CRO statistics
These conversion rate optimization statistics come only from primary sources: the original study, report or paper, opened and checked, with its year and what it actually measured. Most CRO stats pages repeat numbers from other stats pages. This one lists fewer figures, and you can check every one.
The conversion rate optimization statistics most worth knowing come from their original sources: Baymard Institute's average documented cart abandonment rate is 70.22%, extra costs revealed late are the most common avoidable reason, and Optimizely's data on more than 127,000 experiments shows around 12% winning on their primary metric. Use them as context, then measure your own baseline.
Each CRO statistic on this page was found in its original source, not copied from another list. The source is linked, the year is shown, and the note says what the number actually measures: a survey answer, a model estimate, a vendor's own customer data or a controlled experiment. Where the popular version of a statistic differs from the original, the original wins.
The ten conversion rate optimization statistics most worth knowing: about 70% of online carts are abandoned, extra costs are the top avoidable reason, most A/B tests do not win, small speed gains are associated with more conversions, and reviews raise purchase likelihood. Sources and years are in the full tables below.
Cart abandonment statistics come mostly from Baymard Institute, which averages published abandonment studies and runs its own shopper surveys and checkout usability benchmarks. The average documented abandonment rate is 70.22%, and the most common avoidable reason is extra costs revealed late. Many causes are design problems a store can fix.
| Statistic | Source and year | What it measures |
|---|---|---|
| 70.22% average documented online shopping cart abandonment rate | Baymard Institute, cart abandonment rate list, last updated 2025 | An average of 50 third-party studies dated 2006 to 2025, with different methods |
| 42% of US online shoppers have abandoned a cart because "I was just browsing / not ready to buy" | Baymard Institute, 2025 | Self-reported survey of US online shoppers |
| 40% cited extra costs too high (shipping, tax, fees) | Baymard Institute, 2025 | Share of abandoners, excluding the "just browsing" group. The reasons below are on the same basis |
| 20% said delivery was too slow | Baymard Institute, 2025 | Same survey |
| 19% did not trust the site with their credit card information | Baymard Institute, 2025 | Same survey |
| 18% left because the site wanted them to create an account | Baymard Institute, Checkout UX benchmark, 2025 | Survey of 1,026 US adults who had shopped online in the previous 3 months |
| 17% cited a too long or complicated checkout process | Baymard Institute, 2025 | Same survey |
| 12% could not see or calculate the total order cost up front | Baymard Institute, 2025 | Same survey |
| 9% said there were not enough payment methods | Baymard Institute, 2025 | Same survey |
| 35.26% potential conversion rate increase for an average large ecommerce site through better checkout design | Baymard Institute, 2025 | Baymard's estimate from its usability research, not a measured lift on real sites |
| 23.48 form elements in the average US checkout flow (14.88 counting only form fields), against an ideal of 12 to 14 (7 to 8 fields) | Baymard Institute, 2025 | Baymard's checkout benchmark database of US sites |
| 64% of leading US and European ecommerce sites have "mediocre" or worse checkout UX, and only 2% rate "good" | Baymard Institute, Checkout UX benchmark, 2025 | Expert benchmark of leading sites against Baymard's guidelines |
| 62% of sites fail to make guest checkout the most prominent option | Baymard Institute, 2025 | Same benchmark |
| 32% of test participants did not complete at least one required field when a form marked only its optional fields | Baymard Institute, 2025 | Usability testing observation |
What to do with it: show shipping, taxes and fees before checkout, make guest checkout the obvious choice, cut fields, and label both required and optional fields. The checkout optimization guide and the guide to the cart abandonment rate go further.
Page speed statistics consistently link slower pages with fewer conversions and more bounces. The strongest evidence comes from controlled experiments at Bing; the most quoted figures come from observational studies by Google, Deloitte and Akamai, which show association rather than cause. All of them point the same way: on mobile especially, fractions of a second matter.
| Statistic | Source and year | What it measures |
|---|---|---|
| A 0.1 second mobile site speed improvement was associated with 8.4% more retail conversions and 9.2% higher retail average order value | Deloitte, "Milliseconds Make Millions", 2020 | 37 brands, 30 million sessions over 4 weeks, Europe and US. Commissioned by Google. Observational, analyzed with a regression model |
| The same improvement was associated with 10.1% more travel conversions | Deloitte, 2020 | Same study |
| As mobile page load time goes from 1 to 3 seconds, the probability of a bounce increases 32% | Google/SOASTA Research, 2017 | A neural network prediction. A relative increase in bounce probability, not a 32% bounce rate |
| From 1 to 5 seconds, the probability of a bounce increases 90%; from 1 to 10 seconds, 123% | Google/SOASTA Research, 2017 | Same model |
| A 100 millisecond delay in website load time can hurt conversion rates by 7% | Akamai, Online Retail Performance Report, 2017 | About 10 billion visits to top online retailers over one month. Correlation, from the press release |
| A two second delay in page load time increased bounce rates by 103% | Akamai, 2017 | Same data |
| At Bing, every 100 milliseconds of speed improvement increased revenue by 0.6% | Kohavi et al., "Online Controlled Experiments at Large Scale", 2013 | Controlled experiments on Bing, a search engine with very high traffic |
What to do with it: test your key pages in PageSpeed Insights and on a real phone on mobile data, fix the slowest templates first, and measure conversion before and after. See mobile conversion optimization.
The largest published landing page benchmark is Unbounce's Conversion Benchmark Report. Its 2024 edition puts the median landing page conversion rate across all industries at 6.6%, with industry medians from 3.8% to 12.3%. Each page counts its own goal, so a newsletter signup and a sales inquiry sit in the same average.
| Statistic | Source and year | What it measures |
|---|---|---|
| 6.6% median landing page conversion rate across all industries | Unbounce, Conversion Benchmark Report, 2024 | More than 41,000 landing pages, 464 million unique visitors and 57 million conversions, all on Unbounce customers' pages |
| Industry medians range from 3.8% to 12.3% | Unbounce, 2024 | Same data, split by industry |
| 83% of the landing page visits analyzed happened on mobile devices | Unbounce, 2024 | Same data |
What to do with it: use the median as context, not a target, and check your landing pages on mobile first, since that is where most of the visits in this data happened. See landing page optimization.
A/B testing statistics from the companies that run the most experiments agree on one point: most tests do not win. Optimizely's data on more than 127,000 experiments shows around 12% winning on their primary metric, and Microsoft researchers found that about one third of ideas improved the metric they targeted. Conversion rate testing works by finding the few changes that do.
| Statistic | Source and year | What it measures |
|---|---|---|
| Around 12% of experiments win on the primary metric; 88% do not | Optimizely, "Evolution of experimentation", data 2018 to 2023 | Over 127,000 experiments on Optimizely's platforms. Optimizely customers only |
| Fewer than 10% of experiments test 4 or more variants, yet those experiments are more than twice as impactful as A/B tests | Optimizely, data 2018 to 2023 | Same data |
| Only about one third of ideas tested at Microsoft improved the metrics they were designed to improve | Kohavi et al., "Online Experimentation at Microsoft", 2009 | Controlled experiments at Microsoft |
| At Google, about 10% of controlled experiments led to business changes | Kohavi et al., 2013, quoting Jim Manzi | Manzi's figure as cited in the paper |
| A Bing ad headline change increased revenue by 12% without hurting key user experience metrics | Kohavi and Thomke, Harvard Business Review, 2017 | One controlled experiment, run in 2012 |
What to do with it: expect most tests to lose, and plan enough tests to find the winners. Size each test before it starts with the A/B test sample size calculator, read it with the statistical significance calculator, and see the complete A/B testing guide.
Reviews and trust statistics show that proof changes buying behavior. Northwestern University's Spiegel Research Center found purchase likelihood for a product with five reviews was 270% greater than for one with none, with a bigger effect on higher-priced products. Stanford research found that a site's design look was the most mentioned factor when people judged credibility.
| Statistic | Source and year | What it measures |
|---|---|---|
| Purchase likelihood for a product with five reviews is 270% greater than for a product with no reviews | Spiegel Research Center, Northwestern University, 2017 | Purchase likelihood, not sales. Data from PowerReviews, a reviews company |
| Displaying reviews increased conversion rate 190% for a lower-priced product and 380% for a higher-priced product | Spiegel Research Center, 2017 | Same research |
| Purchase likelihood typically peaks at average ratings of 4.0 to 4.7, then falls as ratings approach 5.0 | Spiegel Research Center, 2017 | Same research |
| "Verified buyer" badges on reviews improve the odds of purchase by 15% | Spiegel Research Center, 2017 | Same research |
| The "design look" of a site was mentioned in 46.1% of comments about its credibility, more than any other factor | Fogg et al., Stanford and Consumer WebWatch, 2002 | 2,684 people evaluating live websites. A share of comments, not of people |
What to do with it: show real reviews near the buying decision, including some critical ones, and make the site look as careful as your product is. See the guide to trust badges.
Usability research from Nielsen Norman Group explains why clear, scannable pages convert better: people read only a fraction of the words on a page, and a small usability test with five people finds most of a page's usability problems. Both figures are old, and both still describe how visitors behave.
| Statistic | Source and year | What it measures |
|---|---|---|
| On the average web page, users have time to read at most 28% of the words during a visit; 20% is more likely | Jakob Nielsen, Nielsen Norman Group, 2008 | An upper bound based on time on page, from an analysis of browsing data recorded from 25 users |
| A test with five participants finds about 85% of usability problems | Jakob Nielsen, Nielsen Norman Group, 2000 | A model estimate. The same article says finding all problems takes at least 15 users, and recommends several small tests over one large one |
What to do with it: write for scanning (clear headings, short paragraphs, key facts first), and run small, frequent user tests rather than one big one. See how to reduce cognitive load.
E-commerce accounted for 17.1% of total US retail sales in the second quarter of 2026, seasonally adjusted, according to the US Census Bureau's quarterly estimate released on August 18, 2026. That share is the reason small conversion gains on a store add up: a growing share of buying decisions happens on a web page.
| Statistic | Source and year | What it measures |
|---|---|---|
| 17.1% of total US retail sales were e-commerce in Q2 2026 | US Census Bureau, Quarterly Retail E-Commerce Sales, released 2026 | Seasonally adjusted. Estimates are revised with each quarterly release |
First-hand data
Three results from Convertica's own client tests, each with its metric, timeframe and source from the original case study. They are single tests on single sites, not benchmarks, and no result is typical.
Lead generation
A website marketplace's valuation tool asked for a lot of typed input right after the first click. A rebuilt, step-by-step submission process was split-URL tested, then rolled out.
+51.6%
Conversions (valuation tool submissions), 47 days
Source: Empire Flippers, valuation tool submissions, 47 days, split-URL test then rollout
Read the case study: Empire Flippers: a valuation tool submission process rebuilt
eCommerce
A health supplement store wanted more auto-ship subscriptions and a better return on ad spend. Two split tests ran one after the other, the second against the first winner.
+101.82%
Revenue, cumulative across two sequential split tests, 7 December to 24 March
Source: JustThrive, revenue, two sequential split tests, Dec to Mar
Read the case study: JustThrive: two sequential split tests on a supplement store
Membership eCommerce
A membership store showed locked content in its search results with no clear next step. Adding calls to action and access levels to the results was A/B tested on desktop and mobile.
+49.4%
Paid membership sign-ups, 17-day A/B test
Source: dScryb, paid membership sign-ups, 17-day A/B test, 98% confidence
Read the case study: dScryb: more paid membership sign-ups from search results
Several CRO statistics that appear on most stats pages are not listed here, because the original could not be opened or does not say what is quoted. Where a figure is widely misquoted, the version in the original source is the one in the tables above.
Use conversion rate statistics as context for questions, not as targets. An average mixes sites, goals, devices and years that may look nothing like yours. Before acting on any figure, check what was measured, on whose sites, how it was measured, and when. Then measure your own baseline, which is the only number you can actually improve.
Every conversion rate optimization statistic on this page comes from the original source: Baymard Institute, Deloitte, Google, Akamai, Unbounce, Optimizely, Microsoft research papers, Harvard Business Review, Northwestern University's Spiegel Research Center, Nielsen Norman Group, Stanford and the US Census Bureau. Each row links to the source and gives its year.
Baymard Institute calculates an average documented online shopping cart abandonment rate of 70.22%, based on 50 studies (list last updated September 22, 2025). The studies use different methods and years, so treat it as a broad average rather than a target.
Unbounce's 2024 Conversion Benchmark Report puts the median landing page conversion rate across all industries at 6.6%, with industry medians from 3.8% to 12.3%. It covers more than 41,000 landing pages built by Unbounce customers, each counting its own conversion goal.
Most do not. Optimizely's analysis of over 127,000 experiments run from 2018 to 2023 found that around 12% win on their primary metric. Microsoft researchers reported in 2009 that only about one third of ideas improved the metrics they were designed to improve.
A Google-commissioned Deloitte study (2020) found that a 0.1 second mobile site speed improvement was associated with 8.4% more retail conversions and 10.1% more travel conversions, across 37 brands. It is an observational study, so it shows association, not proof that speed alone caused the change.
It depends on what counts as a conversion. For a landing page, 10% is above the 6.6% cross-industry median in Unbounce's 2024 report. For a first-visit purchase on an online store it would be unusually high. Compare against your own past rate on the same traffic before anyone else's.
Because a number is only listed when the original source could be opened and the figure found in it. Many widely shared CRO statistics trace back only to other roundups, or the original no longer says what is quoted. Those are left out, and some are explained further down the page.
Figures are checked against their sources when the page is reviewed, and each row shows the year of its source. Some sources update on a schedule: the US Census Bureau publishes quarterly e-commerce estimates, and Baymard updates its abandonment average as new studies appear.
Ask Convertica
Convertica
Hi, I'm Convertica's AI assistant. Ask me anything about the free CRO audit, the full report, the one page and full site rebuilds, or any of the free tools.