The Challenge
Zales wanted to improve product discovery with Shopping Muse, a generative AI assistant built into Dynamic Yield. The tool was designed to guide shoppers through the catalogue using natural language. But without real-time context, it was being shown too broadly.
Every visitor, regardless of their behaviour, was offered assistance. That meant wasted impressions, disrupted journeys, and inconsistent results. Some shoppers benefited, while others saw it as noise.
Zales knew the problem wasn’t the tool. It was the timing. They needed a smarter way to deliver help only when it was needed.
The Hypothesis
By showcasing Dynamic Yield's Shopping Muse functionality at the moment a visitor was displaying signs of struggle in the refining buying stage, we would support their discovery and prevent abandonment.
Rather than offering this specific assistance to every visitor, the team believed they could more effectively support product discovery by identifying who actually needed help. If a visitor was looping through pages or actions, hesitating, or showing signs of struggle, the tool could step in as a guide. If they were progressing with focus, the assistant could stay hidden.
Our research in The èƵ Gap shows that 75% of visitors will signal they are struggling at some stage in their journey, which means there are many opportunities to provide assistance. Targeting these struggle moments would move Zales towards a more intelligent, more respectful support experience.
The Solution
Using real-time intent data from Made With èƵ, Zales focused on visitors in the Build èƵ Struggle & Abandon segments from our framework. We define these as visitors who have engaged with an online store but are showing signs of abandonment or struggle behaviours. These include actions such as back-and-forth clicks, revisiting filters or hovering without progressing.
By identifying these segments in real-time, as well as those in the wider Refining buying stage, the Zales team could target the Shopping Muse experience at the most suitable audience.
When a visitor matched this segment criteria, Shopping Muse was triggered contextually, to offer guidance at a key moment of friction. Visitors outside of these segments weren’t prompted to use the tool. Whether in a focused state or in a later phase of the purchase journey, they were left in the flow of their session.
The Impact
Without èƵ, Shopping Muse appeared by default, leading to inconsistent value, missed opportunities and potential friction.
With èƵ, the chatbot only appeared when shoppers needed it, maximising relevance, increasing engagement and driving measurable outcomes.
Key Results: A 9x increase in Shopping Muse interaction, plus a significant boost in incremental revenue.
Integrating our data with Dynamic Yield allowed Zales to turn a static feature into a responsive, real-time support tool. This Play proved that timing transforms the impact of discovery features. By aligning help with behaviour, Zales made AI feel helpful, not interruptive. It validated the hypothesis that relevance comes not only from what you offer, but when you offer it.
They’re now exploring how similar logic can unlock performance across other tools and journeys, from guided selling to promotional overlays.
Real-time intent data is most powerful when used to support visitors at the right moment. This Play shows how responding this way turns powerful features into perfectly timed experiences.
Want to deliver more intelligent support to your online shoppers? Explore more about Features with èƵ.
