Ecommerce conversion rate optimization (CRO) is the work of getting more of the people already on your site to do what you need: add to cart, finish checkout, spend more per order or come back and buy again. Good CRO starts by choosing which of those numbers to move. The test comes second.
Why most ecommerce tests fail before they start
Most tests fail because nobody decided what they were for. Testing tools are cheap and every store has one. The ideas are what most brands are missing.
The usual starting point is a meeting where someone says “let’s test the button color.” Nobody asks which number should move, or why a visitor would buy more because a button turned orange. The test runs, the result is flat, and the team concludes CRO does not work.
The fix is one question before every test: what are we trying to change? For an ecommerce brand there are four real answers.
Pick one of four goals
Each goal lives on a different page and has its own number. A test that helps one can hurt another, so pick one.
Ecommerce A/B testing ideas by goal
| Goal | The number | Test ideas |
|---|---|---|
| More visitors adding to cart | Add to cart rate | Price and what it includes above the fold, shipping cost on the page, photos showing size and use, reviews by the button |
| More carts reaching a finished order | Checkout completion | Shipping cost shown earlier, guest checkout, fewer form fields |
| Bigger carts | Average order value | Free-shipping threshold above today's average order, bundles, a complete-the-set offer |
| More value from customers you already have | Repeat purchase rate, 180-day revenue per customer | Post-purchase offer, reorder reminder timing, subscription offer after the second order |
More visitors adding to cart
The number is add to cart rate on product pages. If it is low, the visitor is not convinced. Test the things that convince: the price and what it includes above the fold, shipping cost shown on the product page, photos that show size and use, and reviews next to the button.
More carts reaching a finished order
The number is checkout completion. Baymard Institute averages 50 studies to a 70.22% cart abandonment rate. Among US shoppers who left during checkout for a reason other than browsing, 40% said “extra costs too high” and 18% said “the site wanted me to create an account.” Test showing shipping costs earlier, guest checkout and fewer form fields before you touch a single color.
Bigger carts
The number is average order value, the average amount spent per order. Test a free-shipping threshold set a little above today’s average order, bundles of products people already buy together and a “complete the set” offer on the cart page.
More value from customers you already have
The number is repeat purchase rate or revenue per customer over 180 days. These tests live after the first order: the post-purchase page, the confirmation email, the timing of a reorder reminder. Our guide to retention marketing covers that side in full.
Why color tests are the stupid test
A button color test has no theory of why people do not buy. If shoppers can see the button, its color is not what stops them. Price, shipping cost, trust and doubt about the product stop them.
Color tests also cost more than they look. Every test needs traffic to reach a trustworthy answer, and most stores only have enough for a handful of real tests a year. Spending one of them on orange versus green is spending money you will not get back.
The same goes for most “best practice” tweaks copied from a blog. If you cannot say which of the four numbers a test should move and why, do not run it.
How to run a test you can trust
- Write it down first. The goal, the number, the change and why you expect it to work.
- Fix the length before launch. Use a sample size calculator, then run full weeks so weekdays and weekends both count. Stopping the moment a result looks good is how false wins happen.
- Change one thing, or list every change in the bundle so you know what you tested.
- Check the next step down the funnel. A test that lifts add to cart but lowers checkout completion is a loss. Judge winners on revenue per visitor as well as the target number.
- Keep a log. Losing tests teach as much as winners, and a log stops the team from running the same test twice.
The goal-first rule works outside the site too
Speed Cube Shop is the paid-media version of the same idea. Its Google Ads efficiency went up 25% while each visitor brought in more revenue, and revenue from Google rose 30%. Cheaper clicks alone would not have counted as a win. The goal was revenue, so revenue was the number.
When not to A/B test
If your store gets too little traffic for a test to finish in a few weeks, skip the test. Fix what research already shows is broken (surprise shipping costs, forced account creation, missing size information) and watch the number for a month.
If a change is plainly right, ship it. A test is for questions you cannot answer any other way.
Questions people ask
What is ecommerce conversion rate optimization?
Ecommerce conversion rate optimization is the work of getting more of your existing visitors to buy, and to buy more. It covers product pages, the cart, checkout and what happens after the first order. Good CRO starts with one goal, such as add to cart rate or average order value, and uses tests to find out what moves it.
What is a good ecommerce conversion rate?
It depends on what you sell, what it costs and where your traffic comes from, so an industry average tells you little. A store selling a $20 consumable should convert far better than one selling a $900 sofa. Track your own rate by traffic source and device, and judge every change against your own baseline.
What should I A/B test first on my ecommerce site?
Start where the most money leaks. Look at your funnel and find the step with the biggest drop: product page to cart, cart to checkout, or checkout to order. Then test the thing most likely causing that drop. For many stores that is extra costs such as shipping, tax and fees, the top reason US shoppers give Baymard for abandoning checkout.
How much traffic do you need to A/B test?
Enough for the test to finish in a few weeks with a trustworthy answer. Put your current conversion rate, daily traffic and the smallest lift you care about into a sample size calculator before launch. If the answer is months, skip the test, fix what research already shows is broken and watch the number instead.
Do button color tests ever work?
Usually not, and that is the point. If shoppers can see the button, its color is not why they leave. Price, shipping cost, trust and doubt about the product are. A color test spends traffic you could use on one of those questions, so save it for after the bigger problems are fixed.
Klaviyo email marketing for DTC brands
Flows, campaigns, design and deliverability, run as one program by senior people.
See how we run itBefore your next test, write one sentence: the goal, the number and why the change should move it. If you cannot write it, do not run the test. If your store converts first-time visitors fine and the real gap is getting them back, that is retention, and our email marketing team runs it for DTC brands end to end. Book a call if you want a second set of eyes on your numbers.
