Personalization Testing for Online Stores

We tested static vs. personalized cart messages. Here’s what happened. Our hypothesis was simple: If we personalize cheering messages based on cart items, users will feel more connected to their purchase, leading to higher engagement and ultimately, higher revenue. Why? → Personalization Principle: When users feel directly addressed, they engage more. → Perceived Value Theory: A tailored message makes products feel more valuable. → Emotional Triggers: Subtle psychological nudges reduce hesitation and drive action. Here’s how we ran the test: Control → Static green cheering message. Variant → Mood image + personalized message directly linked to cart items. And the results? 📈 +0.54% ARPU 📈 +0.73% Conversion Rate Why did this work? Personalized messaging transforms a standard checkout process into an experience where users feel seen and valued. By aligning messages with their cart contents, we created a sense of intentionality, reinforcing their decision to buy. This wasn’t just about optimizing numbers. It was about strengthening the connection between user and product. Every experiment teaches us something new. This one proved that even small, thoughtful tweaks can drive measurable impact. What’s your take on personalization in CRO? Have you tested similar strategies? #cro #abtesting #ecommerce #experimentation

Data is power in DTC. How Le Creuset uses personalization: 1) Recognize more of your site visitors → Use identity resolution to convert anonymous traffic to known. Personalized intent-based popups perform well. Le Creuset increased their daily subscriber signups by 104%. Intent based popups work. 2) Capture zero and first-party data at every opportunity Make sure you consolidate your data across: SMS Email Pop-ups Retargeting A “personalized” experience that feels disconnected can be worse than a generic experience. 3. Activate the data you've captured → Test 1:1 on-page personalization  → Personalize your retargeting Le Creuset saw strong CVR improvement using this simple framework: - 2X triggered email revenue - 60% increase in first-purchase conversion But... Do you know what impacts all of the above? Your site speed. If your site isn’t fast. Your personalization won’t last. Because people will bounce before it triggers. Your site speed silently shapes your Shopify sales. Great CVR experiments are powered by speed. Remember that.

Advanced personalization work involves 'growth engineering' as a new 'role' to connect the dots and architect the data, from tools/data like: • CDPs • Data warehouses • Testing tools that enable adaptive approaches, e.g., Contextual Multi-Arm Bandits (or similar) • And advanced 'intent' or 'propensity' data and models. The last point is where Mr. David Mannheim comes in. He just pushed out a cool Ecom report on intent. (check it here, https://lnkd.in/gFcB-s7f) Whats in there are concepts, vocabulary, taxonomy that influences the last point of propensity data. Things like (from the top of the report): - 63% feel manipulated by ecommerce tactics (only 11% don’t) - 46% feel overwhelmed on ecommerce sites - 83% use discount codes when they would have bought at full price - 1 in 5 will stop shopping if they get an early pop-up The TRICK is to get this data accessible to the testing and engagement platform setup. Feature attribute data: > CDP defined User-level attributes: account tier, number of past upgrades purchased, engagement metrics (time on site, feature usage). > Session-level attributes: current time of day, day of the week, user’s device type, current navigation path or product page. > External attributes (optional): Geo-location, known seasonal promotions, pre-determined propensity model data All this sounds cool, but WHY/WHERE to apply this stuff? Here's my thinking: > Adaptive Learning: A dynamic personalization approach continuously updates the probability distribution of reward for (offer/product/promo) as new data is collected. Unlike a static A/B test, it doesn’t wait for a full experiment cycle to end before updating which offer to show next. (we don't care what wins, just push to what is working best now) > Context Utilization: This setup leverages user and environmental context (e.g., user account age, user’s current usage tier, time of week, location). This allows for personalized experiences rather than one-size-fits-all solutions. Add in explicit propensity and 'intent' data (h/t to David here) and you really get cooking. > Handling Concept Drift: If certain upgrades become more or less attractive over time, the testing/personalization algorithm automatically adjusts. This adaptability ensures that the system remains optimal in the face of changing market or user conditions. Yes this is where AI experimentation tools come into play, but the foundation of tooling and explicit data ontology (use case and model connections) needs to be there first. A personalization (also AB testing) recipe is only as good as it's ingredients. Bottom line? The right data, connected smartly, powers personalization that actually works—and keeps evolving. Want to dig deeper into David’s intent report, architect your own growth engineering setup, or just swap ideas on making this real for your team? I’m all ears—DM me or drop a comment below. Let’s cook up something impactful together!