# Default Review Sort Order A/B Test on Product Pages

Page: https://alidemirbas.com.tr/en/lab/ab-test-library/tests/default-review-sort-order-ab-test-on-product-pages

## Hypothesis

With newest as the default, visitors first see recent experiences; with most helpful, they first see the reviews with the most helpfulness votes. Whichever first view helps the buying decision more could raise purchase conversion rate (CR) more under that default.

## When to run this test

Sessions that stop scrolling after the first few reviews, short or old reviews sitting at the top, or questions that reviews already answer turning into support tickets mean the first reviews visitors see do not help them decide. If one default puts more useful reviews in the first positions, that shows up in purchases. If a product simply has few reviews, sorting will not fix that; before launch, also confirm helpfulness votes cannot be manipulated and that each sorting option has a clear label.

## Primary metric

**Purchase conversion rate (CR)**: Compare purchases per assigned visitor under the two review-sorting defaults.

## Other metrics

- **Review visibility depth**: Count reviews entering the viewport; visibility does not prove they were read.
- **Sort change rate**: Track manual changes without treating each change as dissatisfaction.
- **Return rate**: Track returns after the same follow-up period in both arms.
- **Exit rate after review exposure**: Review exits in sessions exposed to reviews; do not use this subset instead of the outcome for all assigned visitors.

## What to check before and during the test

- **Default sort**: Set either newest or most helpful as the default.
- **Review mix**: Check which ratings appear first without removing negative reviews.
- **Visitor control**: Provide the same sorting options in both arms.
- **Visitor history**: Review new and returning visitors separately.
- **Review volume**: Review predefined review-count groups, including products with few reviews.

## What to watch when reading the result

- Compare the defaults on purchase conversion rate across all assigned visitors; a conversion rise that comes with a higher return rate over the follow-up period is not a gain. Exit rate after review exposure covers only sessions that reached the reviews, and visibility depth does not prove anything was read. Treat both as diagnostics only.
- This test changes only the default order. Keep the review-card design and the way reviews are collected the same throughout, or you cannot tell what caused the difference. Read new and returning visitors separately, along with review-count groups that include products with few reviews.
- Do not read sort changes as dissatisfaction; a visitor may simply want another view. For a small conversion gap, put each arm's visitors and purchases into the [A/B test significance calculator](https://alidemirbas.com.tr/en/calculators/ab-test), and if the gap is not significant, note that this sample could not tell the two defaults apart.

## Frequently Asked Questions

### Can users change review sorting themselves?

Yes. Keep the same choices and review set in both arms. Only the default changes, and the labels must remain understandable.

### Does manually changing the sort prove the default was wrong?

No. Treat it as diagnostic. Use purchase conversion as the primary outcome, and do not equate visible review depth with reading.

### Can most-helpful sorting push negative reviews down?

Yes. It does not guarantee balance. Inspect the initial rating mix without removing negative reviews, and protect helpfulness voting from manipulation.

### How should exits from the review section be interpreted?

That subgroup includes only visitors reaching reviews. Do not substitute it for the assigned-population result. Inspect predefined experience and review-count groups.