88 statistics on A/B test win rates, ecommerce conversion, checkout, site speed and landing pages. Every number links to its original source, and the first section comes from our own database of real test results.
321 real ecommerce A/B test results from 9 practitioners, agencies and testing tools, each linked to the person who shared it. Updated November 2026.
Sources: LiftArchive (November 2026); Kohavi, Deng & Vermeer (2022); Optimizely (2023).
Of the 321 publicly shared ecommerce A/B test results in LiftArchive, 307 (96%) were shared as winners. Practitioners mostly post their wins, so public examples look far more successful than testing really is.
Among shared winning tests that report revenue per visitor, the median reported lift is +9.6% (n = 160).
Among shared winning tests that report conversion rate, the median reported lift is +7.7% (n = 180).
206 of 321 shared ecommerce tests (64%) report revenue per visitor, not just conversion rate.
The product page is the most-tested page: 138 of 321 shared tests (43%).
Large companies with mature testing programs publish how many of their experiments succeed. Most don't.
About one third of experiments at Microsoft improved the metric they were designed to improve.
Only about 15% of experiments at Bing were successful.
Booking.com, Google Ads and Netflix report success rates of roughly 10% for their experiments.
Only about 8% of Airbnb search experiments were successful.
With an 8% success rate and 80% power, about 26% of statistically significant 'wins' would be false positives (false positive risk).
To detect a 10% relative change on a 3.7% conversion rate with 80% power, a test needs about 41,642 users per variant.
Around 12% of experiments win on the primary metric; 88% of tests do not win.
The median company on Optimizely runs 34 experiments per year; the top 3% run more than 500.
Companies with over $1B in revenue had an average experiment win rate of 16.0%, versus 10.7% for companies under $100M.
Experiment win rates by industry ranged from 9.1% (business services) to 17.0% (food and beverage).
Fewer than 10% of experiments test 4 or more variants, yet those experiments are more than twice as impactful as A/B tests.
Most experiment uplifts decay to about 80% of their first-month value after a year, except revenue uplifts, which keep 91%.
Call-to-action clicks are the most common primary metric, used in 34.8% of experiments, with a 23.4% win rate and 38.2% average winning uplift.
Over 92% of all experiments target the top 5 metrics (CTA clicks, revenue, checkout, registration, add to cart).
Three of four experiments on Optimizely are simple A/B tests.
Spotify reports a win rate of about 12%, while roughly 64% of its experiments produce a clear learning.
In 2023 Google ran over 700,000 Search experiments, including 16,871 live traffic experiments, resulting in 4,781 launches.
About 70% of effects tested on a commercial A/B testing platform were truly null (no real effect).
Among website A/B test results significant at 5%, an estimated 18%-25% are false discoveries; about one in five 'wins' is ineffective when deployed.
60% of completed A/B tests on Convert delivered a lift under 20%, and 84% came in under 50%.
A/B tests account for 67.6% of all experiments on Convert; multivariate tests are under 1%.
VWO's 2026 benchmark covers 193K experiments across 38K websites; CTA copy was the most tested tactic (~3,500 times) with a ~10% win rate.
Benchmarks from live store data. Medians, not averages, unless the source says otherwise.
The median Shopify store converts 1.4% of sessions into a purchase; the mean is 1.9%.
A conversion rate above 2.7% puts a Shopify store in the top 20%; the top 10% reach 3.7% or more.
Shopify stores' median conversion rate is 1.5% on mobile and 1.4% on desktop.
The median Shopify store gets 67% of its sessions from mobile phones.
The median Shopify store's add-to-cart rate is 6.8% of sessions.
At the median Shopify store, 49% of sessions that start checkout go on to purchase.
UK retail session conversion rates in September 2026 ranged from 0.44% (Baby & Child) to 5.76% (Arts & Crafts) across 10 markets.
UK fashion, clothing and accessories retailers converted 1.62% of sessions in September 2026.
Website conversion rates fell 6.1% year over year in 2024, according to Contentsquare.
40% of all online visits in 2024 showed signs of user frustration (e.g. rage clicks).
56.4% of US online holiday transactions in 2025 were made on a smartphone, up from 54.5% in 2024.
Why shoppers leave, and how much better checkout design could recover.
The average documented online shopping cart abandonment rate is 70.22%.
42% of US online shoppers abandoned a cart in the past quarter because they were 'just browsing / not ready to buy'.
Extra costs that were too high (shipping, tax, fees) are the top reason for checkout abandonment, cited by 40% of shoppers who abandoned for non-browsing reasons.
20% of shoppers abandoned an order because delivery was too slow.
19% of shoppers abandoned because they didn't trust the site with their credit card information.
18% of shoppers abandoned an order because the site wanted them to create an account.
17% of shoppers abandoned an order because the checkout process was too long or complicated.
12% of shoppers abandoned because they couldn't see or calculate the total order cost up front.
The average large ecommerce site can increase its conversion rate by 35.26% through better checkout design alone.
The average US checkout shows 23.48 form elements by default (14.88 form fields), while an ideal checkout needs 12-14 elements (7-8 fields).
At the median Shopify store, 79% of sessions that add to cart don't purchase.
What faster pages did to conversion in measured studies and A/B tests.
A 0.1-second improvement in mobile site speed was associated with an 8.4% increase in retail conversions.
A 0.1-second improvement in mobile site speed was associated with a 10.1% increase in travel conversions.
Retail consumers spent 9.2% more (average order value) when mobile sites were 0.1 seconds faster.
Luxury sites saw an 8% increase in page views per session with a 0.1-second mobile speed improvement.
As mobile page load time goes from 1 to 3 seconds, the probability of a bounce increases 32%; from 1 to 10 seconds, it increases 123%.
B2B lead-gen sites that load in 1 second have conversion rates 3x higher than sites that load in 5 seconds.
Ecommerce sites that load in 1 second have conversion rates 2.5x higher than sites that load in 5 seconds.
In a Vodafone A/B test, a 31% improvement in Largest Contentful Paint led to 8% more sales.
In a Rakuten 24 A/B test, optimizing Core Web Vitals raised conversion rate by 33.13% and revenue per visitor by 53.37%.
How reviews change what people buy.
A product with five reviews is 270% more likely to be purchased than a product with no reviews.
Displaying reviews increased conversion 380% for higher-priced products versus 190% for lower-priced products.
Purchase likelihood typically peaks for products rated 4.0-4.7 stars and falls as ratings approach 5.0.
95% of consumers say they regularly read product reviews as part of their shopping journey.
Only 43% of consumers say they would buy a product with zero ratings or reviews.
Product pages with at least one review convert 52.2% better than pages with none.
Products with 101 or more reviews see conversion rates more than 250% higher than products with no reviews.
49% of consumers trust online reviews as much as personal recommendations.
Shipping, returns and buy now, pay later.
Buy now, pay later drove $20 billion in US online holiday spending in 2025, up 9.8% year over year.
Smartphones drove 82.2% of BNPL purchases in the 2025 US holiday season.
Online BNPL spending grew from $2.2 billion in 2014 to $342 billion in 2024 worldwide.
Digital wallets accounted for 53% of global ecommerce spending in 2024.
37% of US holiday shoppers cite free shipping or shipping promotions as a factor in where they shop.
82% of consumers say free returns are an important consideration when shopping online.
An estimated 19.3% of online sales will be returned in 2025.
Landing pages, forms and free trials.
The median landing page conversion rate across all industries is 6.6%.
The median SaaS landing page converts 3.8%, 42% below the all-industry median.
For SaaS landing pages, email traffic converts at a median 16.9%, versus 4.1% for paid search and 2.9% for paid social.
83% of landing page visits were on mobile devices, yet desktop converts 8% better than mobile.
Landing pages written at a 5th-7th grade reading level convert at 11.1%, versus 5.3% for professional-level writing.
About 52% of people who start filling in an online form complete it (55.5% on desktop, 47.5% on mobile).
Password fields have the highest abandonment of common form fields, at a 10.50% mean abandonment rate (email 6.41%, phone 6.28%).
Mobile subscription apps with trials of 17-32 days have the highest median trial-to-paid conversion, at 45.7%.
Hard-paywall subscription apps convert 12.11% of downloads to paid by day 35, versus 2.18% for freemium apps.
Survey and platform data on how often teams test and what holds them back.
51% of marketers who A/B test say limited traffic for statistically significant results is their main challenge.
47% of marketers who A/B test cite lack of resources (time, tools or budget) as a challenge.
84% of marketers who A/B test run tests at least monthly; 38% run them weekly and 22% daily.
Websites (60%) and email (57%) are the channels marketers most often A/B test; only 32% test landing pages.
Conversion rate (45%) and customer retention rate (46%) are the metrics marketers rate most important for judging A/B tests.
To be in the top 10% of experiment velocity, companies need to run around 200 tests a year (16+ per month).
These get repeated a lot. Here's where they really come from.
The line usually traces to a 2011 AppSumo guest post with no method. The current, citable version is Optimizely's 2023 benchmark: about 12% of 127,000+ experiments won.
From a 2014 in-app survey of VWO users with no sample size and no definition of a win. Use Optimizely's or Kohavi's figures instead.
The one-third success rate is Microsoft's, from 2009. The same table puts Bing at 15% and Airbnb search at 8%, so it isn't a typical rate.
Most of Google's 700,000+ Search experiments in 2023 were rater tests and side-by-sides. Live traffic experiments numbered 16,871.
A 2016 Google/DoubleClick figure whose original page no longer loads. Google's 2020 Milliseconds Make Millions study is the better source.
Live Shopify data from October 2026 (Littledata, 436 stores) puts the median at 1.5% on mobile and 1.4% on desktop.
Use any statistic here with a link to this page or to the stat itself (each has a # link). For figures from other organizations, please cite the original source too.
LiftArchive, “A/B Testing and CRO Statistics for 2026,” https://liftarchive.com/statistics/ (November 2026).We opened every source ourselves and copied each figure as the source states it, with its sample and year. We prefer original studies over roundups, flag vendor data and older studies, and leave out figures we couldn't trace (see the list above). Our own figures come from the tests in the LiftArchive database; medians need at least 20 results.
It depends on the company. Kohavi, Deng and Vermeer (2022) report about 33% at Microsoft, 15% at Bing, roughly 10% at Booking.com, Google Ads and Netflix, and 8% for Airbnb search. Optimizely's 2023 benchmark of 127,000+ experiments found about 12% won.
The median Shopify store converted 1.4% of sessions into a purchase in Littledata's October 2026 benchmark of 436 stores. The mean was 1.9%, and the top 20% of stores were above 2.7%.
Baymard Institute's average of 50 published studies is 70.22%.
Unbounce's benchmark of 41,000+ landing pages puts the all-industry median at 6.6%, and the SaaS median at 3.8%.
Because they are tests people chose to share. 96% of the 321 public ecommerce tests in LiftArchive are wins, while companies that count every test they run report win rates of about 8% to 33%.