Now that you’ve done the hard work of measuring sentiment through PFM surveys and monitoring the product usage of your On-the-Fence users, it’s time to document your insights, develop some hypotheses, and test those hypotheses.
It’s important to remember that a hypothesis isn’t just an educated guess. A hypothesis should be measurable, testable, and based on data you have at hand.
It's crucial to create testable hypotheses that accurately reflect your goals. Often, however, marketers develop vague hypotheses that lack specificity. For instance, stating that "we need new CTA copy for our website" isn’t insufficient since it fails to identify the problem with the existing copy and what you intend to achieve.
<aside> 💡 To create a better hypothesis, you must make it specific and testable.
</aside>
For example, "changing the CTA copy from 'Submit' to 'Join Our Community' will increase sign-ups by 5%." This hypothesis is testable, and you can validate it by modifying the CTA copy and measuring its impact on the sign-up rate. If you observe an increase in sign-ups, your hypothesis is correct, showing that users require more clarity. On the other hand, if you observe no change or a decrease in sign-ups, your hypothesis is incorrect.
Using this data, you can refine your hypothesis and test another variation, such as changing the CTA copy to "Get Started" and observe how users respond. Creating specific and testable hypotheses is critical for optimizing your marketing strategies and achieving your goals.
Selecting the appropriate metrics to evaluate your marketing experiments is essential to determining the success of your campaigns. Thankfully, if you’ve written your hypothesis well, you should have a core metrics already at hand.
In the example above, our core metric is “sign ups” and our experiments (changing the CTA copy) are designed to move the needle on that one metric.
This should make the importance of designing your hypotheses carefully; if you don’t focus on your highest leverage points, you may end up experimenting with and even improving metrics that don’t matter all that much for your business.
<aside> 💡 “Leverage” in this context refers to actions or activities that give you the highest return for your efforts. In other words, you should focus on identifying the one action or activity that will make the biggest impact on your business goals. Your hypothesis should be built to experiment with actions that move the core metric associated with that activity.
</aside>
Finally, one important rule to remember when selecting metrics is to focus on one core metric at a time. This makes evaluating the performance of your experiment straight forward and clear which allows you to iterate faster.
Remember: metrics should be specific and time bound so that you can actually measure the success of your experiment.
With a testable hypothesis and clear, specific metrics in hand, you’re ready to ideate experiments.
For many, this is the fun part and is an inherently creative process.
You and your team should gather together, in person or virtually, and review the data, hypothesis, and metrics. After this, it’s time to come up with ideas for experiments that you and your team believe will both test your hypothesis and move the needle on the metric you’ve selected.