Introduce React (the wrong way)
We (at the company) were at the weird stage where we have some jQuery-glued libraries that were our ‘framework’ driving frontend. I think at that time it was often the case and usually, you made the call to jump into the Angular or Ember.js bandwagon. React was new on the market (I think it was around 2015) and was kinda… awkward? It was still unclear how to structure some more complex interfaces, where to handle the http layer, how to deal with the complex state, and so on.
We had to do something nevertheless, the UI was getting ridiculously complex. I did some research on what we could potentially use and we simply tried to build anything. So we rebuild a simple section in our app that had some complexity (not too much, not too little) using both Angular and React.
The requirement was to use something that didn’t bring all of its ecosystem with it and that would allow us to rewrite the application section-by-section and React felt really good.
We were quite lucky with the call. Obviously, the first version was thrown out of the window - along with the refluxjs that I added, I had no idea what I was doing - but at least it was the first step and future adaptation was quite smooth overall (but then again - I was lucky to work with very smart and resourceful people).
Now that I reflect on it the most difficult part was making the call - such technical decisions can be very difficult. What if we chose the wrong tool and we shot ourselves in the foot? Doing MVP first and iteration was a good decision - if you don’t have to rewrite your whole system but can adapt technology on the go - so you can control the process with the tech and not the other ways around - means you probably should be a-okey.