Do star ratings and fewer thumbnails lift a holiday deal page?

For a holiday deal, we tested product page changes that had won in earlier tests: star ratings, no thumbnails, a coupon and a lifestyle image. Over 6 days, conversion fell 3.6% with the best version, short of our confidence bar.

Not significant Responsive 6 days Not called

SURFACE

Product page

DEVICE

Responsive

VERTICAL

Ecommerce (B2C)

PRIMARY METRIC

Conversion rate

VARIATIONS

3

The hypothesis

By applying winning changes from earlier tests to the holiday deal page, we will build shopper confidence, resulting in more purchases.

Earlier tests on the product pages had found wins from removing thumbnails, adding star ratings and using a lifestyle image. The holiday deal page was a short window to put those learnings together.

We also tested a version with less content.

What we changed

Stars, coupon and no thumbnails, plus a lifestyle image or less copy

Var A removed the image thumbnails and added star ratings and a coupon. Var B added a lifestyle image on top of Var A's changes. Var C cut back the page content.

Var A: Stars + Thumbnails + Coupon wireframe — VAR A: STARS + THUMBNAILS + COUPON

Image thumbnails removed
Star ratings added
Coupon added

Var B: + Images wireframe — VAR B: + IMAGES

All Var A changes
Lifestyle image added

Var C: Reduce Content wireframe — VAR C: REDUCE CONTENT

Page content cut back

The result

The best version cut conversion 3.6%, short of our bar

CONVERSION RATE

-3.6%

REVENUE PER VISITOR

-4.3%

VAR C CONVERSION

-27.5%

Over 6 days of Black Friday week, the lifestyle image version (Var B) converted 3.6% worse than the original, and Var A converted 7.1% worse. Neither gap reached our confidence bar. Cutting the content (Var C) lowered conversion 27.5%, a loss that cleared our bar, so we paused it early.

What we would do next

Retest these changes outside a short holiday window

Changes that won on the regular product page did not carry a clear lift to the deal page. Shoppers did convert better with the full content, since cutting it lost. We would retest the earlier winners together over at least 14 days.