This year I tried to save lettuce seeds from my garden.
We grew a few plants we liked, let them go so seed, and then harvested the seeds.
I found a few different how-tos on what to do but the first three times I tired it, the seeds wouldn't germinate when I tested them. I was even about to test the test to see if I screwed up the testing process itself (say that three times fast).
Then I changed how I harvested the seeds and the fourth batch germinated well and the fifth batch was a perfect 100%.
The root problem was that some of the advice was harvesting the seeds too early so they weren't viable. Following different advice from other places, I stumbled across a better method that worked. This other method takes a bit longer and more space but it produced the results I needed.
This doesn't mean you shouldn't follow advice of people online. It means you should test advice before you consider it good. The advice could even be okay but you'll find a better option by testing.
That's why you should always be testing something and trying to improve something in your store. Even if you can't do an official Statistically Significant A/B Split Test (TM), you can still try various things and see what works well for you.
It is absolutely fine to test two different approaches and pick the one that you like more. Your "like" can be whatever reason that's important to you, it doesn't have to be more orders, profit, or the regular goals. It could be the easier to understand option, the one that makes you happy, or even the most silly option. It might not be the most profitable option, but there's more to life than just profit.
If you need help measuring the impact of your experiments, Repeat Customer Insights takes a very long-term approach to metrics by focusing on the behavior of repeat customers.
Analyze your customer's behaviors before they defect
Your customers aren't yours forever. Some might have defected today, never to be seen again.
You need to analyze your customer behavior so you can reach them before they defect.