Articles

What the research actually says about the things an audit points out. I open and read every study I use. Each number links to where it was measured, says how (experiment, survey, industry report) and on how many people. Where I could not find a measurement, the article says that too.

• dropoffaudit.me · Articles Prices Hiding the price raised sales in one field experiment and lowered them in another Two field experiments with a real retailer moved the whole price one click later. In the online store more days had a sale; in a deals email fewer people bought. Why the two differ, what surgeons and patients say about prices online, and what nobody has measured for a clinic. 80% to 95% of days with at least one sale when the price moved one click later, randomized field experiment 2 sourcesRead
• dropoffaudit.me · Articles AI search When Google answered in AI Mode, people clicked on far fewer websites A randomized field experiment with 1,100 Google users changed only what Google showed them for a week. What AI Mode did to clicks on websites, what an earlier observational study saw, and what decides which businesses an AI answer names. 18.8 pts lower click-through rate to outside websites with AI Mode, randomized field experiment 3 sourcesRead
• dropoffaudit.me · Articles Booking links Fewer than half believed a real NHS booking link led to the NHS Three randomized experiments changed only the address of a booking link in a health email. What a clearer address did to trust, why a shorter one did not help, and what this can and cannot say about a clinic whose Book button opens another company’s website. 63% of people shown a real NHS booking link were unsure or unlikely to use it, randomized experiments 2 sourcesRead
• dropoffaudit.me · Articles Links and buttons 46% of the top million home pages have a link with no text How often links break, how many home pages have links and buttons with no words, why screen reader users rank them among their top frustrations, and what “Learn more” leaves out. 46.3% of the top million home pages had at least one link with no text, automated scan 8 sourcesRead
• dropoffaudit.me · Articles Typos and trust “One typo cuts sales in half.” I went looking for that study The claim that one spelling mistake halves online sales comes from a single anecdote. What a randomized experiment, a 1,410 person survey and a study that found less actually measured. −8.9 points of trust with five spelling mistakes, randomized experiment 4 sourcesRead
• dropoffaudit.me · Articles Reviews and choice The widely quoted “270%” review stat comes from one gift retailer A half star on Yelp, patient surveys in JAMA, and where the 270% figure comes from. Plus how a result from one clinic lost its context as it was cited. +19 percentage points: 7 pm tables sold out more often with an extra half star 5 sourcesRead
• dropoffaudit.me · Articles Copywriting “75% lean” or “25% fat”? Same meat, different verdict “75% lean” against “25% fat”, instructions that feel harder in a harder typeface, and a website that tested better once the hype was removed. +124% usability score when promotional copy became objective, concise and scannable 3 sourcesRead