Filters on collection pages: -6.4% conversion, -33.2% revenue/visitor
Test run byClean CommitCRO agency
Commentary byLiftArchiveOur analysis, not theirsAdding a filter panel to collection pages lowered conversion rate by 6.4% and revenue per visitor by 33.2%.
Clean Commit ran this test for an anonymized Shopify store and published it in its A/B test library.
Shoppers browsed more and added to cart more, but the drop in order value, mostly on mobile, outweighed it.
- Revenue per visitor: -33.2%, driven by average order value falling 29.2%.
- Conversion rate: -6.4%, and orders: -5.3%. The conversion drop is close to a coin flip, with a 52.4% probability the variant was actually better.
- Mobile average order value fell 40.4%, the main source of the revenue loss.
- Add-to-cart rate: +4.8%, and bounce rate: -40.5%. Shoppers engaged more without buying more.
The change
The variant adds an attribute filter panel above or beside the product grid.
Simplified mockup drawn by LiftArchive from the shared screenshots.
Control
Collection pages had no filtering.
Variant
A filter panel let shoppers narrow products by style, color style, text layout, text arch and city. It sat on the left on desktop and in a collapsible top drawer on mobile, allowed several values per filter, and remembered choices.
Why it likely lost
Our reading of the result, not the practitioner's claim.
- Filters may have pushed shoppers to narrow down to one exact item, rather than browsing and adding several.
- On mobile, a drawer that changes the listing can hide products that would have been picked up while scrolling.
- With a small sample, a few large orders in the control group could explain much of the order value gap.
Caveats: About 5,196 mobile visitors were noted and duration was not shared. The conversion drop was near a coin flip, and the revenue loss may be partly due to a few large orders.
Who ran it
Clean Commit ran this test for a Shopify store (anonymized) and published the results (Clean Commit A/B test library). The numbers are theirs; the summary and commentary on this page are ours.
Read the original case study on Clean Commit →
How to run this test on your store
- Hypothesis
- If we let shoppers filter collections by the attributes they care about, then more will buy because they reach the right product faster.
- Primary metric
- Revenue per visitor
- Guardrails
- Average order value
- Conversion rate
- Mobile revenue per visitor
- Who should test it
- Stores with large collections where products differ on clear attributes. If your collections are small or shoppers tend to buy several items, filters may cut browsing and order size, as here.
Questions
Do filters on collection pages increase sales?
Not always. In this test they raised add-to-cart rate and cut bounce, but revenue per visitor fell 33.2% because orders got smaller.
Why can filters lower average order value?
Filters help people find one specific item fast.
That can reduce the browsing that leads shoppers to add a second or third product.
What should I measure in a filter test?
Use revenue per visitor as the main metric and check order value by device.
Engagement metrics like bounce rate can improve while sales fall.
Similar tests
@oliverkenyon@oliverkenyon
ConvertiblesConvertiblesConvertiblesConvertibles