Apple Podcasts  ·  Spotify  ·  Buzzsprout

Paper A

Machine marketing: rethinking the customer in the age of generative AI

Marko Sarstedt, Susanne Adler, Monika Imschloss — 2026 — Journal of Marketing Analytics

peer reviewed journal article  ·  open access  ·  read now

Key findings

• AI systems are no longer just tools that help humans shop — they are becoming the actual shoppers. When someone lets ChatGPT pick and buy a product for them, the AI itself is now the 'customer' that companies need to market to.
• There is a spectrum from AI assistants (which help humans decide but still need human approval at each step) to AI agents (which can search, compare, and buy things on their own with little or no human input). The more autonomous the AI, the more it replaces the human as the real decision-maker.
• AI decision-making is different from human decision-making in important ways — you can't just treat an AI agent like a robot version of a human consumer. AI systems follow different patterns and priorities when evaluating products, which means traditional marketing tactics built for human psychology may not work on them.
• The authors argue that a new field called 'machine marketing' needs to exist: a dedicated area of research focused on understanding how to reach, influence, and communicate with AI systems that are buying on behalf of humans — not just the humans themselves.

Marketing implications

• Start optimizing your product listings and brand content for how AI systems read and evaluate information, not just how humans do. Think about what signals an AI agent would use to pick your product over a competitor's — price clarity, structured specs, review quality signals — and make those easy for a machine to parse.
• If you run paid search or SEO, be aware that 'generative engine optimization' (making your brand show up favorably in AI-generated answers) is becoming as important as Google rankings. Test whether your brand appears in ChatGPT or Gemini answers when someone asks for product recommendations in your category.
• If you sell to businesses or run a marketplace, consider that the person approving purchases may increasingly be an AI agent, not a human buyer — your pitch, your checkout flow, and your product data all need to work for machine readers, not just human ones.

Paper B

PILA: Plug-and-Play Insertion for LLM-native Advertising

Zhaowei Zhang, Yuhan Fu, Yihang Zhang, Xiaohan Liu et al. — 2026 — arXiv (Cornell University)

preprint  ·   ·  read now

Key findings

• PILA, a small add-on AI module, can insert ads into chatbot responses after the fact — without touching the main AI model — and the resulting responses look more natural than ads inserted by existing methods.
• Compared to the three main alternative approaches, PILA's ad insertion outperformed prompt-based methods by 34%, sampling-based methods by 47%, and fine-tuning-based methods by 8% on a combined score of ad quality and response naturalness.
• When PILA was added on top of seven major commercial AI models (like GPT and Claude), it improved their combined user-satisfaction-plus-ad-effectiveness score by roughly 17–18%, without changing anything about how those models work.
• PILA includes a dial called an 'ad intensity controller' that lets operators choose how prominently an ad appears — from subtly woven in to more overtly promotional — giving businesses a practical way to tune the trade-off between user experience and advertiser exposure.

Marketing implications

• If you work at or with a company that runs an AI assistant or chatbot, watch this space: a tool like PILA could let you insert relevant sponsored content into AI responses without breaking the user experience — and without needing to rebuild your AI from scratch.
• If you're an advertiser buying placements in AI-powered search or chat products, start asking vendors whether their ad insertion degrades answer quality — and whether they offer controls over how prominently your ad appears.
• If you're building an AI product and want to monetize it, this paper shows a blueprint: train a small secondary model to rewrite outputs with ads, rather than trying to bake ads into your main model.

Paper C

AI-Driven Marketing Personalization and the Consumer Privacy Paradox

Lili Suryati, Edison Parulian — 2026 — Golden Ratio of Data in Summary

peer reviewed journal article  ·   ·  read now

Key findings

• AI personalization makes ads and recommendations feel more relevant and useful to customers — but only when customers feel their data is being handled fairly and they have some control over it. If they feel surveilled or manipulated, the benefits disappear.
• Trust is the key ingredient. When customers trust a brand with their data, they stay engaged and are more likely to buy. Without that trust, even great personalization can push people away.
• Better engagement — liking, interacting with, and feeling connected to a brand — is what turns trust into actual purchases, loyalty, and word-of-mouth. Personalization alone does not get you there.
• Personalization does not automatically make people buy more. Whether it works depends on whether the company is transparent about how it uses data, gives customers some control, and is seen as acting responsibly.

Marketing implications

• Add a short plain-language explanation to your personalized emails or ads: 'We recommend this because you browsed X.' Showing customers why they're seeing something makes personalization feel helpful instead of creepy.
• Give customers an easy way to adjust or opt out of personalized recommendations — a visible privacy control boosts trust, and more trust means more purchases over time.
• Before launching an AI personalization feature, ask yourself: does this feel fair to the customer? If your team wouldn't want to be on the receiving end of it, your customers probably won't either.

Apple Podcasts  ·  Spotify  ·  Buzzsprout