Personalized homepage recommendations: +136% clicks to product pages
Test run byOptimizelyTesting and CRO tool
Commentary byLiftArchiveOur analysis, not theirsReplacing static homepage product recommendations with personalized ones lifted click-through to product pages by 136%.
Optimizely ran this test for an online retailer and published it as an Optimizely Field Notes use case. Only click-through was reported, not sales.
- +136% click-through to product pages. Far more visitors moved from the homepage into products.
- No figures for add to cart, orders or revenue were shared, so the sales impact is unknown.
The change
The variant swaps static picks for visitor-specific recommendations.
Simplified mockup drawn by LiftArchive from the shared screenshots.
Control
The homepage showed the same static product recommendations to every visitor.
Variant
The homepage showed recommendations tailored to each visitor.
Why it likely won
Our reading of the result, not the practitioner's claim.
- Products matched to a visitor's interests are more likely to earn a click.
- Static picks show the same items to everyone, so they are irrelevant to many visitors.
- Returning visitors may see items they already looked at, which makes it easy to pick up where they left off.
Caveats: Vendor-reported and anonymized, with no sample size, duration or confidence level shared. The 136% figure measures clicks, not purchases.
Who ran it
Optimizely ran this test for an online retailer (Optimizely customer) and published the results (Optimizely Field Notes use case). The numbers are theirs; the summary and commentary on this page are ours.
Read the original case study on Optimizely →
How to run this test on your store
- Hypothesis
- If we show each visitor personalized product recommendations on the homepage, then more will click into products because the items match their interests.
- Primary metric
- Click-through rate to product pages
- Guardrails
- Add-to-cart rate
- Revenue per visitor
- Homepage bounce rate
- Who should test it
- Stores with a broad catalog and enough traffic and browsing data to personalize. Small catalogs or mostly first-time traffic will see less benefit.
Questions
Do personalized product recommendations work on the homepage?
In this test they more than doubled clicks to product pages.
Check that the lift carries through to orders before rolling it out.
What should new visitors see if there is no data to personalize?
Bestsellers are a common fallback. In a related Optimizely test, bestsellers beat featured products for new visitors.
Similar tests
Convertibles
Blend CommerceBlend Commerce
Clean Commit
SurefootClean Commit