# Photo Reviews vs Text-Only Reviews A/B Test Example

Page: https://alidemirbas.com.tr/en/lab/ab-test-library/tests/photo-reviews-vs-text-only-reviews-ab-test-example

## Hypothesis

If reviews show customer photos, visitors can see how the product looks in real use before they buy and set their expectations to match. The return rate for orders past the follow-up period could fall as a result.

## When to run this test

Return reasons saying the item looked different, support tickets asking about color or texture, or the same complaint repeated across reviews show a gap between studio images and the real product. If customer photos close that gap, buyers in the variant arrive with more accurate expectations and later returns go down. If returns come from shipping damage or sizing problems, look at those first, because photos will not fix them.

## Primary metric

**Return rate**: Compare returns for orders that have completed the same follow-up period in each arm, alongside conversion and the mix of products purchased.

## Other metrics

- **Purchase conversion rate (CR)**: Assess purchases among all assigned visitors, not only those who open a photo.
- **Review section interaction rate**: Use photo opens and review interactions as diagnostics; those controls appear only in the variant.
- **Page weight and load time**: Make sure images do not slow the mobile experience.
- **Moderation workload**: Check that the volume of inappropriate content remains manageable.

## What to check before and during the test

- **Photos**: Show permitted customer photos within the same eligible reviews.
- **Format**: Use one fixed photo presentation and keep review order unchanged.
- **Collection**: Keep the review request process unchanged; prompting for new photo submissions is a separate test.
- **Image set**: Set eligibility rules and a maximum number of displayed photos.
- **Category**: Review relevant categories separately.

## What to watch when reading the result

- Measure returns across all assigned visitors, not only those who open a photo, and read them alongside conversion and the mix of products purchased. If returns fall while conversion falls too, the photos may have changed who buys rather than what buyers expect, so that alone is no gain.
- This test speaks for the categories you tried and this photo eligibility rule; read relevant categories separately and test again before extending it to other categories. Check that the photo strip does not push the CTA down the page and that page weight and mobile load time hold up, and confirm the moderation workload stays manageable.
- Present customer photos so they cannot be mistaken for product photography, do not cherry-pick only positive ones, and do not enlarge low-resolution images. If any of these rules is broken, the return gap reflects how the photos were chosen and shown, not the photos themselves.
- Returns arrive some time after the order, so do not decide on orders that have not completed the follow-up period. With a small or unclear difference, recalculate the orders each arm needs by entering the return rate in the [sample size calculator](https://alidemirbas.com.tr/en/calculators/sample-size-calculator). Update the plan by adding the return window to the time it takes to collect those orders.

## Frequently Asked Questions

### Do fewer returns among photo viewers establish an effect?

No. Viewers are self-selected. Compare assigned arms and interpret matured return rates with purchasing and product mix rather than restricting analysis to photo openers.

### Which customer photos should appear?

Fix eligibility and the maximum count. Do not select only flattering images or enlarge poor-resolution files. Keep customer-review images distinguishable from official product photos.

### What if the photos make the page heavier?

Monitor page weight and mobile loading. Preserve CTA access and verify that moderation can handle the actual content flow.