Library

A test with about 50,000 donors found that a matching offer increased the money from each letter by 19%. A 2-to-1 or 3-to-1 match did not collect more money than a 1-to-1 match.

A random test with 1,612 staff at Trip.com found that 2 home days each week decreased quit rates by 1/3. Performance and promotions did not change over 2 years.

In a random test at large Indian textile firms, free advice on basic management increased productivity by 17% in the first year. The changes were simple. They included tracking, maintenance and stock control.

In a 9-month test at the Chinese travel company Ctrip, call center staff who worked from home performed 13% better. Half as many of them left the company.

In a study of about 20,000 coin flips, people who got a coin result for a big change made the change more often. After 6 months, the people who made the change were happier.

A study of 600,000 US homes found that letters that compared the energy use of each home with its neighbors decreased energy use by about 2%. The effect was similar to a price increase of 11% to 20%.

3 field experiments with 4,867 developers tested an AI code assistant at Microsoft, Accenture and a Fortune 100 company. Developers with the tool finished 26% more tasks.

A study with 758 BCG consultants found that AI made them faster and better on tasks that AI does well. On a task that AI does badly, consultants with AI were 19 percentage points less likely to be correct.

In an MIT experiment with 453 professionals, ChatGPT decreased the time for writing tasks by 40%. The quality of the writing increased by 18%.

A study of 5,179 support agents found that an AI assistant increased the number of problems that agents solved per hour by 14%. New and less skilled agents improved by 34%. The best agents improved very little.