Retail_
18%
Conversion lift
Web and app combined
+24%
Basket size
Personalized cohort
+38%
Email CTR
Versus control
2.8x
First-year ROI
Payback in 6 months
The challenge
ONE-SIZE EMAILS
Aurora's CRM stack was mature but generic. Every loyalty member received nearly the same cadence, offer mix, and homepage layout. Conversion had plateaued for six quarters, open rates drifted down to 7 percent, and store teams saw little impact on foot traffic from campaigns.
The CMO wanted AI for retail that protected the brand and store experience, not a black box that sprayed offers without context.
18M
Loyalty members
Plateau
6 quarters
7%
Email open rate
The approach
FOUR PHASES
We delivered AI for retail stores in four phases so each release drove clear lift, kept merchandisers in control, and made AI agents for retail feel like a copilot rather than a replacement.
Phase 1
Wks 1–6Behavioral Foundation
We established the behavioral data foundation needed for robust AI for retail personalization across channels and stores.
Cleaned and unified Segment events across ecommerce, app, and stores
Defined 14 lifecycle segments tied to real retail behaviors
Built feature store for on-site, in-app, and in-store recommendations
Ran a segmentation audit to show where AI for retail could move the needle fastest
Phase 2
Wks 7–14Recommendation Engine
This phase delivered recommendation quality with guardrails so merchandising teams retained full control over brand and inventory rules.
Two-tower recommendation model for product affinity and cold-start shoppers
Guardrails for brand, pricing, and inventory so AI solutions for retail never break merchandising rules
Merch review loop where buyers could approve, block, or boost patterns
A/B testing harness to compare AI journeys to legacy campaigns
Phase 3
Wks 15–22Channel Activation
We activated personalized journeys across every major customer touchpoint and orchestrated actions with retail AI agents.
Live recs in email, SMS, web homepage, app, and in-store kiosks
AI agents for retail that select the right channel and next best action per customer
Braze integration for triggered campaigns based on behavioral signals
Web and app components for personalized rows, banners, and search results
Phase 4
Wks 23–36Optimization Loop
Post-launch optimization tied personalization performance directly to revenue, margin, and long-term customer value.
Weekly model refresh and creative review with merchandising and brand
Safety checks for discount depth, category mix, and compliance requirements
CMO dashboard tying AI for retail stores impact to revenue, margin, and LTV
Attribution model that credits AI-driven journeys across online and store visits
"
alien.fi understood that personalization in retail is about brand as much as clicks. Their AI for retail approach gave us guardrails our merchants actually trusted, and AI agents for retail that our teams could reason about. The lift followed.
EH
Elaine Hwang
Chief Marketing Officer, Aurora Retail Group
More like this
RELATED STUDIES