# What to A/B Test: 45,000 Real Store Tests, With Examples

Source: https://trymimetic.com/guides/what-to-ab-test

[Guides](/guides) / Revenue QA

# What to A/B test on an online store

Most "what to A/B test" lists are the same twenty ideas copied from one blog to the next. This one comes from what online stores actually test: more than 45,000 A/B tests run by 3,380 stores over the last six months, sorted by what each test changed.

The numbers below are the share of stores that tested each area at least once in that time. They tell you where stores spend their testing effort, which is a good map of where the questions are.

## The eight things stores test most

| What they test | Stores that tested it |
| --- | --- |
| Bundles, upsells and cross-sells | 29% |
| Product description and page content | 29% |
| Homepage hero | 27% |
| Collection pages: filters and product cards | 27% |
| The buy box: add-to-cart and option pickers | 26% |
| Navigation and menus | 24% |
| Product images and gallery | 21% |
| Reviews and social proof | 21% |

### 1. Bundles, upsells and cross-sells

The most widely tested area, and it isn't just on product pages: about half of these tests run on the product page, a quarter in the cart, and some in checkout. Real examples:

- Adding "frequently bought together" picks to the cart
- Swapping a carousel of add-ons for a single add-on offer in the "added to cart" pop-up
- Showing best-selling add-ons during checkout

### 2. Product description and page content

Almost all on the product page, and mostly about how information is organised, not what it says:

- Reordering product FAQ tabs and opening the first one by default
- Adding a comparison table
- Testing a shorter, simpler product page layout

### 3. The homepage hero

- A background video against a still image
- Different wording on the main button
- Reordering the homepage sections

### 4. Collection pages

The area big stores test most heavily: 45% of the largest stores tested their collection pages, against 16% of smaller ones.

- Filters in a sidebar against a filter bar across the top
- A filter bar that stays visible while scrolling
- Switching between on-model photos and product-only photos

### 5. The buy box

The add-to-cart button and everything around it:

- Putting the selected size in the add-to-cart button text
- Moving option pickers so they sit above the button
- Changing colour without reloading the page

### 6. Navigation and menus

- A mobile menu with shopping tabs and product cards
- Seasonal links added to the menu
- A new desktop menu layout

### 7. Product images and gallery

- Making a video the second item in the gallery
- Moving thumbnails to the side on mobile
- Testing a different main image

### 8. Reviews and social proof

- Moving reviews higher on the product page
- Showing the rating as a number, not just stars
- Adding a review carousel to collection pages

## What big stores test that smaller stores don't

The largest stores spend much more of their testing on the plumbing of the store:

| Area | Largest stores | Smaller stores |
| --- | --- | --- |
| Collection pages | 45% | 16% |
| Search | 19% | 2% |
| Checkout | 27% | 8% |
| Navigation | 37% | 15% |

Smaller stores lean toward ad landing pages, whole-theme swaps and price tests. Both make sense. A small store's biggest question is often "does this page convert at all?", while a big store already has a page that works and is tuning how people find and buy across thousands of products.

## How it differs by kind of store

- **Supplements and health stores test subscriptions about four times as often** as other stores: 30% against 8%. Food, drink and pet stores also test subscriptions more than most.
- **Stores selling things people use up test ad landing pages much more.** That's 43% of supplement stores, 46% of pet stores and 37% of food and drink stores, against about a quarter of everyone else.
- **Apparel stores test collection pages more** (35% against 25%) and headline and copy changes about half as often (11% against 20%).
- **Supplement stores test discount and offer wording more:** 32% against 20%.

## The one change stores most often keep

Free-shipping changes are the ones stores most often leave running after a test: a different free-shipping threshold, a progress bar in the cart showing how close the shopper is, or a message when they reach it. If you're not sure where to start, a free-shipping test is a common first step with a clear question behind it.

## How to pick your next test

1. **Start where shoppers drop, not where ideas are fun.** Look at where people leave: product page to cart, cart to checkout, checkout to order. Break it down by phone and desktop, and by browser.
2. **Fix what's broken before you test.** A test can't tell you much if the page has a bug for some shoppers, like a buy button that doesn't respond in one browser or a shipping cost that only appears at the last step. Fixing those isn't a test, it's a fix.
3. **Pick the busiest step you can change.** A small improvement on a page most shoppers see beats a big improvement on a page few see.
4. **Make sure you have the traffic.** A test needs enough orders in each version to tell a real difference from noise. If you don't have the traffic, make the change and compare before and after, and say plainly that it isn't a test.
5. **Decide the measure before you start,** and read revenue and order value, not just conversion rate. A change can lift conversion while shrinking the average order.

## Testing on Shopify

Shopify has its own A/B testing built in, called Rollouts. See [how Shopify Rollouts works](/guides/shopify-rollouts), and [Shopify A/B testing](/guides/shopify-ab-testing) for what Shopify stores test and how to choose a tool.

## Find what to fix first

Mimetic checks your store the way shoppers use it, across phones and browsers, and finds what's costing you sales. That's the list to test, or simply fix, first.

> [Check your store](/) to see what's worth testing.

## Common questions

**What should I A/B test first on my store?**
The busiest step where shoppers drop off, usually the product page or the cart. Check that nothing is broken there first, then test the change most likely to answer a real question.

**What are good A/B test examples for ecommerce?**
The most common are bundles and upsells, product page content, the homepage hero, collection page filters, the add-to-cart area, navigation, product images and reviews. Each section above has real examples.

**How long should an A/B test run?**
Until each version has enough orders to show a real difference, and for at least one full week so every day of the week is covered. Don't stop early because one version looks ahead.

**Can I A/B test with low traffic?**
Only big changes, and slowly. With little traffic, it usually makes more sense to fix clear problems and compare before and after.

Written by Coleman Hindes, founder, Mimetic.
Published 2026-11-18.

### See what is costing your store sales.

[Scan my site](/)
