How Underoutfit Turned Email Marketing Into a Continuous Revenue Growth Engine
Deployed with a single objective — maximize measurable business performance - Eikona delivered consistent gains across the entire email program over a ten-week measurement period.
+45.9%
Click-through rate
Each send taught the next, so insight compounded across the program instead of resetting each campaign.
+29.3%
Revenue per recipient
(up to +169% in best campaign)
+26.7%
Placed order rate
(Across the full program)
+10.5%
Open rate
(Across the full program)
Background
Underoutfit is one of the fastest-growing direct-to-consumer apparel brands in the U.S., focused on comfortable, confidence-building essentials for women. With a highly engaged customer base and a sophisticated lifecycle marketing program, email is one of the company's most important revenue channels.
Like many leading ecommerce brands, Underoutfit had already invested heavily in segmentation and testing. Yet the team faced a familiar gap: while everyone talks about AI-powered efficiency, very few solutions deliver measurable revenue growth.
Approach
Underoutfit partnered with Eikona to put safe, production-grade GenAI to work at scale. After a two-click integration completed in minutes — with no change to the existing marketing stack — the team moved from one-size-fits-all blasts to adaptive campaigns. Each campaign became 20 different emails instead of one, with reinforcement-learning algorithms continuously exploring for winners.
01
Adaptive Campaigns, Not Blasts
- Hundreds of images, messages, and subject lines are generated and tested continuously.
- Every asset is production-grade, brand-safe, and ready to deploy.
02
A Closed Feedback Loop
- Each send’s performance feeds back into the system, so every message informs the next.
- Reinforcement learning continuously shifts delivery toward winning creative.
03
Winners Beyond Email
- Each send’s performance feeds back into the system, so every message informs the next.
- Reinforcement learning continuously shifts delivery toward winning creative.
Business Challenge
Discover new winning creatives and automate testing with AI
Unlock incremental revenue at scale without sacrificing brand quality or product integrity
Deliver impact immediately, integrating into existing workflows with no complex migration
"A/B testing always meant two guesses and a two-week wait. Eikona blew that up — it explores hundreds of intents, learns from every send, and personalizes to each customer on its own. Our CRM stopped being a broadcast channel and became a system that gets smarter every week."
Key Takeaways
This case shows how adaptive, reinforcement-driven creative turned Underoutfit’s lifecycle program into a self-learning revenue engine — where every send compounds the last.
01
Every Message Learns
Each send taught the next, so insight compounded across the program instead of resetting each campaign.
02
Creative at the Speed of Paid Media
Lifecycle email became a continuously tested, production-grade channel — the way paid social already operates.
03
AI Built for Revenue, Not Just Output
Eikona was deployed against one objective: measurable business performance.
Geographically Adaptive Intelligence
In this campaign, Eikona’s Adaptive Marketing Engine factors in user context to find segments that respond differently to different creative. Rather than sending one creative to everyone, it looks for signals that predict what a given segment will respond to — and here it discovered a substantial signal tied to geography, splitting traffic by region and serving each region its own content in real time.
The engine surfaced two completely different winners for the same bra. In one region, simply shifting the background and lighting lifted engagement; in another, swapping to a different model drove the gains — each nearly doubling the control’s click-through rate. These were autonomous, data-driven decisions: reinforcement learning read live performance and matched each segment to the creative most likely to convert, without a marketer choosing per region.
Critically, every variation the engine produces is production-grade — on-brand, compliant, and pixel-perfect on the product itself, never a rough draft a designer has to finish. So these regional winners aren’t just test results; they’re finished, ready-to-deploy assets you can roll straight into paid, social, and every other channel. Measured against the control, the dynamic campaign produced a substantial uplift in revenue and engagement metrics:
+25.9%
RPR
+31.0%
CTR
Intent-Driven Activewear Optimization
In this campaign, a single sports-bra send was turned from a one-size-fits-all blast into a dynamic, self-optimizing campaign. Rather than guessing which message would land, the system tested a more fundamental question on its own: which purchase intent actually moves this customer? Eikona’s Adaptive Marketing Engine pitted two distinct intents against each other — the bra as a high-performance piece (seamless, moisture-wicking, sewn-in pads) versus an everyday fashion item you can wear around the city, not just to the gym.
For each intent, the system autonomously generated genuinely different creative directions: a product shot surfacing the technical features (“Form that performs”), a UGC-style selfie for authenticity (“elevate every move”), a model in a yoga posture, and a lifestyle scene on a city bike (“let’s get active”). Every variation is system-generated yet production- grade — on-brand, compliant with the brand book, and pixel-perfect on the product, which is never altered.
With every send, reinforcement learning read live performance and shifted delivery toward the winning intent and creative. Measured against the control, the dynamic campaign produced a substantial uplift in revenue and engagement metrics:
+30.9%
RPR
+29.8%
CTR
.avif)
