AI & Marketing Research with Dr. Eva Wolf
Not another AI news podcast. This is a research radar — a twice-weekly briefing that surfaces peer-reviewed studies on AI and marketing, tells you what the evidence actually says, and helps you decide what's worth a deeper read.
AI & Marketing Research with Dr. Eva Wolf
AI Marketing Research: Data Gaps, Trust Risks & Personalization
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
You're listening to Avita, an AI-generated research briefing avatar trained on the research framework and methodology of Dr. Eva Wolfe, marketing professor, AI researcher, and founder of Big Plans Media. Every day, Evita scans emerging research in AI, marketing, consumer behavior, psychographics, and business strategy to identify the most relevant developments, opportunities, and risks worth watching. These daily radar reports are designed to help busy professionals stay informed without having to read hundreds of research papers themselves. And every Friday, join Dr. Eva Wolf live for her personally recorded weekly AI Marketing Radar Roundup, where she breaks down the biggest stories, explains what actually matters, and shares practical insights and strategic implications for marketers, educators, entrepreneurs, and business leaders. Now here's today's radar report. Here's the uncomfortable question I kept coming back to this week. You've got data, tons of it. Customer behavior, engagement signals, purchase history, sitting in your CRM, your ad platform, your analytics stack. And your marketing results are still fine. Not great, fine. Why isn't the data doing more? That's the thread running through today's papers. Three studies all circling the same tension between AI's real potential in marketing and the ways it quietly fails you when you don't deploy it right. We screened 140 papers this week. Three made the radar. Quick caveat: this is a first pass research briefing, not a final academic review. I'll tell you what the papers suggest, what they don't prove, and which ones deserve a deeper read. All three are abstract-only access. Everything I'm sharing comes from what the authors published in their abstracts. I'll flag that where it matters. Okay, let's get into it. Paper one. Here's the core idea. Having data doesn't help you. Using AI on that data does. Peer-reviewed quantitative study out of Egypt, B2B companies, 148 managers, path modeling, well suited for this kind of mediation question. The research question was basically: Does information pressure, the stress of having more data and complexity than your team can handle, automatically lead to better marketing performance? The answer was no. When you test whether information overload directly improves marketing results, that link is statistically insignificant on its own. It doesn't work. But when AI adoption sits in the middle, when the company has actually invested in AI tools to process that information, the relationship becomes significant. AI adoption fully explains why information-heavy companies outperform. Without it, all that data is just pressure with no payoff. Okay, here's the thing. This confirms something a lot of us sense but struggle to articulate when we're making the case for AI budget. It's not the data that creates marketing performance, it's the infrastructure that acts on the data. Think about a B2B company with a full CRM, account data, behavioral signals, and they're still running campaigns that feel generic. That's this paper. The data's there, the AI layer isn't. And the finding held regardless of company size or industry sector. At least within this sample. Wait, that last part matters. Within this sample, I'll come back to that. Plain English payoff. If your marketing isn't improving despite having more data than ever, the bottleneck isn't the data. It's that you haven't built the AI layer to act on it. Money move. If you consult for or sell to B2B companies in emerging markets, there's a real opening here. Offer an AI readiness audit. Map their data flows, identify the infrastructure gaps, sell the implementation roadmap. That's a product right now. Try this by Friday. Pull up your current marketing stack and draw a literal line between where your data lives and where your decisions get made. If there's no AI tool touching that gap, you just found your next investment conversation. Evidence check. I want to be direct here. This is 148 managers in Egypt, B2B only, self-reported survey data, and we're working from abstract only. You can't take this finding and apply it to a US consumer brand or a large multinational and call it proven. The geographic scope is genuinely narrow. That said, the logic holds well beyond Egypt. AI as the bridge between data and performance isn't a regional insight. It's a structural one. Radar verdict, use cautiously. The findings clean, the methods credible for a survey design, but the sample size, the geography, and the abstract-only access mean treat this as strong supporting evidence, not proof. Use it to start the conversation about AI infrastructure investment. Don't use it as the only data point in that conversation. Paper two! This one's about what happens when your AI marketing gets good enough to feel unsettling. Synthesis Paper, not an original experiment. One author reviewed academic publications, industry reports, and case studies from 2024 and 2025. Starbucks and Nike show up as examples. The goal? Map how AI, data science, and consumer psychology are converging in modern marketing. Two ideas here that I think are genuinely underappreciated. The first, the paper cites evidence that companies using AI-powered marketing systems have seen returns on ad and marketing spend improve by roughly 20 to 30% with lower cost per acquisition. Big caveat on that number. I'm going to come back to it, but directionally it's consistent with what we're seeing in practice. The second finding is the one I care about more. AI-generated ads can trigger what the paper calls an UNSA NY Valley effect in consumers. You know the concept from robotics. A humanoid robot that's almost human but not quite makes people uncomfortable in a specific, hard-to-articulate way. The paper argues AI-generated creative can do the same thing to brand trust. It can feel off, and that feeling quietly erodes the relationship with your brand, even if the consumer can't name why. And then there's a third thing, a risk the paper calls model collapse. If your AI marketing tools are being trained on AI-generated content rather than real human behavior, the model's understanding of your consumer can drift over time. It starts optimizing for a synthetic version of your customer, not the actual one. I'm telling you, that third one is the sleeper issue in this paper. Most marketers aren't asking their vendors about it. They should be. So here's the picture. AI marketing delivers real efficiency gains at scale. But if you're not checking whether your creative feels human and you're not asking where your AI's training data comes from, you're building on a foundation that can quietly degrade. Plain English payoff. AI can make your marketing faster and cheaper. But if the creative starts feeling fake to consumers or your models learning from synthetic data instead of real behavior, you'll eventually be optimizing in the wrong direction. Try this by Friday. Show your three most recent AI-generated ads to someone outside your marketing team, someone who doesn't know they're AI generated. Ask them how the ads feel, not what they think, how they feel. That's your Uncanny Valley gut check. Evidence check. The 20 to 30% ROI improvement figure. I want to be precise about this. The author didn't run a controlled experiment to produce that number. It comes from industry reports and case studies. And those case studies are Starbucks and Nike. Companies with massive first-party data infrastructure and dedicated AI teams. If you're a mid-market brand with a modest data stack, that number is not yours to claim. Don't put it in a board deck without flagging where it came from. Also, one author, Abstract Only, published in a general engineering journal. The model collapse and uncanny valley ideas are synthesized from theory, not tested by this paper directly. They're real concepts with real evidence behind them elsewhere. But this paper is pointing to them, not proving them. Radar Verdict, Watch List. This paper's a useful map of where AI marketing is heading and where the risks cluster. The model collapse warning and the uncanny valley risk are worth tracking closely, but it's a synthesis paper from a lower-tier venue with abstract only access. Don't cite the ROI figures as empirical proof. Use this to shape your questions, not your conclusions. Paper three, lightning round, but this one earns a few extra sentences. Broad review paper covering the full AI personalization stack: machine learning, NLP, computer vision, generative AI, across major advertising and marketing channels. Three authors trace how personalization evolved from simple if-then rules to real-time neural networks that respond to individual behavior as it happens. The Finding That Matters: AI-driven personalization delivers measurable commercial benefits. And it also creates real risks around privacy, fairness, and consumer manipulation that current regulation hasn't caught up to yet. The authors propose a responsible AI marketing framework: transparency, consumer consent, bias protections. It's not empirically tested, but as a checklist for vetting your AI vendors before you sign, it's useful. The reason I'm keeping this in the lightning round, it's a review paper in a lower-tier journal with zero citations and abstract only access. The survey level view of AI techniques is genuinely helpful for teams new to this space. Use it as a map, not a proof. Plain English payoff. Before you scale any AI personalization campaign, run it through a consent and transparency checklist. Regulators are paying attention. Fixing it later costs more than doing it now. Evidence check. Zero citations, lower tier venue, no original data. Orientation material. The governance framework is the most actionable piece, but treat it as a starting point, not a validated standard. Verdict. Okay, so here's what I think is actually happening this week. All three of these papers are circling the same uncomfortable truth about where AI and marketing actually stand right now. The promise is real. The ROI evidence is real, at least at scale. And the infrastructure gap, the gap between having data and having AI that acts on it, that's real too. But here's what I keep coming back to. The failure modes are getting more sophisticated at exactly the same rate as the capabilities. Paper one tells you AI adoption is the bridge between data and performance. Paper two tells you that once you cross that bridge, you risk building creative that feels subtly wrong to consumers and training models on synthetic outputs that drift away from real human behavior. Paper three says the regulatory and ethical pressure is building underneath all of this. So the story isn't AI marketing works, go use it. The story is AI marketing works when you invest in the infrastructure, check your creative against real humans, ask hard questions about your training data, and build consent in from the start. That's a more demanding bar than a lot of teams are holding themselves to right now. And I think that gap between using AI and using it well is where most of the actual risk lives. Here's the playbook from this week. One, map the gap between where your data lives and where your marketing decisions get made. If there's no AI tool in that space, that's your investment priority. Not another data source, the processing layer. Two, show your recent AI-generated creative to someone outside your team, someone who doesn't know it's AI. Ask how it feels. Flag anything that lands as off. That's your uncanny valley audit. Do it before you scale. 3. Ask your AI personalization vendor directly. What percentage of your training data is synthetic versus real customer behavior? If they can't answer that, red flag. Document consent and transparency practices for any AI-powered campaign before you launch it, not after. Evidence check on all of that? All three of today's papers are abstract-only access. Two are review papers, not original experiments. The B2B mediation finding from paper one is the strongest empirical signal in this batch, and it's still one survey, one country, 148 managers. Use these to decide what to test, not what to blindly believe. Links to all three papers are in the show notes. Read the originals before making major decisions. And remember, two of these are available only as abstracts. If you can track down the full text, do it. See you Thursday, and if something from this episode changed how you're thinking about your AI infrastructure or made you nervous about something you're already running, I want to hear it.