I would absolutely be building my own system when I get into the game. I've built similar systems before, and they're shockingly simple until you start adding bells and whistles.
Automation is great, and it makes things very easy for everyone.
I have no interest in buying a pre-built system for a variety of reasons. First, my needs are different from anyone else's, in the same way that everyone else's needs are different from everyone else's, and I don't fancy a system that tries to cater to everyone. Similarly, I'd want features that they don't have and won't have. Similarly similarly, I know KISS, which too many software vendors seem to forget - I would know the website would work with various browsers / devices / accessibility features, and that people would easily be able to use it.
Second, I would be relying on a two-pronged method of security. One, writing this myself (I'm not a security expert, but I have a bit of formal training and a lot more self-taught) I would not be relying on outsourced code written by a code monkey somewhere who doesn't even understand a prepared statement, let alone laws / regulations / best practices about not storing full SSIDs and CC numbers and so on, or properly hashed passwords, or monitoring systems, or hell, proper testing. Two, by not using a pre-made software package, anyone who wants to attack the system would have to design and execute exploits specifically against what I've done, as opposed to a drive-by script-kiddie attack; this means that I avoid being caught up in popularized exploits against everyone who uses that system as collateral damage, and that anyone who wants to gain access needs not only a bit of specialized knowledge (versus press-the-button script-kiddie attacks) but also a good reason to target me specifically. To summarize, prong one is proper security because I understand it decently well; prong two is security through obscurity. An excellent combination.
Third, I wouldn't want to pay for the package. I also feel that a good system would give me competitive advantage over an off-the-shelf package.
Fourth, I could take care of feature requests myself. People want X? Sure, why not. See above about competitive advantage.
Fifth, it would be fun. A nice, easy, hit-it-out-of-the-park software project, as opposed to my bread and butter: "Why is this not working? Is it the kernel, the RTOS, the drivers, the other drivers, or the hardware? Or the compiler or the makefile? Silicon bug, or a missing resistor? Let me sit here for hours going through various levels." It's fun to do for work, but sometimes I just want something easy :)