The biggest problem I ran into is that the largest segment of user growth were too fickle. They wanted all kinds of magic in new optional features for their personal preferences that took incredible effort. I lacked the analytics to see who used which exotic features. Most of these people just wanted a beautifier more that a diff tool and would drop you in a heartbeat for more popular tools that wouldn’t do what they wanted but were popular.
The tool I wrote did have a strong following mostly around markup language parsing that was not at all exotic but solved problems other tools refused to approach.
My guidance is don’t become a code beautifier. In the languages I was supporting during the time frame I was supporting this code beautifiers were all the rage. Nobody seemed to want a diff tool with extra capabilities. Stick to being a diff tool. The people that are intentionally looking for intelligent diff tools tend to be more engineering focused and make for a loyal audience. People looking for code vanity are just the same as window shoppers walking down a street.