People Think Experimentation Is Slow - Do They Have a Point?
Sep 13, 2023
auto_awesome
Discussion on the slow nature of experimentation and the benefits of going slow in the right direction. Debates on speed versus following instructions using Ikea furniture as an example. The importance of patience in experimentation and avoiding premature endings. Different perspectives on experimentation and signs of 'bullshit management'. Emphasis on providing value and learning through experimentation.
Experimentation is slower than not running a test, but it mitigates risk and ensures better decision-making.
To speed up experimentation, embrace parallelism, involve developers early on, and consider non-inferiority testing.
Deep dives
Experimentation is a calculated slowing down to validate the right direction
Experimentation objectively slows things down, but it's by design. It requires development, designs, hypotheses, research, testing, and analysis. It may be slower than not running a test, but it mitigates risk and ensures better decision-making. It helps avoid launching something that doesn't work, leading to time-consuming corrections later. Experimentation uncovers the unknown and embraces the process, recognizing that being slower but on the right track is better than rushing towards failure.
Recognizing and addressing bottleneck issues in experimentation
Sometimes, the slowdown in experimentation might not be due to the process itself but the individuals involved. It's important for experimentation practitioners to reflect on their own performance and identify if they are the bottleneck. It's crucial to manage expectations, communicate effectively, and prioritize value-driven actions that support the product team. By recognizing and addressing any personal or cultural barriers to speed, experimentation can become more efficient and effective.
Speeding up experimentation through parallelism and value focus
To speed up experimentation, it's beneficial to embrace parallelism by conducting research and testing concurrently. This ensures a continuous stream of learning and value. Planning iterations ahead of time and involving developers early on can also expedite the experimentation process. Additionally, considering non-inferiority testing can enable faster decision-making without compromising the value of experimentation. By focusing on providing value, embracing the learning process, and optimizing the overall experimentation workflow, the perception of experimentation being slow can be addressed and overcome.