Introducing a robust product recommendations strategy to help clients with discovering items
Industry
eCommerce
My role
UX/UI
Campaign duration
26 October 26th 2021 – 30 November 2021 (35 days)
KPI
The success of this campaign was based on by measuring:
- Clicks on product recommendations
- Add to cart
- Bounce rate
- Overall revenue
I identified a potential optimisation opportunity on product detail pages: while recommendations positively impact add-to-cart and revenue, the current 100% display rate lacked strategic variation.
To address this, I proposed testing different recommendation strategies, starting with the following:
- Variant A. PDP — Brand / Recipient / In Stock / Price
- Variant B. PDP — Brand / Recipient / In Stock / Price / Strap
The purpose
The purpose of this 50/50 A/B UX test is due to the fact I’ve identified a significant opportunity to boost product detail page performance. Simply showing them to everyone isn’t maximising their potential. By testing different recommendation strategies as mentioned above, I aimed to uncover the most effective approach, ultimately creating a more personalised shopping experience that translates into higher conversion rates and increased revenue.
The outcome
The campaign was served to 449,255 sessions, 72.1% of whom were on mobile devices.
Variant B (which showed recommendations with a similar strap style) had a statistically significant uplift in add to cart +22.7%.
Variant B (with similar strap) had a significant uplift in clicks by +15.4%.
Variant B had a reduction in bounce rate by 1.5%.
The uplift in add to bag translated into overall revenue where variant B had an uplift of +24.18% totalling £492,485. Compared to variant A at £396,583.
Design
Desktop & mobile
Products shown below were dependent on their recommendation strategy (strap type), for example.
Conclusion
Short summary
Variant B showed that visitors were more likely to click on recommended products, and add them to their basket when the strap type was the same as the product that they were viewing.
If the winning variant was served to 100% of users there would be roughly a further addition of 107 items added to cart over a 30 day period. I therefore started to serve Variant B to 100% of applicable users.
To maximize recommendation effectiveness, I proposed a strategic focus on attribute-driven personalisation. This plan included rolling out tailored solutions across all brands, including Goldsmiths, Mappin & Webb, Mayors, and Watches of Switzerland, considering their distinct product focuses — such as jewellery (ring, bracelets, necklaces etc).