How would you design the experiment itself? What would you define as a good testing process / culture? What are the key metrics you’d pay attention to? How will you define success?
Sometimes I get questions from people about my work. If relevant to a wider audience, I also post the answers here :)
A good testing process involves all stakeholders, is structured and facilitates an easy flow of the experiments from ideation to reporting. At Euroflorist I implemented an online tool to which everyone has access and involves them in the experiment process. It also provides a clear overview of all past research, learnings and reports making it an essential database of our customer knowledge.
Roughly the experiment process follows these stages:
Idea → Hypothesis → Complete test plan → Designing → Development → Q&A → Approval → Live → Analysis → Report.
See also slides 48–54 In my presentation on building an optimization dream team.
Key metrics for experiments will mainly follow company wide KPIs and the micro conversions that lead to these main KPIs:
- Immediate goal: Does the user reach the “next step”?
- Session goal: Does the user end up using the product/ creating an account?
- Lifetime goal: Does the user return X times to use the product (at least until the “aha” moment). How many new users start using the product through this user?
Important elements for a good optimization culture are:
- A broad data literacy
- Leadership & trust
- An inquisitive, questioning culture
- Iterative, learning culture
- Share learnings widely
- Clear north star goal
For more on culture, see my presentation on Optimization Culture.
In the end success will be determined in how much we contribute to the main goal(s) of the company.
Ideas on the above? Let me know in the comments!