When to test and when to trust.
It's tough to argue that a gut feeling can beat hard data, but sometimes that's exactly the case. The challenge for marketers is to know which assumptions to test and to what extent.
Logic (and maybe your boss or client) would tell you to test every hunch to make sure you're correct, but what if you instinctively know you're correct? Do you test anyway to "prove" your idea? Can you tell a client to chill out and trust you?
Here's what I do as a happy medium. (Number 3 is most important.)
1. Form a hypothesis
This is based on industry insight, research or personal experience. If your hypothesis is that you should use Image A over Image B because it has a higher click through rate, then it's not a hypothesis. If you think that your target audience will like Image B because it appeals to a certain shared emotion, that's a hypothesis.
2. Test the hypothesis
This should be as small and fast a test as possible, but it can't be too small or too fast. For example, if you're spending $60/day on Facebook ads, maybe take $10 each day for a week to test something out.
3. Consider results logically… not objectively
Data isn’t a final answer and can't prove a hypothesis on its own; it's only information. Why are you seeing the results you're seeing? What does the data tell you if you read between the lines? What clues does the data give you towards proving or disproving your hypothesis?
For example, if Image A has a high click through rate, but Image B is close behind, the data would say Image A "wins," while you logically understand they're both doing well. If your hypothesis is that Image B appeals to an emotion, the data has started proving that assumption.
You've tested, but you didn't live and die by the results, and you didn't spend unnecessary time doing it. You can keep moving.
When data is used in this way- as part of a feedback loop- it allows you space to find real insight. When data is used objectively, it eventually tends towards the lowest common denominator.
This method isn't meant as a one off experiment. This is a philosophy towards data and marketing in the current climate.
I love Gary Vaynerchuk's concept of "marketing in the year we actually live in," and I don't think it's possible to market for 2017 without calculated assumptions, educated predictions and bold moves. Data is an amazing tool, but if it always worked, computers would be doing 100% of our marketing already.