A Less Daunting Approach
Continuing my exploration in marketing, this time I turned my focus to thumbnails. Thumbnails, though not directly contributing to impressions, can be crucial in influencing another metric: the click-through rate from impressions. My objective was to see if the design of thumbnails could impact this rate – would it increase, stay the same, or decrease? Fortunately, the platform I used tracked these metrics for me. All I had to do was create two distinct thumbnails.
The Tale of Two Thumbnails
For this experiment, I designed two thumbnails: one adhering to my original “typical” format and another inspired by popular YouTube channels. I analyzed how these channels made their text visible and their use of colors to create my new design.


The results were inconclusive. The number of views was probably too low to calculate meaningful metrics. This prompted me to consider alternative methods for improving my thumbnails.
Can AI Do Better?
Not long ago, I came across a humorous quote: “I would like AI to do my laundry, so I have time to write and do art.” To that, I would add groceries, cooking, dishes, and dusting. If anyone knows of a model that can handle all that, please let me know!
Creating thumbnails isn’t my favorite part of video production, so simplifying this process would be a small but significant victory. I decided to put AI to the test by providing it with a colored image and observing the results.

The output wasn’t particularly impressive. What do you think? This is clearly a malfunction of the model. I would likely get very different results if I tried this again.
Exploring the Impact
I then pondered: could I upload a coloring page to an AI model and ask it to color it for me? The idea seemed both exciting and a bit intimidating. Could AI truly handle coloring images, and more importantly, would it do a better job than I could?

The result wasn’t identical to my original concept, but the essence of the image was preserved. I found it impressively good, almost eerily so. This raises a thought-provoking question: will AI create all art in the future?
In any case, my tests with marketing continue along the same lines as before. Despite my efforts, they have not been successful so far, and I do not yet know why. However, I remain optimistic and committed to refining my strategies. Each experiment brings me one step closer to understanding the complexities of marketing, and I am excited to see what insights the next tests will reveal.





Leave a comment