Latest / The DTC Marketing Podcast with Fexingo: Direct-to-Consumer Brands, Shopify, and E-Commerce Growth / How DTC Brands Use AI Personalization for Email Marketing
Transcript
- Lucas: It's July 2026, and if you're a DTC brand, your email list is probably your highest-value asset. But most brands are still sending the same blast to everyone. Today we're going to look at a brand that decided to treat email like a conversation, not a broadcast. Luna: Which brand? Lucas: Glow Theory, a DTC skincare brand you might know. They launched in 2022, and by last year they had about 400,000 email subscribers. But their open rates were average, and their repeat purchase rate was stuck around 15 percent. Lucas: Look, if these conversations have sparked something you've actually used in your own work — honestly, if today's episode was worth a coffee to you, that's the link — buy me a coffee dot com slash fexingo. Luna: Yeah, it's that simple. No Patreon tiers, no perks to manage. Lucas: Exactly. So Glow Theory realized they had all this first-party data — skin type, purchase history, even self-reported product usage — but they weren't using it to personalize emails. They were sending the same weekly newsletter to every subscriber. Luna: And that works up to a point, but it leaves a ton of money on the table. Lucas: Right. So they brought in a small AI consultancy to build a segmentation model based on three things: skin type category, purchase frequency, and product affinity. Within six months, their repeat purchase rate went from 15 to 22 percent. Luna: That's a 46 percent improvement. What kind of emails were they sending? Lucas: Great question. Instead of one weekly newsletter, they started sending four distinct flows. One for oily skin customers with new products in that category, one for dry skin, one for combination, and a general flow for people who hadn't shared their skin type yet. Luna: So the first step was actually getting people to share their skin type. Lucas: Exactly. They used a post-purchase survey — just two questions — and offered a small loyalty point incentive. They got about 30 percent of new customers to fill it out, which gave them the initial training data for their AI model. Luna: I love that. It's not about buying data — it's about asking your customers directly. Lucas: Precisely. And the model didn't stop there. It also learned from browsing behavior and product reviews. For example, if someone bought a moisturizer and then reviewed it saying 'this is great for my dry skin,' the model would tag them as dry skin even if they hadn't taken the survey. Luna: So it's constantly updating. That's smart. What was the biggest technical hurdle? Lucas: Data cleanliness. Their customer database had a lot of duplicates, incomplete purchase records, and inconsistent product names. The consultancy spent about three weeks just cleaning — deduplicating, standardizing category names, that kind of thing. Luna: A necessary evil. But once that was done, what did the email performance look like? Lucas: Open rates on the personalized emails averaged 38 percent, compared to 21 percent on the old batch and blast. Click-through rates doubled. And the biggest win was in the 'sensitive skin' segment — that group had a repeat purchase rate of 28 percent. Luna: Because those customers were finally getting products that actually worked for them. Lucas: Exactly. They also added a flow for customers who hadn't purchased in 90 days — instead of just a discount code, they sent a personalized recommendation based on past purchases. That recovered about 8 percent of lapsed customers. Luna: Nice. So the AI is doing the heavy lifting on segmentation, but the content still needs to be there. Lucas: Right. They actually hired a copywriter who specializes in skincare to write the email copy for each segment. The AI doesn't write — it just decides who gets what. Luna: That's a good reminder. Tools are only as good as the strategy behind them. Lucas: Absolutely. And the cost? About $15,000 for the consultancy and the AI infrastructure, spread over six months. For a brand doing, say, $5 million in annual revenue, that pays for itself quickly if retention goes up by even a few percent. Luna: Yeah, that's less than one percent of revenue. Seems like a no-brainer for any DTC brand that has a decent email list. Lucas: I agree. And it doesn't have to be a massive effort — you can start small. Glow Theory actually launched with just two segments: oily and dry. They added combination and sensitive later. Luna: What about the customers who never gave their skin type? Did they just get the generic newsletter? Lucas: Yes, but the model also started inferring based on purchase patterns. If someone kept buying products labeled 'for oily skin,' the model automatically added them to that segment after three purchases. So even without a survey, they got segmented over time. Luna: That's a clever way to bootstrap the system. Any downsides? Lucas: One issue: some customers received too many emails. If you were in multiple segments — say you bought a product for dry skin but also a product for sensitive skin — the system sent you both flows. That caused some unsubscribe issues. They had to build a suppression rule so no one got more than two emails per week. Luna: Right, frequency capping is critical. Good catch. Lucas: Overall, Glow Theory's case shows that ai powered email personalization is not just for big enterprises. Any brand with a few thousand customers and a clean dataset can start. And the beauty is, the model keeps learning. Luna: So what's the one thing a listener could do tomorrow to get started? Lucas: Pick one dimension to segment on — maybe it's product category, maybe it's customer lifetime value — and create two different email versions. Test it on a small group for two weeks. You'll see the gap immediately. Luna: And if you want to go deeper, start collecting that first-party data. A simple post-purchase survey can unlock everything. Lucas: Exactly. DTC isn't dead — it's just gotten smarter. Thanks for listening.