If you approach searchandising with a “set it and forget it” mentality, you may be leaving revenue on the table. You have the power to fine-tune the AI-powered algorithms that optimize the sort order of your products and add rules that apply to your specific business and customer context.
Don’t forget the influence your facets and in-grid UI components have on customer experience as well. A/B testing can help you make data-driven decisions on all of the above. Let’s explore the top five testing strategies of Browse and Collection pages.
Searchandising Rules
A/B testing use cases for searchandising rules are as broad as your imagination — just keep in mind that not all rule conditions are significant enough to provide an A/B test lift.
To see the biggest revenue lifts, try testing more complex ‘recipes’ of search rules against your existing rules, rather than testing rule-by-rule.
Default sort
Look for opportunities where a new rule set may outperform your existing rules (i.e., Sale and Clearance pages for a fashion catalog).
Is showing newer items first more appealing to your customer base than deeper discounts? Is your goal to turn over aging items more quickly or drive more daily revenue?
Sale collection results ranked by newest
Newer items feel more in-season while deeper discounts feel like a score.
Sale collection results ranked by percent off
Remember, A/B tests thrive on sample sizes. So, choose your testing strategies carefully. Global rules will get you significant results faster than testing a single page unless it consistently receives a lot of traffic per week.
Facets
Filters are powerful tools to help shoppers customize their results, helping them make faster buying decisions — but they’re often difficult to use, especially on mobile.
The more granular your facet options, the more complicated this can be.
Facet rank
Optimizing facet rank can be achieved through your search and discovery tool when machine learning is enabled to re-rank based on user behavior.
Facet visibility
Also, consider which facet values to display open versus closed by default.
A/B testing lets you challenge your existing approach with alternatives. For example, on a prominent U.S. sporting goods site, only Brand is expanded by default when you open the filter.
Sporting goods site’s filter menu with Top Brands facet open by default (Live)
You may hypothesize that product type would do a better job guiding users to a focused result set than brand.
A proposed alternative with Product Type facet by default
Or, you may hypothesize that a fully collapsed menu reduces visual clutter and streamlines the experience for better usability for most users.
A proposed alternative with all facets collapsed
Consider running such A/B experiments across clusters of browse pages with similar attributes. For example, top-level department categories return more results than sub-categories and may benefit more from visible Product Type facets.
Besides testing facet visibility, you might also want to A/B test a fixed sort order against dynamic reranking if your search application offers this capability.
Facet slotting and merging
With Constructor, you can also test slotting (pinning facets to specific spots) and merging facets to reduce complexity (i.e., for colors, size scales, and synonymous tags).
Attribute badges
Highlighting useful product attributes within card details can help shoppers better understand their features and benefits without having to pogo-stick back and forth from product list to product pages.
Showing the right product attributes can also help ‘pre-sell’ shoppers pre-click to increase add-to-cart rates.
Product listings with attribute badges
Product attributes may also underperform due to visual clutter and confusion. Consider A/B testing product listings without attribute detail and testing with different UI treatments.
Next Steps
Your browse and collection pages are more than just lists of products. They're opportunities to guide customers to the right purchase decisions.
Start with the tests that promise the biggest impact, choose high-trafficked pages for these tests, and commit to a rigorous, iterative testing program for continuous learning.
For more A/B testing insights and strategies from the Constructor experimentation team, check out the Experiments Blog.