Latest / The Email Marketing Podcast with Fexingo: Newsletters, Open Rates, and Subscriber Growth / How Behavioral Segmentation Boosted Revenue by 40 Percent
Transcript
- Lucas: So there is this outdoor gear brand — name I'll keep vague — that was sending the same promotional email to their entire list of about sixty thousand subscribers. Open rates hovered around eighteen percent. Revenue per send was basically flat. Luna: Sounds like every brand's first year of email marketing. Lucas: Exactly. So they decided to try behavioral segmentation — not just by age or location, but by what subscribers actually did. Did they browse tents but never buy? Did they open every email but never click? They built a simple engagement score using recency, frequency, and monetary value — that classic RFM model — plus how many product pages someone viewed in the last thirty days. Luna: That sounds data-heavy. Did they have a marketing automation platform or were they stitching it together in spreadsheets? Lucas: They were using a mid-tier ESP — nothing fancy — but it supported custom fields and conditional logic. They created three tiers: high-value frequent buyers, mid-engagement window-shoppers, and dormant subscribers. For each tier they built separate flows. High-value got early access to new arrivals and a VIP discount code. Window-shoppers got browse abandonment emails with product recommendations based on what they'd viewed. Dormant got re-engagement sequences every sixty days. Luna: And what happened to the revenue from the people who were already loyal? Did they get annoyed by all the attention? Lucas: Good question. Actually, the high-value segment saw a twenty-two percent increase in average order value because the early access made them feel exclusive. The window-shopper segment converted at nearly double the rate of the old blast campaigns. Overall, after three months of running these segmented flows, the brand reported a forty percent lift in revenue attributable to email — compared to the same period the year before. Luna: Forty percent is huge. But I wonder — how did they track attribution? Was it last-click or did they use a multi-touch model? Lucas: They used last-click, but they also ran A-B tests on their biggest product launch. Half the list got the old one-size-fits-all email, the other half got the segmented version. The segmented group outperformed by thirty-three percent in click-through rate and twenty-one percent in direct conversions. So they had a clean control. Luna: That's the kind of proof that gets a CFO to approve a better ESP. Okay, but what about the technical lift? Did they need a data scientist to build the engagement score? Lucas: Not really. They pulled a one-time export from their e-commerce platform — purchase history, email engagement stats from the ESP, and page view data from their analytics — then used Excel to assign points. Each subscriber got a score out of one hundred. High-value was above seventy-five, window-shoppers between forty and seventy-four, dormant below forty. The whole thing took one marketer about two days. After that, they updated the scores monthly via a simple CSV import. Luna: So it's replicable for a lot of small teams. I'm also thinking about the privacy side — with third-party cookies dying, behavioral data from on-site browsing might get harder to collect. Lucas: That's a sharp point. The brand actually supplemented behavioral data with zero-party data — preference quizzes during signup and a simple 'what are you into?' survey after the first purchase. So they owned the data directly. They also made sure their tracking was transparent in the privacy policy. I think that's the direction a lot of email marketers need to move — away from spying and toward asking. Luna: Yeah, it's more respectful and honestly better signal. If someone tells you they hike in spring and camp in summer, you know exactly what to send them. Lucas: Exactly. And that brand found that subscribers who completed the quiz had a fifty-seven percent higher open rate and a thirty percent higher click-through rate compared to the rest of the list. So the segmentation paid off even more when they layered on that explicit intent data. Luna: I want to try this with my own tiny list. But one hang-up — how do you avoid over-segmenting? Like, if you have twelve segments and only a hundred people in each, it's not statistically meaningful. Lucas: You're right. They started with just three segments and expanded to five after they saw the data was solid. The rule of thumb is: if a segment has fewer than one thousand subscribers, don't build a dedicated flow — instead, use dynamic content within a more general campaign. That way you still personalize without the overhead. Luna: Practical. So what's the next frontier for this brand? Predictive segmentation — machine learning forecasting which window-shoppers are about to buy? Lucas: Actually, they've started testing that. They're using a simple logistic regression model — nothing crazy — to score the probability of a window-shopper converting in the next seven days. If the score crosses a threshold, that subscriber gets a different email with a more urgent call to action. Early data suggests it's lifting conversion rates another eight to ten percent. Luna: It's really impressive what's possible with just a spreadsheet and some curiosity. Speaking of things that are worth it… Lucas: Yeah, we put a lot of research into each episode, and listener support is what lets us stay independent and focused on what actually works. So thank you to everyone who's already contributed. Luna: Alright, back to segmentation — one more thing I thought was clever. That brand also used their engagement score to suppress low-scoring subscribers from their main sends, only hitting them with a quarterly re-engagement campaign. Their deliverability improved, and their open rates overall went up twelve points. Lucas: That's a smart play. It's the same logic as a sunset policy but using your own behavioral data instead of just time. I think the big takeaway from this whole case is: don't treat your email list like a megaphone. Treat it like a conversation where you listen to what people do, not just what they say. Luna: And you can start today. Export your list, add two columns — last purchase date and last email open — and just sort. You'll find patterns immediately. Lucas: Exactly. So for next episode, we're going to explore something a little different — how email list personalization intersects with ai generated content. Can a machine write subject lines that outperform human-written ones? We'll have data. Luna: Looking forward to it. Until then, go segment something.