We do rapid experiments
to gather new data.
We can’t always find the answers we need in academic research or corporate data. So, we use experiments to test our hypotheses about business opportunities. In the process, we generate new data. Here are some examples:
- We needed to understand the relationship between physical activity and health-care use. So, we put accelerometers on hundreds of patients, tracked their activity, and linked it to their health-care data.
- We wanted to know whether the number of accessory styles available in a retail store influenced accessory sales. So, we removed about 50% of the accessory styles from 16 randomly selected stores and measured sales.
- We wanted to know whether pre-appointment phone calls could increase the number of patients who received preventive testing and recommended lab work. So, we randomly selected hundreds of patients with upcoming physician appointments. We called them and asked them to complete recommended tests before their appointments, and we measured completion rates.
- We wanted to know which order form would sell the most tech-support services with computer purchases. So, we designed four different forms, started using them in randomly selected stores, and measured tech-support sales.
No matter what we’re trying to learn, our experiments have several things in common:
- They are scientifically designed.
- They are measured with reliable data.
- They are quick, iterative, and inexpensive.
- They don’t interfere with normal business operations.
Sometimes these experiments lead to big ideas for new products. Sometimes they lead to more questions, more research, and more experiments. In every case, they provide actionable data and a deeper understanding of our partners’ businesses.