A group of respected statisticians say that they can predict the success of a new Kickstarter with 84% accuracy with 36 hours of it's launch. Resarchers at Éccole Polytechnique Fédérale de Lausanne created statistical models using funding data and discussions on Twitter from 16,000 Kickstarter campaigns that had raised a total of $158 million. 50% of them failed.
"The scientists collected Twitter data by searching for the word 'kickstarter,' then matching tweets to the Kickstarter project using URLs included in the tweets," according to ars technica. "They also culled information from each project’s 'Backers' page to get a list of what users pledged money, and how much they had pledged collectively. The second step was time-intensive, so the authors only completed it every two days." Next they added the tweets.
When combined with financial data, the models for the first 36 hours could predict success with 84% accuracy and 87% at the end of the first six days.
The researchers drew no conclusions regarding what factors led to Kickstater success or failure.
