Ocado delivers
fresh experiences

Introduction
Ocado, the world’s largest online grocery, operates three marquis brands in addition to their flagship supermarket. To boost conversion rate and average order values across Fetch, Fabled and Sizzle, Ocado used Dynamic Yield to personalize recommendations and send individualized messaging based on user behavior. Ultimately, the retailer drove 8X more conversions from recommendations while also boosting high-value orders by more than 20%!

Dynamic Yield has made our marketing team a lean, mean machine, allowing us to seamlessly test and deploy unique variations and strategies. We can quickly analyze our data in real-time and craft experiences that significantly improve our core KPIs.
The Challenge
Across its portfolio of lifestyle brands, Ocado oversees a vast product set targeted towards a plethora of consumer profiles.
To delight and monetize all of these personas, Ocado required a solution to
- Test recommendations strategies on product pages to drive more conversions
- Encourage serendipity by exposing users to more relevant items
- Increase average order value across its properties
- Boost conversion rates from different traffic sources
Execution
PDP recommendations in particular demonstrated a massive uplift, delivering 8X more conversions than a control group in which no recommendations were shown. The Homepage recommendations in which Ocado tests “Free Gift” recommendations vs. “Newest is” has demonstrated a 54% uplift in CTR.


When a user first arrived on site, an overlay informed her that she would receive a £10 voucher if she spent £75 or more. As she navigated the site, various versions of that message would appear based on items in cart and onsite behavior. The highly targeted nudges paid off, resulting in 20% more baskets stocked with more than £75 of merchandise.



After launching Dynamic Yield’s real-time social proof capabilities, which highlights the number of views and purchases every high-demand product elicits, Ocado saw a 9.8% uplift in revenue per user.
