How Maëlys More Than Doubled RPR, and Automated Its Creative Process — With Eikona
+54.7%
Revenue per recipient
(up to +490% in best campaign)
+104%
Click-through rate
(up to +119% in best campaign)
Approach
Maëlys layered Eikona's adaptive marketing engine onto its existing Klaviyo stack - integration took minutes, nothing was migrated - and the sending model changed overnight: single blasts gave way to portfolios of calculated variations - grounded in past performance, exploring new intents and messages - steered by reinforcement learning and measured against a randomized control. An engine trained on the brand: fresh off a substantial brand uplift, Maëlys needed every output 100% on-brand, so Eikona's creative pipeline adapted to the partner's brand - here, tuning its AI models on the new guidelines and assets - with continuous feedback from the brand and creative team sharpening the engine until its emails were indistinguishable from the in-house team's work. Calculated exploration, not blasts: every campaign ships as up to 20 production-grade variations, each probing a different benefit, message, or visual - with traffic shifting toward winners in real time - and Eikona's Adaptive Marketing Engine gives each user the creative their first-party data and past reactions point to, the logic acquisition channels have always run on, while send frequency adapts per segment based on response to volume, trimming sends by over 10% while lifting revenue and deliverability. Learning that travels: each test feeds the next, so knowledge accumulates instead of resetting with every campaign, and creative analytics explain why winners won - those concepts moving into paid, social, and beyond.
01
Business Challenge
Swap guesswork and occasional A/B testing for always-on testing with measurable, repeatable results against a randomized control.
Use AI to scale creative testing - without compromising the brand or the product.
Right-size frequency, sending each person the amount of email that works for them.
More than doubling RPR is the kind of number you double-check - and it held, against a randomized control. Just as important: we came out of a rebrand our team fought hard for, so 'AI-generated' was a scary word, yet Eikona's emails became indistinguishable from ours. This is truly the future of retention and CRM marketing.
Key Takeaways
Adaptive creative and reinforcement learning turned Maëlys' email program into a source of both revenue and understanding - every send doubles as a test, and every test sharpens the next.
01
Measured, Not Marketed
(up to +490% in best campaign)
02
On-Brand Is the Last Mile
Out-of-the-box models don't clear a brand team's bar; it took a vertical AI stack tuned on Maëlys' guidelines to make outputs indistinguishable from in-house work.
03
From Winners to Understanding
Users get the creative they respond to, and the reasons behind each win become playbook material for every channel.
Campaign Example 1
In this campaign, Eikona transformed a single GET-DREAMY send into a portfolio of distinct creative directions, each built around a different reason to believe in the product. The system explored overnight firming and smoothing, confidence, consistency, and the idea of a nightly body-care ritual.
Each concept was translated into a production-grade execution using Maëlys' existing brand language, product assets, and real model photography. No models were generated or altered with AI, allowing Eikona to expand creative exploration while staying fully aligned with the brand's visual identity and creative decisions.
Reinforcement learning measured how customers responded to each direction and progressively shifted traffic toward the creative most likely to drive engagement and revenue.
+27.8%
OR
+204.6%
CTR
Campaign Example 2
In this campaign, Eikona tested which creative language would communicate the same offer most effectively. The promotion remained fixed, while the engine explored multiple ways to frame it - from bold discount-led messaging to lifestyle imagery, product-led storytelling, and benefit-focused creative centered on 24/7 moisture.
Every model featured in the experiment came from Maëlys' existing photography and was not generated or altered with AI. Eikona worked within the brand's established visual language, using the client's own assets to create new, performance-ready compositions.
Each variation preserved consistency across typography, color, product presentation, tone, and claims. Reinforcement learning then shifted traffic toward the creative framing most likely to convert.
+407.9%
RPR
+158.2%
CTR
.avif)
