In trying to define a problem you can solve with experimentation, you want to start with a solid understanding of your company’s goals and how they work together. A wide variety of metrics feed into a company’s goals. One way to organize those metrics is to create a goal tree based on the hierarchy of your company’s goals. Creating a goal tree will help you look at not just one possible experiment goal, but to reflect on a variety of opportunities so you can decide which goal is best to pursue first.
Why would you want to take the time to build a goal tree? Because they’re the foundation of your ideation strategy. They’ll help you:
- Decide where to focus your optimization efforts
- Break down broad organizational imperatives into bite-sized experiments
- Find multiple avenues to generate impact through optimization
- Optimize for more granular KPIs ( or key performance indicators) and understand how these impact your business goals
- Help stakeholders grasp the value of your optimization efforts in a language they understand
You’ll only have to build your goal tree once, but you’ll return to it regularly when prioritizing your experimentation backlog or designing experiments (in fact, we’ll return to it in just a little while to inform data analysis, too). When you’re analyzing experiment results, the goal tree will also help you visualize the impact of the results on other KPIs.
The optimization program manager should be the owner of goal tree content and will be responsible for updating it if the company’s business model changes. Team members who focus on experimentation strategy and analysis will also work closely with this document.
Let’s look at a goal tree for Attic and Button, our example retail site, to help you get an idea about how to build one.
Like the hierarchy we built previously, revenue is at the top, but in this case the broad organizational goal is broken into a several more granular goals that feed into the larger overall company goal. In other words, you can increase revenue by increasing either the number of visitors to your site or by increasing the amount of revenue per visitor. Each high-level metric descends into lower-level and more specific KPIs. Each category also branches into smaller categories that should equal the original in value.
Once you’ve built a goal tree, it’s easy to see how a change at a lower stage can have a cascading influence up the “branches” to the ultimate goal at the top. Knowing how different goals work together at each level will be extremely helpful as you assemble data and begin to build a hypothesis for your experiment.