiveness than VRC for Efficient Reach alone. ✔️ Adding Google AI-powered Demand Gen to Search and Performance Max campaigns delivers 10% higher ROAS and 12% higher sales effectiveness than those without Demand Gen. It's already clear that #AI isn’t just a fad or a futuristic planning tool – it’s a performance driver in #advertising.

save the 2‑page cheat sheet for your next budget meeting — and if your CFO asks you to justify your brand spend, slide them the 31‑page model guide below https://lnkd.in/gzWTFAbK -- for anyone who's ever had to justify a Brand marketing budget to a skeptical CFO, this one's ^ for you. the old myth says you can't measure the long-term impact of Brand building. you just have to "trust the process." well, I just pulled myself out of a far-too-deep rabbit hole looking for research digging into the idea of whether Brand marketing can be measured — and if so, how. and once measured, what the actual results were for the study participants in terms of **actual cash money** this one's a massive study that seemingly isn't talked about enough it analyzed over 3,500 campaigns from brands in CPG, retail, and tech, and looked at a ton of spend and the results over a long period of time tl;dr: let's go with the classic iceberg analogy the short-term, last-click sales are the part you can see above the water. but the real mass—the part that can sink your competition—is the long-term Brand base building up beneath the surface. the key question tho - how big is that hidden part? big the study found a wild 60% of total advertising ROI comes from long-term effects. for tech & durables, it's a knee-wobbling 76%. it also gave us a new rulebook for media channels: The Old Rule: "Meta is for short-term DR." The New Rule: Wrong. For retail brands (which is all i know), the long-term ROI from Facebook & Instagram was more than 2.5x its short-term impact. It’s a brand-building powerhouse... ***...if we know how to use the platform correctly...which did NOT match how we used it for the first 6 yrs we were in business***. The Old Rule: "TV is dead and can't be measured." The New Rule: Also wrong. Across every industry studied, TV showed a massive long-term ROI, often driving a higher total ROI than many digital channels. my takeaways (validated what seemed to work for us ultimately) 1. separating "Brand" and "performance" is a one way ticket to stagnation and inability to scale spend (been there. done that. doesn't end well) 2. brand building is performance marketing 3. brand building makes your performance marketing better. 4. if you want more performance, do Brand building. 5. if you're not doing Brand building, you're not doing performance marketing right

5% of apps had embedded agents Jan 2024. That number is going to hit 40% by Dec 26 (Gartner). Agents are the new distribution channel. Most teams are scrambling to understand what to build for this inevitable future. Here's the roadmap: 1. Why it matters Your users used to find your product. Now their agents choose it for them. If agents can't discover and use your API programmatically, you don't exist in this channel. 🔗 AI Agents for PMs: https://lnkd.in/eeey5Cxr 🔗 AEO/GEO Guide: https://lnkd.in/e8X-Qt_C 2. The agent access stack Your API is the engine. A CLI wraps it for composability. An MCP server wraps it for discoverability. They're parallel layers, not a sequence. 🔗 Context Engineering: https://lnkd.in/eUUPMmJK 🔗 How to Build AI Products: https://lnkd.in/eDGmsvZ5 3. When to use what API when you need bulk ops and latency control. CLI when your users compose with grep/jq/xargs. MCP when you want tool discovery and multi-client reach. The guide has the full decision matrix. 4. 5 things agents need Discoverability. Programmatic auth. Structured I/O. Idempotency. Rate limits. Skip any one and agents route around you to a competitor that got them right. 🔗 PM Prompt Library: https://lnkd.in/eHU88esH 🔗 AI Evals: https://lnkd.in/eGbzWMxf 5. The numbers Anthropic, Linux Foundation, Gartner, and GitHub are all pushing this forward: • 97M+ MCP SDK downloads per month • 60,000+ repos with AGENTS.md • 8,600+ servers on PulseMCP 6. 10 production MCP servers Stripe, Linear, Sentry, Asana, Shopify, Cloudflare, GitHub, Notion, Zapier, Pendo. I analyzed them all to see what patterns matter most. 🔗 Analysis: https://lnkd.in/gHHGngYe 7. The 7 mistakes CLI checkbox with 3 endpoints and "done." Vague tool descriptions agents can't parse. Day-one write access where the agent deletes prod. The fix for all 7: treat the agent as a first-class user with a PM who owns the experience. 8. Your roadmap • This week: 5-question audit + ship AGENTS.md • This month: read-only MCP server + list on PulseMCP • This quarter: approval flows, agent analytics, agent-specific pricing 9. Every resource in one place Comment below and Direct Message me 'Agent distribution links' and I'll send you the high resolution infographic + links. 🔗 Full Deep Dive: https://lnkd.in/gHHGngYe 🔖 Save this. Repost so your product team sees it too. ➕ Follow Aakash Gupta for daily AI PM content. The companies building agent access now are going to own this channel the same way early SEO adopters owned search. Here's your roadmap.