Etsy has improved its experimentation process by using a method called CUPAC, which builds on the earlier CUPED approach. CUPAC has significantly reduced the average experiment duration by about three days, allowing for quicker insights and more frequent testing. The original CUPED implementation achieved a 7% variance reduction across experiments, but CUPAC has ramped this up to an average of 27%. This means that if a typical experiment was expected to last ten days, it can now wrap up in just seven, enabling teams to conduct ten or more additional experiments each year. CUPED and CUPAC work by leveraging historical data to enhance the accuracy of experiment outcomes, reducing the noise that can skew results. Specifically, CUPED uses pre-experiment visitor data to better estimate metrics like conversion rates. By applying CUPAC, which further reduces variance, Etsy can achieve the same statistical power with a smaller sample size, directly impacting the speed of learning in product development. This shift is critical for Etsy, as the platform continuously seeks to enhance user experience for both buyers and sellers. The advancements in these methods not only streamline the experimentation process but also emphasize Etsy’s commitment to using data-driven insights to refine its offerings. The changes in methodology reflect a broader trend toward efficiency in tech-driven companies, where the speed of iteration can lead to better products and services.