May I know what was your needs that required you to code your own survey?

in psych, "coding a survey" _can_ mean "tagging qualitative survey responses against a list of traits (called the codebook)" rather than "writing computer code to present a survey". GP may have meant that although Qualtrics provides tools for this process, it was easier to just do it by hand.

guess: doing away with very poor user-interfaces. Most survey systems are downright horrible.

Guessing: privacy

I doubt this is the reason. Any academic studies which are going to be published need Institutional Review Board (IRB) approval to ensure they won't violate ethics guidelines. Qualtrics has done a lot of work to be the go-to academic survey software and meets the privacy criteria required by most if not all IRB's. Using Qualtrics for surveys a breeze to get through my institution's IRB - all they cared about was the content of the questions.

It would have been a nightmare trying to do surveys on some home-brew web server survey system. I saw something similar happen once (study needed real-time tracking of user inputs to a degree which wasn't possible with other systems), and it involved a physical security audit of the server as well as university IT being involved. Project was delayed by a few months getting it through IRB.

For what it's worth, I found Qualtrics to be a a flexible, extendable platform for my unique survey needs. The study I was working on involved people comparing 3d models. Even though an STL viewer wasn't provided by Qualtrics, I was able to use an iframe to display the models in the survey.

The simple answer is that they (mostly) serve different markets