60 Ecommerce A/B Test Ideas, Sorted by Page and Traffic
60 A/B test ideas for online shops, each tagged with the research signal that should trigger it, the size of the change and how much traffic it needs.
By Joost BakkerPublished 17 min read
On this page
- How to use this list
- Homepage (ideas 1–8)
- Category pages and filters (ideas 9–17)
- Product pages (ideas 18–31)
- Cart (ideas 32–39)
- Checkout (and its limits on Shopify) (ideas 40–46)
- Site-wide: navigation, search and trust (ideas 47–54)
- Pricing and offers (ideas 55–60)
- Ideas that rarely win
- Turning this list into your backlog
- FAQ
How to use this list
Most "A/B test ideas" lists give you 50 things to try and no way to choose. That leads to random testing: you test whatever sounds good, most tests show no difference, and the team loses faith in testing.
Use this list differently:
- Start from your research. Look at heatmaps, recordings, surveys and analytics first. Our guide to ecommerce CRO explains how.
- Find the matching signal. Each idea has a "Run it when you see…" column. If you don't see that signal on your shop, skip the idea.
- Check the traffic tag. If the page doesn't get enough visitors for that kind of change, pick a bolder idea or a page with more traffic.
- Score what's left. Put the remaining ideas in your backlog and rank them with PXL, ICE or PIE. See how to prioritize A/B test ideas.
- Write a real hypothesis for each. "Because we saw X, we believe Y will cause Z." See the hypothesis template.
What the tags mean
Change size
- Big: visitors will clearly notice. A new layout, a new offer, a different page structure.
- Medium: a visible change to one important element. A rewritten value proposition, new product photos.
- Small: a detail. Button color, a word in a heading, an icon.
Traffic needed (to the page you're testing, per week)
| Tag | What it can detect | Visitors per week to the page |
|---|---|---|
| L (low) | A change of about ±20% in orders (or about ±10% in add-to-carts) | about 7,000 (about 9,500 for the add-to-cart case) |
| M (medium) | A change of about ±10% in orders | about 27,000 |
| H (high) | A change of about ±5% in orders | about 104,000 |
How we calculated this: a 3% order conversion rate, two versions (A and B), 95% confidence (two-sided), 80% power, and a 4-week test. For a 20% relative lift (3.0% → 3.6%) you need about 13,900 visitors per version, so about 27,800 in total, or about 7,000 a week over four weeks. A 10% lift (3.0% → 3.3%) needs about 53,200 per version (about 26,600 a week). A 5% lift needs about 208,000 per version (about 104,000 a week). If you measure add-to-cart instead (say 8%, aiming for +10%), you need about 18,900 per version, or about 9,500 a week.
These are rough guides. Your numbers depend on your conversion rate. Use the sample size calculator with your own data. If even L is out of reach, read A/B testing with low traffic.
A rule of thumb: small changes rarely move orders by 20%. If your page only qualifies for L, test big changes.
Homepage (ideas 1–8)
The homepage is often the most visited page but not the most important one. Many visitors land on product or category pages from ads and search. Check your landing page report before investing here.
| # | Idea | Run it when you see… | Change size | Traffic |
|---|---|---|---|---|
| 1 | Replace a rotating carousel with one static hero and one clear message | Clicks on slides 2–5 are near zero in the heatmap | Medium | M |
| 2 | Rewrite the hero headline to say what you sell and for whom | Survey answers like "I wasn't sure what this shop sells" | Medium | M |
| 3 | Show top categories as image tiles right below the hero | Scroll map shows many visitors never reach the category links | Big | L |
| 4 | Add best-sellers above the fold on mobile | Mobile visitors bounce from the homepage more than desktop visitors | Big | L |
| 5 | Move USPs (free delivery, returns, delivery time) into a bar under the menu | Exit survey mentions shipping cost or delivery time | Medium | M |
| 6 | Add a search bar that's visible by default on mobile (not just an icon) | Recordings show visitors scrolling through the menu looking for a product | Medium | M |
| 7 | Show reviews or a rating summary near the top | First-time visitors ask "Is this shop reliable?" in surveys | Small | H |
| 8 | Remove an entry pop-up or delay it | Recordings show visitors closing the pop-up within seconds; high bounce on mobile | Medium | M |
Category pages and filters (ideas 9–17)
Category pages decide whether visitors find a product they like. Problems here are often visible in recordings: endless scrolling, filters opened and closed, back-and-forth between products.
| # | Idea | Run it when you see… | Change size | Traffic |
|---|---|---|---|---|
| 9 | Show filters as visible buttons on mobile instead of hiding them behind one "Filter" button | Few mobile visitors use filters; recordings show long scrolling | Big | L |
| 10 | Add the most used filters (size, price, color) as quick chips above the products | Heatmap shows most filter clicks go to one or two filters | Medium | M |
| 11 | Change the default sort order (for example, best-sellers first) | Products at the top of the list get few clicks | Medium | M |
| 12 | Show more products per row on mobile (2 instead of 1) | Visitors scroll far but click little | Big | L |
| 13 | Add a second product image on hover (desktop) or swipe (mobile) | Visitors open many product pages and return quickly | Medium | M |
| 14 | Show delivery time or "in stock" on product cards | Survey answers mention delivery time; recordings show visitors checking stock on product pages | Medium | M |
| 15 | Show price per unit or a "from" price for products with variants | Visitors open product pages and leave after seeing the real price | Small | H |
| 16 | Replace pagination with "load more" (or the other way around) | Click data shows few visitors go past page 1 | Medium | M |
| 17 | Add a short buying guide at the top of complex categories | Survey answers like "I don't know which one to choose" | Medium | M |
Product pages (ideas 18–31)
The product page is where most shops win or lose the sale. It's also usually the page template with the most traffic, because a test on "all product pages" pools visitors from every product. That makes it the best place for small and mid-size shops to test.
| # | Idea | Run it when you see… | Change size | Traffic |
|---|---|---|---|---|
| 18 | Put the delivery date ("Order today, delivered Thursday") under the price | Delivery is a top answer in exit surveys | Medium | L* |
| 19 | Show shipping cost (or "free shipping from €X") next to the price | Survey answers or recordings show visitors going to the cart only to check shipping | Medium | L* |
| 20 | Move reviews higher or add a rating summary near the title | Scroll map shows few visitors reach the reviews | Medium | M |
| 21 | Add a size guide link right next to the size selector | Recordings show visitors hunting for size info; returns due to size | Medium | L* |
| 22 | Replace a size dropdown with visible size buttons | Recordings show repeated dropdown opening; rage clicks on the selector | Medium | M |
| 23 | Make the add-to-cart button sticky on mobile | Mobile visitors scroll down into the description and never scroll back up | Medium | L* |
| 24 | Rewrite the product description as short benefit bullets above the long text | Low scroll depth; survey answers ask questions the description already answers | Medium | M |
| 25 | Add product photos in use (on a person, in a room) | Visitors zoom on images a lot; survey answers about size or look | Big | L |
| 26 | Add a short product video | Same as above, especially for products where movement or fit matters | Big | L |
| 27 | Show stock level when it's genuinely low | Many add-to-carts but slow follow-through; only if the stock data is real | Small | H |
| 28 | Add an FAQ block with the top 5 questions from customer service | Repeated questions in survey answers or support tickets | Medium | M |
| 29 | Add trust signals near the button (returns policy, secure payment, guarantee) | Survey answers mention doubt about returns or trust | Small | M |
| 30 | Show "frequently bought together" below the button | Low items per order; customers buy accessories in separate orders | Medium | M |
| 31 | Remove distractions near the button (social share icons, extra links) | Heatmap shows clicks spread over elements that don't lead to a sale | Small | H |
* These ideas often move add-to-cart clearly before they move orders. If your shop is small, use add-to-cart as the primary goal and keep orders as a secondary goal. In Varly, for example, you can set Added to cart as the primary goal and Placed an order as a secondary goal, so you see both. See read your test results.
Revenue tip: before you test an element, check whether visitors who use it actually buy more. A "click impact" view (orders and revenue of visitors who clicked an element vs. those who didn't) helps you pick the elements worth testing. It shows a link, not a cause, so use it to choose tests, not to skip them.
Cart (ideas 32–39)
The cart is the last page most testing tools can change. Visitors here have shown real intent, so even small leaks matter.
| # | Idea | Run it when you see… | Change size | Traffic |
|---|---|---|---|---|
| 32 | Show the total including shipping (or a shipping estimate) in the cart | High exit rate on the cart; survey answers about extra costs | Medium | M |
| 33 | Add a "free shipping from €X" progress bar | Many orders just below the free-shipping threshold | Medium | M |
| 34 | Make the checkout button the only primary button, above the fold on mobile | Recordings show mobile visitors scrolling to find the button | Small | M |
| 35 | Move the discount code field into a collapsed link | Recordings show visitors leaving the cart to search for codes | Small | M |
| 36 | Show payment method logos your customers use (iDEAL, PayPal, Klarna…) | Survey answers ask which payment methods you accept | Small | H |
| 37 | Add a short reassurance block (returns, delivery date, support contact) | Exit survey answers mention doubt at the last moment | Small | M |
| 38 | Use a slide-out cart drawer instead of a cart page (or the other way around) | Many visitors open the cart and go back to shopping | Big | L |
| 39 | Remove cross-sells from the cart | Cross-sells get few clicks and recordings show them pushing the button down | Medium | M |
Remember that the reasons people abandon carts are mostly about money and trust. In Baymard Institute's checkout research, the top reason was "extra costs too high" (40% of shoppers who abandoned, excluding those just browsing), followed by slow delivery (20%), not trusting the site with card details (19%) and forced account creation (18%) (Baymard). Ideas 18, 19, 32 and 33 go straight at the first two.
Checkout (and its limits on Shopify) (ideas 40–46)
Checkout tests are valuable but harder to run:
- On Shopify, your theme code doesn't run in the checkout, so client-side testing tools can't change it. Shopify's own Rollouts can test different checkout configurations against each other on the Grow plan and higher (Shopify Help Center). Some apps offer checkout tests only on Shopify Plus (Intelligems pricing).
- On WooCommerce and Magento, the checkout is part of your site, so you can test it, but take care with payment fields and privacy.
- Privacy: many tools (Varly included) don't track or change checkout pages by default. That's on purpose: checkout pages hold personal and payment data.
| # | Idea | Run it when you see… | Change size | Traffic |
|---|---|---|---|---|
| 40 | Offer guest checkout as the first option | Survey answers or support tickets about forced account creation | Big | L |
| 41 | Reduce the number of form fields (for example, one name field, address lookup) | Funnel shows a big drop on the address step | Medium | M |
| 42 | Show the order summary on mobile (collapsed, with the total visible) | Mobile drop-off in checkout higher than desktop | Small | M |
| 43 | Show the delivery date on the shipping step | Delivery time is a top abandonment reason in your surveys | Medium | M |
| 44 | Put the most used payment method first | Payment step drop-off; analytics show one method dominates | Small | H |
| 45 | One-page vs. multi-step checkout (WooCommerce, Magento) | Visitors drop between steps rather than within one | Big | L |
| 46 | Add express checkout buttons (Shop Pay, Apple Pay, PayPal) earlier | Many mobile visitors; high drop on form steps | Medium | M |
Measure checkout tests on completed orders, not on "reached the next step." A step that's easier to start but harder to finish doesn't help.
Site-wide: navigation, search and trust (ideas 47–54)
Site-wide tests change something on every page. They pool all your traffic, which helps small shops, but a site-wide change can also break things in many places at once. Preview every page type before you launch.
| # | Idea | Run it when you see… | Change size | Traffic |
|---|---|---|---|---|
| 47 | Simplify the main menu (fewer top-level items, clearer names) | Menu click data spread thin; recordings show hesitation in the menu | Big | L |
| 48 | Show the search bar open on every page | Visitors who search convert better than those who don't (check your analytics) | Medium | M |
| 49 | Improve "no results" pages with suggestions and popular products | Many searches with zero results | Medium | M |
| 50 | Add a USP bar (delivery, returns, payment) to every page | Same doubts appear in surveys on many page types | Medium | L |
| 51 | Show a customer service contact (chat, phone, WhatsApp) in the header | Survey answers like "I had a question but couldn't find how to ask" | Small | M |
| 52 | Make the logo, menu and cart icon larger on mobile | Rage clicks or dead clicks around the header on mobile | Small | H |
| 53 | Add a trust mark or review score in the footer and header | First-time visitors convert much worse than returning visitors | Small | H |
| 54 | Change how the cookie banner looks (smaller, bottom instead of full-screen overlay) | Bounce on first page view is high on mobile. Check with your privacy advisor first; never make "reject" harder to find | Medium | M |
Pricing and offers (ideas 55–60)
Price and offer tests can have the biggest effect, and the biggest risk. Two warnings:
- Showing different prices to different visitors can be illegal or damage trust, depending on your country and how you do it. Get advice first.
- Most visual testing tools only change what's on screen, not the price your shop charges. Real price tests need a tool that changes the price in your shop system, such as Shopify catalogs in Rollouts or a price-testing app.
| # | Idea | Run it when you see… | Change size | Traffic |
|---|---|---|---|---|
| 55 | Change the free-shipping threshold | Many orders cluster just below or just above the current threshold | Big | L |
| 56 | Free returns vs. paid returns (message and policy) | Returns are a top worry in surveys | Big | L |
| 57 | Percentage vs. fixed-amount discount in a promotion ("20% off" vs. "€10 off") | You run regular promotions and want to know which framing works | Medium | M |
| 58 | Bundle offer ("buy 2, save 10%") on products often bought together | Many orders with two or more of the same product type | Medium | M |
| 59 | Show "pay later" options (Klarna, etc.) near the price | Higher-priced products; survey answers about price | Small | M |
| 60 | Show a price anchor (original price, value of the bundle) | Only if the anchor is genuine and allowed in your market | Small | H |
Ideas that rarely win
These come up in many test-idea lists. They rarely produce a clear result on their own, mostly because the change is too small to move orders by an amount a shop can detect. You may still have a good reason to test them; just don't start here.
- Button color. Unless the button is genuinely hard to see, a color change is too small to detect without huge traffic.
- Tiny copy changes ("Buy now" vs. "Add to bag"). Same problem.
- Countdown timers that aren't real. Visitors notice when the timer resets. Fake urgency can hurt trust, and may be unlawful in some places.
- More pop-ups. Each pop-up competes with your product.
- Trust badges no one recognizes. A logo from an unknown "secure shop" scheme adds nothing.
- Changes on low-traffic pages (About us, blog). Even if they help, you'll rarely have the visitors to prove it.
If you want to test something small, bundle several small improvements into one bigger change. You lose the ability to say which part worked, but you gain a test that can finish.
Turning this list into your backlog
- Copy the rows that match signals you've actually seen into your backlog (a spreadsheet works).
- Add a hypothesis to each: the evidence, the change and the metric.
- Add a traffic check: visitors per week to the page, and whether that's enough for the change size.
- Score them with ICE, PIE or PXL and start with the top three. See PXL vs ICE vs PIE.
- Pick one primary metric per test (orders, or add-to-cart if traffic is low), and add guardrails such as orders when the primary goal is add-to-cart.
In Varly, the Ideas backlog does steps 2–4 in one place: each idea keeps the research it came from, a PXL, PIE, ICE or custom score, and a Can we test this? check that says whether the page gets enough visitors and roughly how many weeks a test needs (prioritization help). Varly costs €99 per website per month.
For the full process from idea to result, read A/B testing for ecommerce. Shopify-specific ideas and limits are in CRO on Shopify.
FAQ
What should I A/B test first on my online shop? The product page template, if your research points to a problem there. It usually has the most traffic and the most direct link to sales.
How many visitors do I need to run an A/B test? It depends on your conversion rate and how big the change is. With a 3% conversion rate, detecting a 20% lift in orders takes about 27,800 visitors in total to the tested page. Use the calculator for your own numbers.
Can I A/B test the Shopify checkout? Not with tools that run in your theme. Use Shopify Rollouts (Grow plan and higher) to test checkout configurations, or an app that supports checkout tests on Shopify Plus.
Should I test prices? Only with a tool that changes the price in your shop system, and only after checking the legal side in your market. Free-shipping thresholds and bundles are safer places to start.
What if my test shows no difference? That's a result. The change didn't matter enough to your visitors. Keep the simpler version, write down what you learned, and test something bolder next. See the A/B test report template for how to write it up.
Varly turns research into scored ideas and ideas into A/B tests, in one tool. Request access.
