# Filters Visible or Behind a Button: A/B Test Example

Page: https://alidemirbas.com.tr/en/lab/ab-test-library/tests/filters-visible-or-behind-a-button-ab-test-example

## Hypothesis

When filters are visible on the page, visitors see their options right away and may narrow the listing to a suitable product more easily. Keeping them behind a button leaves more room for products, so which layout lifts purchase conversion rate more depends on how those two effects balance out.

## When to run this test

Prioritize this test when filters and products seem to compete for space on your listings, for example when few products fit on the first screen, heatmaps show filters being passed over, or session recordings show visitors hunting for them. If visible filters win, more visitors narrow the list and reach a product page; if the button layout wins, the extra product visibility should show up as fewer listing exits. If visitors find the filters but dislike what they return, review the filter options and labels first, and keep the number of filters the same in this test.

## Primary metric

**Purchase conversion rate (CR)**: Compare purchases among all eligible assigned visitors who can see the filter controls.

## Other metrics

- **Filter usage rate**: Compare eligible visitors who apply a filter.
- **Product page visit rate**: Measure product-page visits per assigned visitor over the same follow-up window.
- **Products visible in the list**: Compare the number of distinct products actually visible under the same screen and list conditions.
- **Listing exit rate**: Compare exits among assigned visitors who are shown the product listing.

## What to check before and during the test

- **Visibility**: Show the same filter set by default in one arm and behind a clearly labeled button in the other.
- **Opening**: Report panel opens in the collapsed arm as a diagnostic measure.
- **Scope**: Keep the visible filter set the same across both arms.
- **Screen space**: Measure actual product visibility in both layouts.
- **Device**: Compare arms within each device type before considering device-specific defaults.

## What to watch when reading the result

- Filter usage rate will likely be higher when filters are visible, and that alone does not make a winner. Judge the layouts on purchase conversion rate and product page visit rate among all eligible assigned visitors, including those who never filter; products visible in the list and listing exit rate explain why a gap appears.
- In the collapsed arm, make sure the indicator showing how many filters are applied is always visible. Without it, visitors may not notice the list has been narrowed, and the two arms are no longer compared on fair terms.
- Keep the visible filters the same across categories for the whole test, or you cannot tell which difference came from visibility. Your finding covers the categories and devices you measured; test again before applying it to categories with a very different filter structure.
- A small overall gap calls for comparing the arms within each device first, since opposite effects on mobile and desktop can cancel out. For the overall result, enter each layout's visitors and conversions into the [A/B test significance calculator](https://alidemirbas.com.tr/en/calculators/ab-test); if the gap is not significant, do not crown either layout, but do not read that as proof the two are equal either.

## Frequently Asked Questions

### Why do visitors who never filter remain in the denominator?

Visibility changes who uses filters. Restricting analysis to filter users would bias the comparison. Compare purchases among all eligible assigned visitors within the common window; product-page visits show where a gap opens.

### How can I assess the extra space gained by hiding filters?

Compare distinct products actually visible under equivalent screen conditions, then product-page visits and purchases. List exits are diagnostic. Keep a clear filter-button label and applied-filter count.

### Can the result be applied to other categories?

Do not generalize to categories with very different filters. Compare variants within each device before considering device-specific defaults. Keep filter count fixed.