The Power of Experimentation

Abstract

Organizations have long run experiments to test the appeal of new products, and in today’s digital economy, running speedy, low-cost experiments has never been easier. Firms like Alibaba, Facebook, and Google conduct tens of thousands of experiments each year, testing the impact of small changes to algorithms, user features, and web design. While each experiment may be small, the cumulative impact is not. For most companies, experimentation is not yet a deep and distributed capability. Here’s why. The vast majority of employees don’t have the latitude to launch and run small-scale experiments. In most organizations, the ability to design and run trials remains the province of specialists in R&D, data science, or product marketing. Even for employees in those functions, doing anything more than a narrow A/B test usually requires management approval. It’s not surprising that, in our survey of 10,000 Harvard Business Review readers, 61 percent of respondents from large companies said it’s “very difficult” for frontline employee to try something new when doing so requires a small team and a bit of seed funding. Another 34 percent said that bottom-up experiments are possible only THE POWER OF EXPERIMENTATION