Conversion rate optimization: what moves the number
Most CRO advice is a list of tweaks: change the button color, add a countdown timer, move the reviews. Some of those help, most do little, and without a process you never know which is which. This guide covers what we do with our own clients: find where people leave, understand why, fix the biggest leak first, and test in a way that gives answers you can trust.
What CRO is for
The goal is more money from the visitors you already pay for, and a higher conversion rate is only one way to get there. So the number we watch first is revenue per visitor: conversion rate times average order value. A change that lowers conversion rate a little but raises order value a lot is a win. A discount that lifts conversion rate and cuts margin often isn’t.
When revenue per visitor goes up, the same ads earn more, and you can afford to pay more for traffic than competitors can. That’s where CRO meets paid ads.
Find the leak first
Before changing anything, look at where people leave. Every store has one step that loses more people than the others, and that’s where the first work should go.
Where 10,000 visitors go
The funnel from visit to order, and where each step loses people.
For most DTC stores, the biggest drop is between viewing a product and adding it to the cart. That’s the product page’s job, and it’s usually where the biggest wins are. Check the same funnel split by mobile and desktop, and by new and returning visitors. A step that looks fine on average is often broken on one device.
The CRO process
Five steps, on repeat
How a CRO program runs, month after month.
Research: numbers first, then words
The numbers tell you where people leave; only people can tell you why.
- Analytics: the funnel above, landing pages by bounce rate, device split, page speed.
- Heatmaps and session recordings: where people tap, how far they scroll, where they hesitate.
- On-site surveys: one question on the product page or at exit, such as “What’s stopping you from ordering today?”
- Reviews and support emails: the questions people ask before buying, and the complaints after.
- User tests: watch five people from your target group try to buy. It’s humbling and fast.
Hypothesis: one clear sentence
Write every idea like this: “Because we saw [research finding], we believe [change] will [result] for [visitors]. We’ll know by [metric].” If you can’t fill in the first part, it’s a guess, not a hypothesis.
Prioritize: impact, confidence, effort
Score each idea from 1 to 10 on how much it could move the number, how sure the research makes you, and how easy it is to build. Start with the highest total.
| Idea | Impact | Confidence | Ease | Score |
|---|---|---|---|---|
| Show the size guide next to the size picker (surveys: “not sure about my size”) | 8 | 8 | 9 | 25 |
| Make 3-for-2 the default option on the product page | 9 | 6 | 8 | 23 |
| Add express checkout buttons | 6 | 7 | 8 | 21 |
| Redesign the homepage | 5 | 3 | 2 | 10 |
Best practices, page by page
These aren’t rules to copy blindly; they’re the places we check first, because they fix the most common problems.
The product page
Seamless leggings that stay up through every squat
- Images that answer questions. In use, on different bodies, close-ups of the fabric, and one that shows the size.
- A headline that says what it does, with the product name below it. The same promise the ad made.
- Reviews right under the title, with the count. Proof before the price.
- The offer as the default choice, with the price per unit visible.
- One clear button, sticky on mobile so it’s always in reach.
- Shipping and returns next to the button, not hidden in the footer.
Below the first screen: answer the top objections from your research (fit, quality, washing, delivery), show more reviews with photos, and repeat the button. If most visitors come from ads, the page should continue the ad’s promise, in the same words. More on that in our direct response funnel guide.
The cart
- Show the total, including shipping, as early as possible. Surprise costs are the most common reason people leave at checkout.
- A progress bar to free shipping or a gift, and one relevant add-on. Our AOV offers guide shows the options.
- Keep the checkout button visible without scrolling, especially on mobile.
The checkout
- Guest checkout. Asking for an account before the first order loses sales.
- Express payment buttons (Apple Pay, Google Pay, PayPal, Shop Pay) at the top.
- Only the fields you need. Every extra field is a reason to stop.
- The payment methods your market expects. In some countries, pay-by-invoice or local methods matter more than cards.
Everywhere
- Speed. Heavy images, apps and scripts slow the page down. Check it on a mid-range phone on mobile data, not your office Wi-Fi.
- Mobile first. Most ad traffic is on phones. Review every change on a phone before a laptop.
- Search and navigation that find the product in two taps.
How to run A/B tests properly
An A/B test shows half of visitors version A and half version B, and compares the results. Simple in principle, easy to get wrong in practice.
1. Decide the sample size before you start
Small differences need a lot of traffic to prove. Visitors needed per version (95% confidence, 80% power):
| Your conversion rate | To detect +10% | +20% | +30% |
|---|---|---|---|
| 1% | 163,100 | 42,700 | 19,900 |
| 2% | 80,700 | 21,200 | 9,800 |
| 3% | 53,300 | 14,000 | 6,500 |
At a 2% conversion rate, proving a 10% lift takes about 80,000 visitors per version. Many stores don’t have that. Test bigger changes, which produce bigger differences, or use the methods further down.
2. Run it for full weeks
People buy differently on Monday than on Saturday, and around payday. Run every test for at least one full week, ideally two, even if the sample size is reached sooner.
3. Don’t stop when it looks good
The peeking problem
Why a test that looks like a winner on day 3 often isn’t.
Checking every day and stopping the moment one version crosses 95% is the most common way to “find” a winner that doesn’t exist. Set the sample size and the end date up front, and only call the result then.
4. Pick one main metric
Choose the metric that decides the test before it starts. For most tests that’s revenue per visitor, or conversion rate for tests that don’t touch price or offer. Look at others (AOV, add-to-cart rate, device split) to understand the result, not to find a winner after the fact.
5. Check that the split is fair
If you planned 50/50 and one version got 53% of visitors, something is broken in the setup. Fix it before reading the results.
Beyond A/B testing
Not every store has the traffic for clean tests, and not every question needs one. Other ways to improve conversion:
- Fix what’s clearly broken. A slow page, a bug on one browser, a missing size guide. No test needed; just fix it.
- Before and after. Make one big change, compare four weeks before with four weeks after, and account for season and ad spend. Less exact, still useful.
- Test in the ads, not on the site. Send two ad sets to two different landing pages. Meta splits the traffic, and you can compare cost per purchase.
- Test the offer, not the button. A new bundle or guarantee changes behavior far more than design tweaks, so the difference shows up with less traffic.
- User tests and surveys. Five user tests find most of the obvious problems, with no traffic needed.
- Look after the purchase. Returns, reviews and repeat orders. A change that lifts conversion but brings more returns isn’t a win.
Common mistakes
- Copying another brand’s page without knowing why it works for them.
- Testing small details on a page with a big problem, like an unclear offer.
- Calling tests early, or running them on too little traffic.
- Judging by conversion rate alone when the change affects price or order value.
- Forgetting the ad. If the ad and the page don’t match, no page test will fix it.
- Not writing down results, so the same ideas get tested twice.
The short version
- The goal is more revenue per visitor, not only a higher conversion rate.
- Find the step that loses the most people first. For most DTC stores it’s the product page.
- Research before you test: numbers show where people leave, surveys and recordings show why.
- Decide the sample size up front, run tests for full weeks and don’t stop when it looks good.
- Without enough traffic, fix what’s broken, test bigger changes and test offers in the ads.
FAQ
What is conversion rate optimization?
Conversion rate optimization (CRO) is the process of finding where and why visitors leave a store, making changes to fix it, and measuring the result, usually with A/B tests. For e-commerce, the real goal is more revenue or profit per visitor, not only a higher conversion rate.
What is a good conversion rate for an online store?
It varies a lot by product, price, traffic source and device, so general benchmarks are of limited use. Compare your store with itself over time and by segment (mobile vs. desktop, new vs. returning, each traffic source), and focus on the step in the funnel that loses the most people.
How long should an A/B test run?
Until it reaches the sample size you set before starting, and for at least one full week, ideally two, so every day of the week is included. Stopping early because one version looks like a winner leads to false results.
How much traffic do I need for A/B testing?
More than most people expect. At a 2% conversion rate, detecting a 10% improvement needs about 80,000 visitors per version at 95% confidence and 80% power; a 30% improvement needs about 10,000. Smaller stores should test bigger changes or use other methods.
What should I test first?
The step that loses the most visitors, usually the product page, starting with ideas that your research supports, that could have a big impact and that are easy to build. Score ideas on impact, confidence and effort.
What can I do if I don’t have enough traffic to A/B test?
Fix obvious problems without testing, run user tests and surveys, compare before and after for big changes, test offers rather than small design details, and split test landing pages through your ad platform.