AI in Ecommerce Marketing

Yesterday, in the flood of mind-blowing, benchmark-setting, GPU-melting AI announcements, it was easy to overlook the quiet little beta announcement coming out of Amazon - one that focuses less on the tech and more on the consumer, asking a question as old as innovation itself: “Cool tech bro, how do you monetize that tho?” Enter: Interest AI. ✨ Amazon’s new LLM-powered assistant, now in beta, lives inside the shopping app. It’s trained not on the open internet - but on YOU. What you’ve browsed, bought, returned, reviewed, streamed at 2 a.m., and forgotten in your cart. Ask it: “What’s a good beginner camera?” Get: “Here’s one based on your budget, your previous purchases, and your mild obsession with aesthetically pleasing home decor.” It doesn’t just answer questions. It answers your questions. Personalized, contextual, and commercial from the jump. But here’s the real play: Interest AI doesn’t just respond to intent - it generates it. It constantly scans Amazon’s massive, ever-expanding catalog to surface new items tied to your passions - travel, fitness, cooking, your cat’s wardrobe. It transforms how you discover, not just how you shop. It's not just a smarter search bar - it's a predictive, personalized discovery engine at scale. Interest AI not sexy. It won’t pass a Bar exam. But it might get you to click “Add to Cart.” And that, of course, is the point. Amazon isn’t chasing AGI. It’s chasing 💰 CLV (customer lifetime value) 💰 . While others build general-purpose LLMs, Amazon builds contextual commerce machines. This could quietly become one of the most monetizable use cases of LLMs we’ve seen to date. And it leans into Amazon’s real edge: first-party data, not foundational models. While the market experiments with AI co-pilots, Amazon just strapped a personalized sales engine to the world's biggest mall.

At the start of my career, pricing was often treated as an afterthought. Decisions were made based on instinct, outdated models, or by simply matching competitors. I witnessed how this approach consistently led to underperformance, weak positioning, and lost revenue opportunities. That experience shaped my belief that pricing is one of the most overlooked drivers of business growth. To solve this, we built the Predictive Sales Engine an AI-powered tool that brings clarity to pricing strategy. It analyzes actual market behavior to forecast revenue and sales volume at different price points. More importantly, it segments data to reveal how different audiences respond to pricing, allowing companies to set prices with precision and confidence. After working with hundreds of companies, the pattern is clear. When pricing aligns with how customers perceive value, businesses grow faster and more profitably. In a competitive market, using AI to guide pricing decisions is no longer a luxury. It’s a requirement for those aiming to lead rather than follow. #PricingStrategy #ArtificialIntelligence #PredictiveAnalytics #RevenueGrowth #ProductMarketing

Amazon just launched Brand+, a new AI-powered solution that identifies consumers likely to purchase in the next three months. This is the thin end of the wedge. With Brand+, they're not just serving ads—they're predicting who will buy in the next three months. How? By analyzing trillions of shopping, browsing, and streaming signals. This means: • Ads on Prime Video, Twitch, and Fire TV • Targeting based on real-time purchase intent • Access to third-party platforms like BuzzFeed and Fox Amazon isn’t selling ad space. They’re selling predictive consumer behaviour. And here’s why that matters: For years, performance marketing was about fine-tuning data signals—finding micro-optimizations in ad spend, creative, and bidding strategies. Now, the value of human decision-making is shifting upstream: • Identifying customer pain points • Crafting compelling narratives • Positioning brands meaningfully When AI optimizes everything else to the point of indifference, the last competitive advantage will be what you say, not just who you reach. In the next 3-5 years, brand storytelling won’t be a “nice to have.” It will be the only thing separating winners from the noise. #marketing #business #career

Scott Brinker's 3 Domains of AI Agents in Marketing cuts through the noise for retail and CPG teams trying to figure out where to invest. Looking at this through a commerce lens, it clarifies exactly where the disruption is happening. Reading the spectrum from left to right. That's not just a taxonomy. It's a map of where you're losing control of the customer journey. 🔵 BLUE (Left side): Agents FOR Marketers High control, high visibility. Adobe, Canva, HubSpot, Salesforce for creative production and decisioning. GrowthLoop, Treasure Data for agentic CDPs. This is where 68.9% of current AI adoption sits. Safe. Productive. Completely disconnected from where many shopping decisions are actually happening. 🟢 GREEN (Middle): Agents FOR Customers Inside/outside boundary. This is the battleground. Shopper Concierges like Bloomreach and Manifest AI. Customer Service Agents like Sierra and Zendesk. AI Sales Agents like Regie.ai & 11x. Only 14.6% have deployed shopping assistants. Meanwhile, your customers are already asking ChatGPT "what's the best [your product category]" & transacting without ever touching your site. Eventually these will become monetizable assets via firms like Thrad. B2C is ahead here (69.4% have customer service agents vs 46.3% in B2B), but mostly in reactive support, not proactive commerce. 🟡 GOLD (Right side): Agents OF Customers Low control, low visibility. ChatGPT, Claude, Perplexity, Gemini making decisions on behalf of shoppers. Agentic Browsers like Opera Neon and MaxAI. Consumer Agents deciding what to buy, when & at what price. This is where McKinsey & Company's $750B in consumer spend flows by 2028. Notice the dashed line in the middle? That's the control cliff. The Commerce Gap: 🛒 Everyone's investing left (internal productivity) 🛒 The customer journey is moving right (external AI intermediation) 🛒 The middle domain (your customer-facing agents) is underfunded and undermeasured Looking at middle again. See those AI Adaptive Websites? AI Sales Agents? That's the infrastructure for agentic commerce. Bloomreach and Manifest AI guiding product discovery, Nosto and Constructor adapting experiences in real-time based on AI understanding of intent. But here's the brutal math from Brinker's data: 63.1% are optimizing content for AI discovery, but only 13.6% are measuring AI inclusion rate or agent-referred conversions. As a collective, we're building for a channel we can't see. The vendors aren't the insight. It's the control gradient. As you move from left to right, your ability to own the experience evaporates. Your martech stack was architected for the left domain. Your customers are shopping in the right domain. The key leverage point? The middle. Customer-facing agents, machine-readable product feeds, API-first commerce infrastructure, real-time context engineering. Looking at your stack through this lens, where's the gap? Full report from Scott Brinker and Frans Riemersma at chiefmartec below 👇

Commerce Media powered by AI Agents: The playbook may look very different 🛒🤖 Rufus is Amazon’s AI shopping guide: it interprets intent, evaluates options, and surfaces products based on relevance, not keywords. And the latest research from Profitero+ and Mars United Commerce highlights just how different this AI layer behaves. They compared Rufus results with page-1 search for the same prompts over two months. Here’s what stood out 👇 1️⃣ Only 22% of page-1 products appeared in Rufus. The old playbook of “win page-1 = win the shopper” won’t survive in AI-driven commerce media. Rufus is curating results, not just mirroring search. 2️⃣ 36% of Rufus picks weren’t even on page-1. AI is elevating products with zero traditional visibility, based purely on relevance. A very different model of influence. AI commerce assistants aren’t replacing search yet… but you can already see the blueprint of how AI-powered commerce media will operate. What rises to the top will be driven by: ✅ Structured, high-quality product data ✅ Clear, attribute-rich descriptions ✅ Audience and context relevance signals ✅ Shoppable, intent-led experiences Not by: ⛔️ Keyword stuffing ⛔️ Legacy SEO rankings ⛔️ Traditional shelf logic Are brands and Retail Media Networks getting ready for these foundational shifts? #advertising #media #tech #ecommerce #ai

Delta Air Lines is piloting AI-driven dynamic pricing on a portion of its fares, with plans to expand the program substantially by year's end. Framed as a modernization of pricing strategy, this shift warrants a deeper examination of how algorithmic systems are shaping access and at what cost. Dynamic pricing is often described as demand responsive. But in execution, it frequently introduces volatility that obscures fairness. Similar approaches in retail have led to disproportionate price increases in lower income communities, raising concern that these systems are less responsive to human need than to data correlations detached from context. Several issues demand scrutiny: - Bias and disparity: Pricing algorithms can reproduce regional, racial, and economic inequities, particularly when data reflects underlying structural imbalances. - Loss of predictability: Consumers face fluctuating costs without the tools to understand or anticipate those changes, making budgeting and planning increasingly difficult. - Opaque logic: There is little transparency around how these models are developed, what inputs are prioritized, or what safeguards exist to ensure equitable outcomes. Delta reports that early results are "amazingly favorable." Without clarity on who benefits, how outcomes are defined, or which metrics are being used, these claims raise more questions than they resolve. This initiative signals a broader transformation in how corporations deploy AI across consumer-facing systems. These models are increasingly designed to maximize extraction without transparency or accountability. The consequences are rarely confined to the checkout screen. They affect who has access, who carries the burden, and who is excluded from the benefits of technological progress. This also applies to healthcare. As AI becomes more embedded in clinical decision making, triage, and resource allocation, the same concerns apply. We acknowledge that a lot of policies and stands in the field have been adopted from aviation. Hello, Universal Protocol! An algorithm that controls pricing today could soon influence how risk is scored or how treatment urgency is determined. Without safeguards, these systems risk distorting clinical judgment and widening disparities in care. What begins in commerce often finds its way into health systems, especially when the underlying logic is left unchallenged. In addition to technical efficiency and "optimization," we need governance frameworks that prioritize equity, transparency, and accountability across every domain touched by AI. "It's not about what it is, it's about what it can become."- Dr. Seuss #healthcareonlinkedin #aiethics #consumerrights #aiinaviation

🥤Just explored an incredible customer story at Dreamforce that perfectly illustrates what "becoming an Agentic Enterprise" actually means in practice. PepsiCo - yes, the company behind Lay's, Doritos, Gatorade, and Pepsi - is deploying Agentforce across their global operations. And the scale is staggering. 🎯 THE CHALLENGE
👥 100,000+ employees using Salesforce to manage: * 🛒 120,000 field sales reps keeping shelves stocked * 🏪 1M+ small retailers (mom-and-pop shops, bodegas, gas stations) * 🚚 25,000+ delivery routes * 🚛 One of North America's largest private fleets ⚠️ Problem: Big retailers like Walmart get daily visits. Small retailers don’t — but they represent huge volume. Missing one day of shelf presence = significant revenue loss. 💡 THE SOLUTION
Agentforce deploying across 100+ contact centers to give smaller retailers instant assistance through online portal or phone (Agentforce Voice). 🧩 The stack: * 📊 Data Cloud — Unifies data across all systems, zero-copy architecture * 🤖 Agentforce — Handles basic inquiries with access to real-time catalogs, inventory, promotions * 🚚 Consumer Goods Cloud — 120,000 field reps get planograms, route planning, metrics * 📞 Service Cloud — B2B portal for retailers * 📱 Marketing Cloud — Personalized retailer communications * ⚙️ Field Service — Managing fleets and equipment * 🔗 MuleSoft — Integrating enterprise systems 💬 THE QUOTE
Athina Kanioura, Chief Strategy & Transformation Officer:
"Embracing an AI-first world means reimagining an enterprise where humans and intelligent agents don't just coexist, they collaborate." 📈 THE IMPACT * ✅ Agentforce handling nearly all basic inquiries ("Where's my order?" "Why short shipment?") * 🙌 Human teams freed for high-value relationships * 🤝 Better retailer engagement (feedback becomes faster, simpler) * 🚀 20 future use cases identified — personalized recommendations to AI-generated planograms 🎯 WHY THIS MATTERS
This isn't a pilot. PepsiCo (320,000 employees, $92B revenue) is one of the first major CPG companies deploying Agentforce at scale. This is what the Agentic Enterprise looks like in traditional, asset-heavy business: * ✅ 100,000+ employees empowered by unified platform * ✅ Digital labor scaling support to millions of retailers * ✅ Human workers elevated to strategic work * ✅ Data unified across apps and geographies * ✅ Integration as competitive advantage 🙌 Not replacing 120,000 field reps. Augmenting them. Giving small retailers who couldn't get daily visits the same service as major accounts. ⚡️ That's the digital labor economy in action. When a company this size goes "all in" on Agentforce, it validates this isn’t just tech hype. It’s the future of complex global operations. Check out my interview with Tracy Matis VP Transformation and CX at PepsiCo. #Salesforcepartnership

AI-powered ads are coming. Nobody can seem to agree on how to do them. Perplexity tried ads but pulled back due to concerns about objectivity. OpenAI launched ads in ChatGPT last month. And now Amazon Ads is making a move that I think deserves more attention than either of those. Amazon Publisher Services is in early conversations with sites and external firms to power ads within their chatbots. This isn't Amazon building its own chatbot. It's Amazon approaching from the publisher side first: help platforms monetize, then let the demand follow. The move would extend Amazon's ad tech influence well beyond retail media, give it valuable data on user behavior outside its own ecosystem, and put it in direct competition with OpenAI. Here's why I think this is the one worth paying the most attention to. The AI platforms entering the ad space, like OpenAI and Perplexity, are genuinely innovative. But they're building ads and commerce expertise from scratch. They know AI. They don't necessarily know ad tech infrastructure, advertiser relationships, measurement, or how to build trust with brands managing millions in spend. Case in point: it was recently reported that OpenAI has scaled back its direct checkout ambitions in ChatGPT after finding that users browse but don't buy. Commerce is hard. The infrastructure behind it is even harder. Amazon has spent decades building both. They also bring mature data, measurement, and ad tools that marketers already need and expect. That's fundamentally different from selling placements inside your own AI product. And frankly, it's the approach most likely to work. The proof is already in the numbers. Having spent nearly four years at Amazon Ads, I've seen firsthand how the company builds infrastructure that scales beyond its own walls. ADSP now reaches 300+ million ad-supported users in the U.S. It powers everything from Prime Video to Netflix inventory. It helped drive $21.3 billion in ad revenue in Q4 alone, up 23% YoY. Our data at Skai tells the same story from the buy side: upper-funnel DSP investment surged 72% YoY last quarter while costs actually fell 24%. That's not experimentation. That's marketers voting with their budgets because the infrastructure delivers. Powering chatbot ads for third-party platforms? Same playbook. Next frontier. For brands, this is encouraging. Instead of navigating yet another walled garden with unproven ad tech, you could potentially manage chatbot advertising through an ecosystem you already know and trust. Discovery, recommendations, and transactions blending into AI-mediated experiences, powered by the same infrastructure that's already working across your commerce media program. The complexity isn't going away. But the right infrastructure can make it navigable. And right now, nobody's better positioned to build those pipes than Amazon.

Yesterday, Amazon Ads dropped TWO announcements that are about to change retail media strategy. 1) Amazon Marketing Cloud is now self-service This is HUGE. Until now, AMC's sophisticated capabilities were locked behind barriers. You needed technical expertise, SQL knowledge, the whole thing. Not anymore. Any advertiser can now access these insights through intuitive, no-code templates with AI assistance. You can finally connect AMC learnings across your entire retail media strategy. Budget shifts, audience adjustments, keyword targeting, all of it unified in one place. 2) Creative and media activation are no longer separate. Amazon is rolling out integrated creative tools for Sponsored Brands, DSP, and more, so you can build, launch, and measure creative in one workflow. That means less lag time between concept and execution, as well as the ability to measure creative performance against campaign results in real time. The bigger picture is clear. Amazon just set the new standard for what retail media should look like.

Retail just crossed a line. Target “Good & Gather” eggs are $1.99 if you’re browsing from Rochester, NY… and $2.29 if you’re sitting in Tribeca. The difference isn’t logistics or supply chain. It’s not cost of goods. It’s not regional operations. It’s an algorithm—one that Target now discloses, quietly, behind an “i” icon—using your personal data to decide what you pay. New York State’s new law forced the disclosure. And that’s the only reason we’re seeing this at all. Businesses that set prices using consumer data now must acknowledge it, but they still don’t have to explain what data they use, how it’s weighted, or why one customer sees one price and their neighbor sees another. Let’s be clear about what this means: AI is no longer just optimizing supply chains or search results. It’s calibrating the price of basic goods based on who you are, where you are, and what the algorithm thinks you’ll tolerate. The FTC has already warned about “surveillance pricing”—the practice of using granular data, including precise location, to set individualized prices. This isn’t dynamic pricing for airline seats or surge pricing for rideshares. This is grocery pricing tied to personal data and opaque profiling. Privacy concerns aside, the brand implications are enormous. Once consumers realize they’re not getting the same price as the person next to them, trust breaks instantly. No amount of loyalty marketing or brand storytelling fixes that. This is the opposite of transparency. It’s the opposite of value. And it undermines every promise retailers make about fairness and consistency. As an industry, we’re about to face a reckoning. Not about AI—AI is here to stay. But about how far we’re willing to push personalization before it becomes exploitation. If retailers want long-term loyalty, the answer is simple: Innovate on experience. Innovate on efficiency. Innovate on value. But don’t monetize consumers’ identities through invisible, algorithm-driven price discrimination. That’s not the future of retail. That’s a fast track to losing customer trust—and once it’s gone, it’s gone for good.