The best checkout A/B tests in our archive mostly removed something: an extra step, a code box, a second button, a timer.
The biggest documented wins were AnotherSole merging its discount-code step into the cart (+131% revenue), a checkout cut from eight steps to one (+68% order completions), and full-size express payment buttons in the cart drawer (+47.75% and +37.33% conversion in two Blend Commerce tests).
How we ranked them: we took the cart and checkout tests in the LiftArchive research database that were run as A/B or split tests and won, then ranked them by the lift each one reported on its main metric.
Metrics differ (revenue, conversion rate, orders), and some sources give no sample size, so each entry says how solid its evidence is.
The 20 checkout tests at a glance
| Rank | Test | Change | Result | Who ran it |
|---|---|---|---|---|
| 1 | AnotherSole one-step cart | Merged discount-code step into cart | +131% revenue | ConvertPolo |
| 2 | Bingo Card Creator fast cart | Fast home-built cart | +94% conversions | Patrick McKenzie |
| 3 | Eight steps to one | Removed forced sign-up and extra steps | +68% order completions | MECLABS |
| 4 | Fame Shop Pay button | Full-size Shop Pay button in cart drawer | +47.75% conversion rate | Blend Commerce |
| 5 | Media subscription checkout | On-brand page, navigation removed | +40% subscriptions | MECLABS |
| 6 | Aeromotive PayPal Express | PayPal Express button replaced logos | +37.33% conversion | Blend Commerce |
| 7 | Vacation cart single CTA | One Checkout button, extras as add-ons | +36.5% completions | MECLABS |
| 8 | Readers.com localized checkout | Local shipping, dates and currency | +30% conversion rate | One Click |
| 9 | Checkout timer removed | Hid the countdown timer | Revenue per visitor +26.18% | Kasper Dolk |
| 10 | Bionic Gloves code fields | Removed coupon and gift card boxes | +24.7% revenue | Sq1 |
| 11 | Cart trust block | Dispatch, returns and payment block | +23.8% revenue per visitor | Oliver Kenyon |
| 12 | Shipping protection toggle | One toggle, off by default | +23.7% conversion | Convertibles |
| 13 | Reviews at checkout | Customer reviews on checkout page | +23.13% revenue per visitor | Dave Diederen |
| 14 | Fresh Clean Threads icons | Guarantee, review, quality icons | +21% conversion | Visually |
| 15 | BioLite contact block | Explained email and phone fields | +18% transactions | Surefoot |
| 16 | Guarantee under button | 30-day money-back line | +18% revenue per visitor | Dave Diederen |
| 17 | Ashley Furniture address step | Address entry moved before checkout | +15% conversions | AB Tasty |
| 18 | Kickz cart drawer | Cart page replaced by one-CTA drawer | +14.2% RPU | DRIP Agency |
| 19 | Odd Muse no-fees line | 'No extra fees' for US shoppers | +14% conversion | Visually |
| 20 | Unlocked free shipping copy | Confirmed free shipping in cart | +13.93% revenue per visitor | Dave Diederen |
Want every result, including the small ones? Browse all checkout tests and cart tests.
The 20 best checkout A/B tests, ranked
1. AnotherSole: discount code merged into a one-step cart
The change: The cart and the discount-code screen were two separate steps.
The variant merged them into one step with a clearer visual order.
The result: +131% revenue, plus +51% thank-you page visits.
Run by ConvertPolo for AnotherSole on VWO. A/B test, but no sample size or significance was published, so treat the size with care.
Why it likely worked: A whole extra screen just to enter a code is a step where people stall or leave to hunt for codes. Folding it in removed the stop.
Try it if: your cart sends people to a separate page or modal for codes, gift cards or other extras before checkout.
2. Bingo Card Creator: a fast home-built cart
The change: Patrick McKenzie replaced a slow hosted cart with Continue Shopping and Update Cart steps with a fast lightbox cart that worked out prices locally. Payment stayed with the same provider.
The result: +94% conversions. Run by Patrick McKenzie in 2009 on downloadable software.
Several cart changes were bundled, and sample size and confidence were not stated.
Why it likely worked: Speed and fewer buttons.
Every pause and every extra button before payment gives a buyer a chance to drift off.
Try it if: you sell software or digital products through a hosted cart you never chose, and it loads slowly.
3. Eight checkout steps cut to one
The change: The original flow took eight steps from product page to receipt, with forced account creation across three pages and a double order confirmation.
The variant was one basic step, with registration and the extra confirmation removed.
The result: +68% order completions. Run by MECLABS; the industry was not named.
Why it likely worked: Forced sign-up is a classic checkout leak.
Removing it and the second confirmation took out pages that did nothing for the buyer.
Try it if: your checkout makes people create an account or confirm their order twice.
4. Fame: a dedicated Shop Pay button in the cart drawer
The change: The cart drawer had only a standard Checkout button.
The winning arm added a Shop Pay button of the same size below it.
A third arm with payment icons raised conversion but cut checkout starts by 7%.
The result: +47.75% conversion rate, with +69.30% revenue per visitor and +14.59% average order value.
Run by Blend Commerce on Intelligems as an A/B/C test. No significance level was stated.
Why it likely worked: Returning Shop Pay users can skip typing an address and card.
A full-size button makes that shortcut obvious, where small icons only hint at it.
Try it if: you're on Shopify and a large share of your buyers already use Shop Pay.
5. A media subscription checkout cleaned up
The change: The first checkout step used off-brand colors, barely showed the logo and had two levels of navigation.
The variant was a simple on-brand page with the navigation removed.
The result: +40% subscriptions for a large media publisher.
Run by MECLABS; the publisher's name was blurred.
Why it likely worked: An off-brand page makes people wonder if they're still on the right site, right when they're asked for card details. Navigation links give them an exit.
Try it if: you sell subscriptions and your checkout runs on a third-party template that doesn't look like your site.
6. Aeromotive: PayPal Express instead of payment logos
The change: The cart drawer showed Affirm and credit card logos.
The variant swapped the logos for a working PayPal Express button.
The result: +37.33% conversion rate, +35.60% average order value and +75.48% revenue per user. Run by Blend Commerce for Aeromotive.
Why it likely worked: Logos tell people what they can pay with.
A button lets them pay that way right now, skipping the forms.
Try it if: your cart shows payment logos but no express wallet buttons.
7. A vacation cart cut to one Checkout button
The change: The cart gave visitors three equally weighted options to choose between.
The variant folded the extras into the product details as plus-sign add-ons and left one main Checkout button.
The result: +36.5% completions for a package-vacation company. Run by MECLABS.
Why it likely worked: Three equal buttons force a decision the buyer didn't come to make. One clear next step removes it.
Try it if: your cart offers several buttons of the same weight, such as protection plans, upgrades and Checkout side by side.
8. Readers.com: checkout localized by country
The change: Everyone used to see the same checkout.
The variant showed local details by visitor country, such as Canadian shipping rates, delivery estimates and currency for Canadians.
The result: +30% conversion rate. Run by One Click for Readers.com on VWO, as a segment-targeted test. No sample size or significance was given.
Why it likely worked: Foreign visitors worry about surprise duties, shipping costs and currency.
Local numbers answer all three before they're asked.
Try it if: a real share of your orders come from outside your home country.
9. Removing the checkout countdown timer
The change: A Shopify Plus checkout showed a countdown timer.
The variant hid it. A post-purchase survey had found Dutch shoppers felt the timer was too much pressure.
The result: +26.18% revenue per visitor (€2.96 to €3.73), with conversion up 25.61% (0.82% to 1.03%).
Run by Kasper Dolk. The conversion range stayed above zero; the revenue range did not. One store, one market.
Why it likely worked: By checkout the shopper has already decided.
A ticking clock at that point adds anxiety, not speed.
Try it if: you run a timer at checkout.
This one is worth testing in both directions; see the losers below.
10. Bionic Gloves: coupon and gift card fields removed
The change: The cart page had boxes for a special-offer code and a gift card code. The variant removed both.
The result: +24.7% revenue and +17.1% revenue per visitor.
Run by Sq1 for Bionic Gloves on VWO.
Small sample: about 1,400 visitors over 48 days, significance not stated.
Why it likely worked: An empty code box tells people a discount exists that they don't have.
Some leave to look for one and never come back.
Try it if: few of your orders use codes, or you can apply offers automatically through links instead.
11. A dispatch, returns and payment block under the cart button
The change: A plain cart drawer got three changes: a "You've unlocked free shipping" note, the size shown as a dropdown, and a block under Checkout with tracked next-day dispatch, 14-day easy returns and payment icons.
The result: +23.8% revenue per visitor ($0.771 to $0.955) at 99.41% confidence, on about 203,000 visitors over 16 days.
Run by Oliver Kenyon on a fashion store.
This is the best-documented test on the list, but three changes were tested together.
Why it likely worked: It answers the delivery, returns and payment questions at the exact moment of checkout, and the size dropdown handles fit worries without leaving the cart.
Try it if: you sell items where fit or delivery time is a common doubt.
12. Shipping protection as one toggle, off by default
ConvertiblesThe change: A shipping protection widget showed two checkout buttons, with and without protection.
The variant used one toggle, off by default, with a single checkout button showing the live total.
The result: 1.53% to 1.90% conversion (+23.7%) and +16.4% revenue per visitor over 19 days, at 97% probability to beat control.
Run by Convertibles for an electronics accessories brand.
Average order value fell 6%, partly from fewer protection fees.
Why it likely worked: Two checkout buttons split attention at the most important click.
One button and an optional toggle keep the path clear.
Try it if: a protection or insurance app has added a second button or a pre-ticked fee to your cart.
13. Customer reviews on the checkout page
The change: The normal checkout had no reviews.
The variant added a few customer reviews. Dave Diederen suggests picking reviews that answer common doubts, such as shipping speed, sizing or durability.
The result: +23.13% revenue per visitor ($1.03 to $1.27), with conversion up 17.96%.
Run by Dave Diederen. The ranges were wide and crossed zero, so the real lift may be much smaller.
Why it likely worked: Most sites put proof on the product page.
Doubts return at payment, and reviews there meet them in the moment.
Try it if: your checkout can show custom content, such as Shopify checkout blocks or a custom checkout.
14. Fresh Clean Threads: guarantee, review and quality icons at checkout
VisuallyThe change: A plain, payment-only checkout got icons for the satisfaction guarantee, 5-star reviews and product quality.
The result: +21% conversion rate and +20% profit per session.
Credited to Visually. The test ran under one week and comes from vendor marketing without sample size or significance.
Why it likely worked: Three short reassurances fill an empty space at the step where people hesitate most.
Try it if: your checkout is bare and you have a real guarantee to show.
15. BioLite: explaining why checkout asks for contact details
The change: Shopify checkout asked for email and phone up front with no explanation.
The variant grouped them into a labeled Contact Information block below the shipping address, with helper text on how each would be used.
The result: +18% transactions, a 14% drop in step-1-to-step-2 drop-off and more email capture. Run by Surefoot for BioLite.
Why it likely worked: People hold back a phone number when they fear spam calls.
Saying what it's for, and asking after the address, lowers that worry.
Try it if: you require a phone number or email and drop-off is high on the first checkout step.
16. A 30-day money-back guarantee under the checkout button
The change: Nothing sat under the cart's Checkout button.
The variant added one line with a checkmark: "30 Day Money Back Guarantee."
A third arm with payment icons also beat control but trailed the guarantee.
The result: +18% revenue per visitor (£1.76 to £2.07) and +17% conversion (3.77% to 4.41%) over 30 days.
Run by Dave Diederen. The revenue range ran from -18% to +71%.
Why it likely worked: Doubt peaks at the button.
A refund promise right there answers it without adding clutter.
Try it if: you offer a guarantee but only mention it in the footer or on the product page.
17. Ashley Furniture: address entry moved out of checkout
The change: Shoppers without saved addresses had to type delivery and billing addresses inside the purchase funnel.
The variant asked them to log in and fill in the address before checkout.
The result: +15% conversions and -4% bounce rate.
Credited to AB Tasty for Ashley Furniture. No sample size or significance was given.
Why it likely worked: The funnel itself got shorter, so the steps that felt like buying came faster.
Try it if: logged-in shoppers make up a big share of your orders and often lack saved addresses.
18. Kickz: cart page replaced with a one-button drawer
DRIP AgencyThe change: Desktop shoppers went to a separate cart page with several calls to action.
The variant removed it and used a cart drawer with one CTA and an upsell section.
The result: +14.2% revenue per user, desktop only.
Run by DRIP Agency for Kickz. Significance and sample size were not stated.
Why it likely worked: Shoppers stay on the page they were browsing, see one clear next step, and still get offered an add-on.
Try it if: your cart page has many buttons competing with Checkout.
19. Odd Muse: 'no extra fees' message for US shoppers
The change: US shoppers saw nothing about duties.
The variant told them clearly there would be no extra fees at checkout.
The result: +14% conversion rate. Credited to Visually for Odd Muse.
Vendor marketing; sample size and significance not stated.
Why it likely worked: Overseas buyers fear duty bills on delivery.
One plain sentence removes that fear before they pay.
Try it if: you ship across borders and cover the duties, but don't say so.
20. 'You unlocked FREE shipping' instead of 'will be applied at checkout'
The change: Once the cart passed the free shipping threshold, it said free shipping would be applied at checkout.
The variant said "Congrats! You unlocked FREE shipping!"
The result: +13.93% revenue per visitor ($3.07 to $3.49), with conversion up 11.08%.
Run by Dave Diederen. The ranges crossed zero.
Why it likely worked: "Will be applied" asks people to trust a later step.
"You unlocked" confirms the benefit now and frames it as a reward.
Try it if: you have a free shipping threshold and your cart copy hedges about when it applies.
Patterns across the winners
Fewer steps beat clever steps
At least six of the top ten removed a step, a field or a choice: AnotherSole, Bingo Card Creator, the eight-step checkout, the media publisher, the vacation cart and Bionic Gloves. For more, see checkout flow tests.
Express payment works as a button, not a logo
Fame and Aeromotive both won by giving shoppers a working wallet button in the cart drawer.
In the Fame test, icons alone cut checkout starts.
Compare more in cart payment and express checkout tests.
Answer the last doubt where the button is
The guarantee line, the trust block, the checkout icons and checkout reviews all put reassurance next to the final click.
See checkout guarantee tests and checkout review tests.
Remove cost surprises
Readers.com, Odd Muse and the free shipping copy test each told buyers exactly what they'd pay.
Cross-border sellers in particular should state duties and currency up front. More in cart shipping messaging tests.
It isn't only ecommerce
The media subscription checkout (+40% subscriptions) and Bingo Card Creator (+94% conversions) are subscription and software sales.
Outside retail, payment gateway PayU dropped the email field from its hosted checkout and kept only the phone number for +5.8% checkout rate, a statistically significant result.
Checkout tests that lost
Every idea above has a version that failed somewhere.
These are worth reading before you copy a winner:
- Adding a checkout timer: Dave Diederen added one and lost -7.1% revenue per visitor, though another Diederen test of a timer won +3.16% revenue per visitor. Kasper Dolk's removal win (rank 9) points the same way as the loss. See checkout timer tests.
- Skipping the cart: sending shoppers straight to checkout lost -6.58% revenue per visitor in a Dave Diederen test. The cart was doing selling work (upsells, a free shipping bar, trust lines) that vanished.
- Stripping the mini cart to just the button: a checkout-first minicart without recommendations lost -9.9% conversion for Clean Commit.
- Moving app buyers to a web checkout: Zumba saw immediate conversions -25%; LTV +17%, so it lost on first-day conversion but counted as a win on lifetime value. Judge app checkout tests on revenue over time, not first-day conversion.
What to test next
- Audit your steps. Count every screen and field between the cart and the receipt. Remove anything that doesn't take payment or delivery details.
- Add a real express button to the cart. Shop Pay, PayPal or Apple Pay as a full-size button, then watch checkout starts as well as orders.
- Put one reassurance line under the checkout button. Use a guarantee, returns window or dispatch time you actually offer.
- Hide or remove the coupon box. Apply offers through links so the empty field stops sending people away.
- Test your timer, both ways. If you run one at checkout, a removal test is cheap and the archive shows it can go either way.
- Fix cross-border surprises. Local currency, delivery dates and a "no extra fees" line for foreign buyers.
Checkout changes affect every order, so judge them on revenue per visitor and order count together.
For a sense of how often tests win at all, see our A/B test win rate research.
Real results
@oliverkenyon
@DaveDiederen
Blend Commerce
Surefoot
@kasper_dolk@DaveDiederenWatch out for
Bundled changes. Several winners changed more than one thing at once, so you can't tell which part did the work. Retest the parts you copy.
Thin evidence. Many vendor and agency case studies give no sample size or significance.
A +131% result with no sample size is a lead worth testing, not a forecast.
Average order value tradeoffs. The shipping protection toggle raised conversion but cut order value 6%. Track both before you call a winner.
Fake urgency. Timers that reset or lie damage trust at the step where trust matters most.
If you'd rather have someone plan and run checkout tests for you, Oliver Kenyon ran the cart trust block test above.