Latest / Marketing Made Simple with Fexingo: Branding, SEO, and Digital Strategy for Beginners / How Brands Use AI to Personalize Email Marketing in 2026
Transcript
- Lucas: Alright Luna, I want to talk about something that's quietly reshaping one of the oldest digital marketing channels: email. Luna: Email? In 2026? I feel like everyone's obsessed with AI chatbots and social commerce. Lucas: Right, and that's exactly why email is the sleeper hit. It's still the channel with the highest ROI — something like thirty-six to forty dollars for every one dollar spent. But what's changed in the last eighteen months is how brands are using AI to personalize at scale. Luna: We're not just talking 'Hi ' anymore, I assume. Lucas: Exactly. That's table stakes. The new wave is generative AI writing entire email bodies, subject lines, even product recommendations, tailored to each individual's recent behavior. And I want to look at one company that's been doing this exceptionally well: Stitch Fix. Luna: Oh, the clothing subscription service? They've always been data-heavy. Lucas: Totally. Their whole model relies on personal styling algorithms. But recently they've started using large language models to generate what they call 'narrative emails' — instead of just saying 'Here are your picks,' the email tells a mini-story based on what you bought last time, the weather in your area, even upcoming events on your calendar if you've synced it. Luna: Wait, they access your calendar? That feels invasive. Lucas: Only if you opt in. And the data shows about twenty-two percent of their active users do. The payoff: open rates on those narrative emails are forty-three percent higher than their standard product-push emails. Click-through rates nearly double. Luna: Okay, that's impressive. But doesn't it risk feeling like a robot wrote it? I think consumers are getting pretty good at spotting generic AI tone. Lucas: You're right — and that's the challenge. Stitch Fix actually fine-tunes the language model on their brand voice guide and past successful copy. They also have human editors reviewing a sample of the outputs. But the key insight is: ai generated emails perform best when they're surprising but relevant, not when they're perfectly on-brand. Luna: Surprising how? Give me an example. Lucas: So imagine you bought a navy blazer from them three months ago. A typical email might say 'Complete your look with these chinos.' The ai generated version might say 'That blazer you got in April? It's going to pair perfectly with our new linen shirts — and since you're in Chicago where it's hitting ninety degrees this week, we picked a breathable cotton option.' That's contextual personalization. Luna: That is actually useful. It doesn't feel like a generic upsell. Lucas: Exactly. And the technology behind it isn't that new — OpenAI's GPT-4 and Anthropic's Claude have been around. What's new is the integration with real-time data pipelines. Now brands can trigger an email within minutes of a customer action, not hours. Luna: So what does this mean for a small business that can't afford a data science team? Is this only for the Stitch Fixes of the world? Lucas: Not anymore. There are now several platforms — Klaviyo, Mailchimp, ActiveCampaign — that have built-in AI personalization features. You can feed them your product catalog and past purchase data, and they generate dynamic email content. You don't need to write a single line of code. Luna: But the quality of output depends on the quality of data, right? If your data is messy, the AI will just generate fluent nonsense. Lucas: Bingo. That's the biggest pitfall. I've seen brands jump into AI email without cleaning up their segmentation first. Then the AI suggests products that customers already returned, or references a location the customer hasn't lived in for years. It backfires fast. Luna: So the prerequisite is still good old-fashioned data hygiene. Lucas: One hundred percent. But once that's in place, the results can be dramatic. I spoke with a mid-size outdoor gear brand — they don't want me to name them — that started using AI subject line optimization last quarter. They tested ai generated subject lines against their copywriter's best ones. The AI won on open rate by eleven percent. Luna: Ouch. How did the copywriter feel about that? Lucas: She actually loved it — because it freed her up to work on strategy and higher-level creative. She started using the AI as a brainstorming partner rather than a competitor. That's the healthy dynamic. Luna: So the human role shifts from writing to editing and strategizing. I can see that working. Lucas: And there's another layer: send-time optimization. AI can predict the optimal send time for each individual subscriber based on their historical open behavior. One study I saw from a marketing analytics firm showed that using ai based send time optimization improved click-through rates by an average of seventeen percent across retail brands. Luna: Seventeen percent is huge. But doesn't that mean emails go out at all hours? Like 3 AM for some people? Lucas: It can, but most platforms let you set boundaries — say, only between 6 AM and 10 PM local time. The AI works within those constraints. And honestly, if someone always opens your emails at 2 AM, maybe that's when they're browsing. Why not meet them there? Luna: Fair point. So what's the ethical line here? At what point does personalization become surveillance? Lucas: That's the million-dollar question. The backlash against hyper-targeting is real. Apple's Mail Privacy Protection already limits open tracking. Google's been phasing out third-party cookies. So brands have to rely on first-party data and explicit permissions. The rule I like is: if you wouldn't feel comfortable telling a customer you're using that data point, don't use it. Luna: That's a good gut check. I think customers are willing to share data if they see clear value. Like Stitch Fix's calendar integration — that's a tradeoff they opted into. Lucas: Exactly. And the brands that communicate that value transparently tend to earn trust. One thing I've noticed is that the best ai personalized emails actually include a line explaining why you're receiving them: 'Because you bought X last month, we thought you'd like Y.' It sounds simple, but it works. Luna: Okay, so let's say I'm a small business owner listening. What's the first step I should take this week to start using AI in email marketing? Lucas: First, audit your data. Make sure your customer segments are accurate and clean. Then pick one campaign — maybe a welcome series or a re-engagement email — and test an ai generated version against your current one. Run the test for two weeks and measure open and click rates. The data will tell you whether it's worth scaling. Luna: And if the AI version wins, do you switch entirely? Lucas: Not entirely. I'd recommend a hybrid approach: use AI for the initial draft, then have a human review for brand voice, accuracy, and emotional resonance. The best results I've seen come from that collaboration. The AI handles the data-driven personalization, the human adds the empathy. Luna: It's like having a really fast assistant who never sleeps but still needs a manager. Lucas: Exactly. And that manager role is more important than ever. Because without human oversight, the AI can generate stuff that's technically correct but tone-deaf. I remember a travel company that sent an ai written email about 'adventure getaways' to a customer who had just lost their job. They had no way of knowing that context from the data. Luna: Oh no. That's a PR disaster waiting to happen. Lucas: It was. And it highlights the limits of personalization. AI can see purchase history, browsing behavior, maybe even location. But it can't see life events. So brands need to build in safeguards — like exclude segments based on negative signals, or offer a 'no email' option for certain situations. Luna: So it's less about replacing human judgment and more about augmenting it. Lucas: Right. And that's the theme I'm seeing across all marketing channels in 2026. AI is a tool, not a strategy. The brands that win are the ones that use it thoughtfully, with a clear understanding of their customer's context. Luna: Before we wrap, I want to mention something that ties into this. If these marketing conversations have sparked something you've actually used in your own work, we really appreciate when listeners consider supporting the show. Lucas: Yeah, it's what keeps us ad-free and focused on the content. You can find us at buy me a coffee dot com slash fexingo. No pressure, but every bit helps us keep digging into these topics. Luna: And now back to AI email. Lucas, you mentioned that the future might involve even more dynamic content. What's next? Lucas: I think the next frontier is real-time personalization based on on-site behavior while the email is being opened. Imagine opening an email and seeing a product that was just restocked five minutes ago, because the AI knows you were looking for it. Some brands are already testing that with server-side rendering. Luna: That's wild. But also a little scary — how much data is being processed in that split second? Lucas: A lot. But it's all done server-side, so the user's privacy isn't compromised in the way client-side tracking was. And the processing happens in milliseconds. The key is that the brand already has permission to use that data. Luna: So the winners will be the brands that combine data permission with genuine value. Lucas: Exactly. And that's the takeaway for today: AI personalization in email is powerful, but only if you respect your customer's boundaries. Use the technology to serve them better, not to squeeze more out of them. Luna: Well said. Thanks for the conversation, Lucas. I think I'll go audit my own data now. Lucas: Ha, good luck. And listeners, if you try an AI email test, let us know how it goes. We'd love to hear about it.