课程: Agile Requirements Foundations

Experiments

课程: Agile Requirements Foundations

Experiments

“

- Have you ever been in a situation where you have multiple options and your research, data and even your gut instinct don't point in one direction? Experiments can provide a process driven approach to solving for this ambiguity. Experiments help us get away from fixed requirements where unknowns are high, and yet still move forward to provide real outcomes and value. Experiments help teams learn while delivering, rather than simply cranking out features that may not deliver value. When a product feature is being discussed and the team isn't sure what approach to take, the team needs to quickly get from idea spin mode to action. Action can then mean building the wrong thing, or it can mean quickly building an experiment to address the critical unknowns, that will give the team the insight to move forward. Experiments involve creating a hypothesis and reflecting on the outcome. The results are learning that the team uses to either plan more experiments, or move forward with designing and developing a future product. To design an experiment first we need to frame up the hypothesis. We do this by using a statement similar to this. Next, design the leanest experiment possible to test the hypothesis. It's easy to both over and under-engineer this part. So, think carefully about how robust it needs to be and which pieces matter most. Then make sure to identify the assumptions that you have, the things you're looking to measure, and what outcomes you expect. Then conduct the experiment and evaluate the results. Lets walk through an example. In this example, we're working with our online coffee store case study. The team thinks there may be a requirement for the testimonials to be a more prominent part of each page. The hypothesis would look like this. With this hypothesis I can plan an experiment to build a few pages with proposed testimonial enhancements. And I'll want to make sure I have the metrics tracking in place and run the experiment. Experiments can easily get out of hand in scope, so we want to make them quick, cheap, and to the point. We want to really know what we're trying to learn and don't over-engineer the experiment to miss out on critical time. Experiments can be amazing tools to provide real-time customer insights. They can reduce ambiguity and point your team in the right direction.

内容