The marketing and media agency industry is about to make the same mistake it made with digital. Invest in AI. Use it to cut costs. Discount services. Erode margins. Repeat. The Solow Paradox tells us that productivity gains from major technology shifts take a decade or more to materialise — and the benefits go disproportionately to early movers who rethink their model, not just their tools. Klarna tried to replace customer service with AI. It didn't work. The jobs that require judgement, nuance and human connection are the last to go — and the first to define your value. The agencies that come out of this well won't be the ones that automated fastest. They'll be the ones that worked out what they're actually worth — and repriced accordingly. Thanks to Kieron McCann from A Few Good People for his guest article. Read the full piece at The Media Stack or in the link in the comments 👇
Marketing Agencies Must Rethink Their Model, Not Just Tools
This title was summarized by AI from the post below.
More Relevant Posts
-
𝐓𝐋:𝐃𝐑 old man yells at clouds: "retail is hard!" (It's me. I'm the old man.) OpenAI's retreat from native checkout places it in a familiar position: among the many digital ecosystems before it that tried to extend discovery power into transaction ownership. Google experimented with its own checkout and marketplace models. Meta invested heavily in in-app Shops. TikTok and Snap have each pushed into integrated commerce. The logic has been consistent, if rarely appropriately interrogated: 𝑖𝑓 𝑑𝑖𝑠𝑐𝑜𝑣𝑒𝑟𝑦 𝘩𝑎𝑝𝑝𝑒𝑛𝑠 𝘩𝑒𝑟𝑒, 𝑡𝑟𝑎𝑛𝑠𝑎𝑐𝑡𝑖𝑜𝑛𝑠 𝑠𝘩𝑜𝑢𝑙𝑑 𝑓𝑜𝑙𝑙𝑜𝑤. What I find striking is how lazily we are collapsing a set of assumptions about how AI agents will reshape commerce into a single, confident narrative. The prevailing discourse from “𝐴𝐼 𝑐𝑎𝑛 𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑒 𝑑𝑖𝑠𝑐𝑜𝑣𝑒𝑟𝑦” to “𝐴𝐼 𝑤𝑖𝑙𝑙 𝑜𝑤𝑛 𝑡𝘩𝑒 𝑡𝑟𝑎𝑛𝑠𝑎𝑐𝑡𝑖𝑜𝑛 𝑙𝑎𝑦𝑒𝑟” happened with surprisingly little friction. What has remained comparatively stable through decades of tech disruption in commerce is where transactions live. They tend to concentrate with players that have infrastructure, payment rails, logistics capabilities, and most importantly, consumer trust. I say this as someone who grew up in retail and later edged into digital advertising. I have been repeatedly struck by how often our industry underestimates the business of retail itself. Retail is thin margins, operational complexity, returns, fraud, inventory management, and service. It is 𝐧𝐨𝐭 simply a monetizable surface where a checkout button can be dropped in and scaled. Retail media has had to learn this lesson as well, and I would argue we are slowly but surely getting better at respecting the underlying business. But the broader tech ecosystem continues to collide with the same reality: owning attention is not the same as owning commerce. AI platforms are on their way to becoming influential discovery channels (current forecast anticipates that nearly 9% of US retail ecommerce sales will be attributable to AI platforms by 2029). But unless it can absorb the structural realities of retail, including margins, logistics, trust, and service, it is far from obvious that it will own the transaction layer. Retail economics do not bend simply because a new interface captures attention.
-
-
Sarah Marzano saying the quiet part out loud: most attempts at moving conversion up funnel have been rocky at best. AI/agentic commerce will surely have an impact at some point but we aren’t close yet. People have been comparing this to e-commerce, which started slowly at under 5% of sales and has steadily grown. But that took years, decades even. Those of us working in this space will, again, construct the future of retail, but don’t expect companies like OpenAI to lead the way. They have too many other opportunities that will pay off faster.
Retail & Commerce Media at EMARKETER | Analyst & Industry Thought Leader | Keynote Speaker
𝐓𝐋:𝐃𝐑 old man yells at clouds: "retail is hard!" (It's me. I'm the old man.) OpenAI's retreat from native checkout places it in a familiar position: among the many digital ecosystems before it that tried to extend discovery power into transaction ownership. Google experimented with its own checkout and marketplace models. Meta invested heavily in in-app Shops. TikTok and Snap have each pushed into integrated commerce. The logic has been consistent, if rarely appropriately interrogated: 𝑖𝑓 𝑑𝑖𝑠𝑐𝑜𝑣𝑒𝑟𝑦 𝘩𝑎𝑝𝑝𝑒𝑛𝑠 𝘩𝑒𝑟𝑒, 𝑡𝑟𝑎𝑛𝑠𝑎𝑐𝑡𝑖𝑜𝑛𝑠 𝑠𝘩𝑜𝑢𝑙𝑑 𝑓𝑜𝑙𝑙𝑜𝑤. What I find striking is how lazily we are collapsing a set of assumptions about how AI agents will reshape commerce into a single, confident narrative. The prevailing discourse from “𝐴𝐼 𝑐𝑎𝑛 𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑒 𝑑𝑖𝑠𝑐𝑜𝑣𝑒𝑟𝑦” to “𝐴𝐼 𝑤𝑖𝑙𝑙 𝑜𝑤𝑛 𝑡𝘩𝑒 𝑡𝑟𝑎𝑛𝑠𝑎𝑐𝑡𝑖𝑜𝑛 𝑙𝑎𝑦𝑒𝑟” happened with surprisingly little friction. What has remained comparatively stable through decades of tech disruption in commerce is where transactions live. They tend to concentrate with players that have infrastructure, payment rails, logistics capabilities, and most importantly, consumer trust. I say this as someone who grew up in retail and later edged into digital advertising. I have been repeatedly struck by how often our industry underestimates the business of retail itself. Retail is thin margins, operational complexity, returns, fraud, inventory management, and service. It is 𝐧𝐨𝐭 simply a monetizable surface where a checkout button can be dropped in and scaled. Retail media has had to learn this lesson as well, and I would argue we are slowly but surely getting better at respecting the underlying business. But the broader tech ecosystem continues to collide with the same reality: owning attention is not the same as owning commerce. AI platforms are on their way to becoming influential discovery channels (current forecast anticipates that nearly 9% of US retail ecommerce sales will be attributable to AI platforms by 2029). But unless it can absorb the structural realities of retail, including margins, logistics, trust, and service, it is far from obvious that it will own the transaction layer. Retail economics do not bend simply because a new interface captures attention.
-
-
Absolutely love this post from Sarah at EMARKETER giving a reality check on how likely it is that LLM's will own a significant share of native commerce. As Sarah says so many digital platforms have failed to convert eyeballs to meaningful commerce share. Attention does not = certainty in commerce revenue. Attention does = discovery but running a retail business is a whole different business than advertising (the inverse of what retailers have had to learn over the last 5 years).. Retail is hard and the shopper is savvy and well versed in piecing together different tools to research, shortlist, compare and find offers. Add to this, that the place of purchase is: 1) almost always selected based on underlying trust, and 2) for many product categories that place is still more often than not a physical store and suddenly OpenAI or Claude stealing vast commerce share seems far less of a certainty..stealing share from search is almost a certainty but not that the fulfillment will happen inside an LLM..
Retail & Commerce Media at EMARKETER | Analyst & Industry Thought Leader | Keynote Speaker
𝐓𝐋:𝐃𝐑 old man yells at clouds: "retail is hard!" (It's me. I'm the old man.) OpenAI's retreat from native checkout places it in a familiar position: among the many digital ecosystems before it that tried to extend discovery power into transaction ownership. Google experimented with its own checkout and marketplace models. Meta invested heavily in in-app Shops. TikTok and Snap have each pushed into integrated commerce. The logic has been consistent, if rarely appropriately interrogated: 𝑖𝑓 𝑑𝑖𝑠𝑐𝑜𝑣𝑒𝑟𝑦 𝘩𝑎𝑝𝑝𝑒𝑛𝑠 𝘩𝑒𝑟𝑒, 𝑡𝑟𝑎𝑛𝑠𝑎𝑐𝑡𝑖𝑜𝑛𝑠 𝑠𝘩𝑜𝑢𝑙𝑑 𝑓𝑜𝑙𝑙𝑜𝑤. What I find striking is how lazily we are collapsing a set of assumptions about how AI agents will reshape commerce into a single, confident narrative. The prevailing discourse from “𝐴𝐼 𝑐𝑎𝑛 𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑒 𝑑𝑖𝑠𝑐𝑜𝑣𝑒𝑟𝑦” to “𝐴𝐼 𝑤𝑖𝑙𝑙 𝑜𝑤𝑛 𝑡𝘩𝑒 𝑡𝑟𝑎𝑛𝑠𝑎𝑐𝑡𝑖𝑜𝑛 𝑙𝑎𝑦𝑒𝑟” happened with surprisingly little friction. What has remained comparatively stable through decades of tech disruption in commerce is where transactions live. They tend to concentrate with players that have infrastructure, payment rails, logistics capabilities, and most importantly, consumer trust. I say this as someone who grew up in retail and later edged into digital advertising. I have been repeatedly struck by how often our industry underestimates the business of retail itself. Retail is thin margins, operational complexity, returns, fraud, inventory management, and service. It is 𝐧𝐨𝐭 simply a monetizable surface where a checkout button can be dropped in and scaled. Retail media has had to learn this lesson as well, and I would argue we are slowly but surely getting better at respecting the underlying business. But the broader tech ecosystem continues to collide with the same reality: owning attention is not the same as owning commerce. AI platforms are on their way to becoming influential discovery channels (current forecast anticipates that nearly 9% of US retail ecommerce sales will be attributable to AI platforms by 2029). But unless it can absorb the structural realities of retail, including margins, logistics, trust, and service, it is far from obvious that it will own the transaction layer. Retail economics do not bend simply because a new interface captures attention.
-
-
Thanks Paul Frampton-Calero ⚡️ for reposting to bring to my attention and Sarah Marzano for the insight. The $144 billion forecast assumes consumers won’t notice. I think they will, and the reason is far deeper than platform economics. This is the right structural argument about owning discovery versus owning transaction, and there’s a layer underneath it worth adding. Even if AI platforms solve the checkout problem, the margins question, the logistics question, the trust question, there’s a more fundamental issue with using LLMs as buying agents that doesn’t get discussed enough. These models are trained to satisfy the person asking. That is the optimisation target. Retail trust requires something close to the opposite: a system that tells you when to wait, when the review pattern looks wrong, when something isn’t right for you. Honest brokerage sits in direct tension with a reward signal built around approval. I assumed, as many did, that development pace would simply outrun these problems. Having gone deeper on the architecture, I don’t think that’s right. This isn’t a UX problem or an integration problem. It’s a reward signal problem, and I don’t think it resolves until the models themselves change in a more fundamental way than current trajectories suggest. The $144 billion gets there only if consumers don’t notice the difference between an assistant that’s genuinely on their side and one trained to sound like it is. I think OpenAI know it, which is not only why the roll back but the desperation to generate ad revenue a tad too early in my view.
Retail & Commerce Media at EMARKETER | Analyst & Industry Thought Leader | Keynote Speaker
𝐓𝐋:𝐃𝐑 old man yells at clouds: "retail is hard!" (It's me. I'm the old man.) OpenAI's retreat from native checkout places it in a familiar position: among the many digital ecosystems before it that tried to extend discovery power into transaction ownership. Google experimented with its own checkout and marketplace models. Meta invested heavily in in-app Shops. TikTok and Snap have each pushed into integrated commerce. The logic has been consistent, if rarely appropriately interrogated: 𝑖𝑓 𝑑𝑖𝑠𝑐𝑜𝑣𝑒𝑟𝑦 𝘩𝑎𝑝𝑝𝑒𝑛𝑠 𝘩𝑒𝑟𝑒, 𝑡𝑟𝑎𝑛𝑠𝑎𝑐𝑡𝑖𝑜𝑛𝑠 𝑠𝘩𝑜𝑢𝑙𝑑 𝑓𝑜𝑙𝑙𝑜𝑤. What I find striking is how lazily we are collapsing a set of assumptions about how AI agents will reshape commerce into a single, confident narrative. The prevailing discourse from “𝐴𝐼 𝑐𝑎𝑛 𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑒 𝑑𝑖𝑠𝑐𝑜𝑣𝑒𝑟𝑦” to “𝐴𝐼 𝑤𝑖𝑙𝑙 𝑜𝑤𝑛 𝑡𝘩𝑒 𝑡𝑟𝑎𝑛𝑠𝑎𝑐𝑡𝑖𝑜𝑛 𝑙𝑎𝑦𝑒𝑟” happened with surprisingly little friction. What has remained comparatively stable through decades of tech disruption in commerce is where transactions live. They tend to concentrate with players that have infrastructure, payment rails, logistics capabilities, and most importantly, consumer trust. I say this as someone who grew up in retail and later edged into digital advertising. I have been repeatedly struck by how often our industry underestimates the business of retail itself. Retail is thin margins, operational complexity, returns, fraud, inventory management, and service. It is 𝐧𝐨𝐭 simply a monetizable surface where a checkout button can be dropped in and scaled. Retail media has had to learn this lesson as well, and I would argue we are slowly but surely getting better at respecting the underlying business. But the broader tech ecosystem continues to collide with the same reality: owning attention is not the same as owning commerce. AI platforms are on their way to becoming influential discovery channels (current forecast anticipates that nearly 9% of US retail ecommerce sales will be attributable to AI platforms by 2029). But unless it can absorb the structural realities of retail, including margins, logistics, trust, and service, it is far from obvious that it will own the transaction layer. Retail economics do not bend simply because a new interface captures attention.
-
-
I really enjoyed this piece from Sarah Marzano. Retail is hard indeed — and that’s exactly why assumptions about AI “owning the transaction layer” deserve scrutiny. One shift I do believe we will see: the Product Detail Page (PDP) increasingly becoming the new search page. Historically, both retailers and brands have looked at search pages as the epicenter of retail media. And that will remain important — likely even more powerful as agentic shopping experiences enhance discovery. But I believe the role of the PDP will increasingly move into focus. There is already massive impression potential on PDPs today, and that opportunity will only grow as more shopper journeys land directly on product pages. For brands seeking market share gains, this means fine-tuning PDP conquesting and bidding strategies — influencing the decision at the exact moment shoppers decide what goes into their cart. Looking at new buyers acquired through PDP bidding can also unlock meaningful insights into incremental demand and competitive switching. For retailers, it means thinking about new ad formats and smarter decisioning that help grow the basket through relevant additional items — while keeping the shopper experience seamless and maximizing add-to-cart. The PDP is no longer just where conversion happens. It is increasingly where discovery, persuasion, and purchase converge.
Retail & Commerce Media at EMARKETER | Analyst & Industry Thought Leader | Keynote Speaker
𝐓𝐋:𝐃𝐑 old man yells at clouds: "retail is hard!" (It's me. I'm the old man.) OpenAI's retreat from native checkout places it in a familiar position: among the many digital ecosystems before it that tried to extend discovery power into transaction ownership. Google experimented with its own checkout and marketplace models. Meta invested heavily in in-app Shops. TikTok and Snap have each pushed into integrated commerce. The logic has been consistent, if rarely appropriately interrogated: 𝑖𝑓 𝑑𝑖𝑠𝑐𝑜𝑣𝑒𝑟𝑦 𝘩𝑎𝑝𝑝𝑒𝑛𝑠 𝘩𝑒𝑟𝑒, 𝑡𝑟𝑎𝑛𝑠𝑎𝑐𝑡𝑖𝑜𝑛𝑠 𝑠𝘩𝑜𝑢𝑙𝑑 𝑓𝑜𝑙𝑙𝑜𝑤. What I find striking is how lazily we are collapsing a set of assumptions about how AI agents will reshape commerce into a single, confident narrative. The prevailing discourse from “𝐴𝐼 𝑐𝑎𝑛 𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑒 𝑑𝑖𝑠𝑐𝑜𝑣𝑒𝑟𝑦” to “𝐴𝐼 𝑤𝑖𝑙𝑙 𝑜𝑤𝑛 𝑡𝘩𝑒 𝑡𝑟𝑎𝑛𝑠𝑎𝑐𝑡𝑖𝑜𝑛 𝑙𝑎𝑦𝑒𝑟” happened with surprisingly little friction. What has remained comparatively stable through decades of tech disruption in commerce is where transactions live. They tend to concentrate with players that have infrastructure, payment rails, logistics capabilities, and most importantly, consumer trust. I say this as someone who grew up in retail and later edged into digital advertising. I have been repeatedly struck by how often our industry underestimates the business of retail itself. Retail is thin margins, operational complexity, returns, fraud, inventory management, and service. It is 𝐧𝐨𝐭 simply a monetizable surface where a checkout button can be dropped in and scaled. Retail media has had to learn this lesson as well, and I would argue we are slowly but surely getting better at respecting the underlying business. But the broader tech ecosystem continues to collide with the same reality: owning attention is not the same as owning commerce. AI platforms are on their way to becoming influential discovery channels (current forecast anticipates that nearly 9% of US retail ecommerce sales will be attributable to AI platforms by 2029). But unless it can absorb the structural realities of retail, including margins, logistics, trust, and service, it is far from obvious that it will own the transaction layer. Retail economics do not bend simply because a new interface captures attention.
-
-
Stripe executives predict agentic commerce will develop in gradual stages rather than emerge as a fully autonomous system. AI agents capable of searching, comparing, and purchasing products independently could influence up to $1 trillion in U.S. retail sales by 2030, fundamentally transforming how consumers discover and buy products. This technology represents a fundamental shift in retail operations, with global implications extending far beyond current e-commerce models. Read the full article to learn more. #AgenticCommerce #AIinRetail #CommerceAutomation #RetailTechnology
-
Stripe executives predict agentic commerce will develop in gradual stages rather than emerge as a fully autonomous system. AI agents capable of searching, comparing, and purchasing products independently could influence up to $1 trillion in U.S. retail sales by 2030, fundamentally transforming how consumers discover and buy products. This technology represents a fundamental shift in retail operations, with global implications extending far beyond current e-commerce models. Read the full article to learn more. #AgenticCommerce #AIinRetail #CommerceAutomation #RetailTechnology
-
𝐖𝐞𝐞𝐤𝐥𝐲 𝐑𝐞𝐭𝐚𝐢𝐥 𝐚𝐧𝐝 𝐞𝐂𝐨𝐦𝐦𝐞𝐫𝐜𝐞 𝐍𝐞𝐰𝐬 𝐑𝐨𝐮𝐧𝐝-𝐮𝐩 Is the "Buy" button moving? 🛒 In my years consulting for global brands, I have seen plenty of hype cycles, but 2026 is delivering a massive reality check. We are currently witnessing a pivot from "AI as a Storefront" to "AI as a Concierge." The headline news this week? 𝗢𝗽𝗲𝗻𝗔𝗜 𝗶𝘀 𝗽𝘂𝗹𝗹𝗶𝗻𝗴 𝗯𝗮𝗰𝗸 𝗳𝗿𝗼𝗺 𝗱𝗶𝗿𝗲𝗰𝘁 𝗶𝗻-𝗮𝗽𝗽 𝗰𝗵𝗲𝗰𝗸𝗼𝘂𝘁𝘀. Why? Because while shoppers love using ChatGPT for research and discovery, they still prefer the security and familiarity of a merchant’s own app or site to actually swipe the card. It is a reminder that in eCommerce, trust is still the ultimate currency. 👇 Here is your weekly breakdown of the moves shaping the retail landscape: ✨ AI-Driven Payments and Agentic Commerce Expand ✨ Global eCommerce Hits $370B - China Leads Growth ✨ Ramadan Drives Saudi eCommerce Growth ✨ Warby Parker Turns Profitable and Expands Retail Footprint ✨ EU Mandates ‘Cancel Contract’ Button for Online Shops by June 2026 ✨ Klaviyo + Google: AI-Powered Customer Experiences Take Off ✨ Abercrombie & Fitch Co. Posts Record FY2025, Beats Q4 Estimates ✨ OpenAI Pulls Back on ChatGPT Direct Checkout ✨ Criteo Joins OpenAI ChatGPT Ad Pilot ✨ Middle East Conflict Disrupts Global Shipping and Food Supply ✨ Google’s AI-Generated Landing Pages: Focused on Shopping & Ads ✨ Meta Tests AI Shopping Assistant to Rival OpenAI & Google ✨ Apple Eyes Google’s Gemini for AI-Powered iPhone Features ✨ TikTok Shop Expands Cross-Border to Mexico ✨ JD.COM Sports Launches £200M Share Buyback The big takeaway? 𝗗𝗼 𝗻𝗼𝘁 𝗯𝘂𝗶𝗹𝗱 𝘆𝗼𝘂𝗿 𝗲𝗻𝘁𝗶𝗿𝗲 𝘀𝘁𝗿𝗮𝘁𝗲𝗴𝘆 𝗼𝗻 𝗯𝗲𝗶𝗻𝗴 𝗶𝗻𝘀𝗶𝗱𝗲 𝗮 𝗰𝗵𝗮𝘁𝗯𝗼𝘁. 𝗕𝘂𝗶𝗹𝗱 𝗮 𝗯𝗿𝗮𝗻𝗱 𝘁𝗵𝗮𝘁 𝗔𝗜 𝗮𝘀𝘀𝗶𝘀𝘁𝗮𝗻𝘁𝘀 𝘄𝗮𝗻𝘁 𝘁𝗼 𝗿𝗲𝗰𝗼𝗺𝗺𝗲𝗻𝗱. Which of these moves will impact your Q3 planning the most? Let’s discuss in the comments. 👇 #EmerceConsulting #eCommerce #RetailNews #AICommerce
-
On March 18, Brands Factory is teaming up with Stripe to host an exclusive meeting for DTC leaders in Guangzhou. What we'll unpack: → 2026's defining global market trends → AI agents and intelligent customer interactions → Next-gen revenue models and payment experiences → Financial infrastructure for complex cross-border environments → The payment evolution in an agent-driven commerce era The intersection of AI and real-time payments isn't coming. It's here and it's transforming how brands scale globally. Why this matters: Smart customer acquisition and seamless transactions are no longer separate plays. The brands winning in 2026 are integrating both from day one. Excited to partner with Stripe's world-class payment infrastructure to help more brands break through in the AI era. The future of global commerce starts with conversations like this. 📍 Guangzhou | March 18, 2026
-
-
I am delighted to share a piece that my colleagues at Stripe co-authored with Bain & Company (Aaron Cheris + Justin Miller), diving into retail’s new distribution channel...one of my favorite topics! TLDR: Agentic AI is moving commerce beyond simple search & toward autonomous systems that can research, compare, and buy products for consumers. This shift creates a new landscape where retailers must decide how to interact with 3rd-party AI agents while protecting their own customer data and brand loyalty. Players like Stripe are enabling brands to monetize this new distribution channel. 🤖 AI agents are becoming a new "front door" for shoppers. ⚖️ Success likely won't be a binary choice, but a blend of participating in 3rd party ecosystems while building proprietary agents to maintain loyalty. 💳 Success depends on modernizing checkout systems with secure tokens so AI can transact safely without increasing fraud risk (a Stripe superpower). 📊 Businesses must prioritize owning the checkout process to ensure they don't lose access to vital customer insights. 🚀 A "do-nothing" approach isn't an option as 30-45% of consumers already use generative AI to research their purchases. #stripe #bain #agenticcommerce #stripeforretail #futureofretail #futureofcommerce https://lnkd.in/ehbC95MW
https://www.themediastack.co.uk/p/the-solow-paradox-gives-us-clues