This is a bad idea. Imagine if you have a functioning website, and it's successful. AND it has visitors. All it would take for someone to screw with you is to get your current version of the website with your customer data and create a spoof website, by essentially forking you.
Good point, but it doesn't mean that descentralized websites are pointless. "Customer data" or other personal data should never be exposed in a public way. So yes, you should never create a "decentralized website" with personal information of anyone (besides of the site owner if he really wants to).
Decentralized websites should be treated the same way than static HTML websites. Everything what you publish there is public. If someone forks your site without your authorization, essentially it's a copyright violation, but other harm would not be done. (A CC-BY-NC-ND Creative Commons license would be an interesting option to prevent this, at least legally. It would only allow non-commercial forks, and forking without commercial intentions is kind of pointless. Even better would be a custom "no fork" license which only allows redistribution of the original, not a fork.)
So what would be the use cases? Mainly, blogs and other informational sites, and open-source/open content projects. One of the prime examples, which is often mentioned of decentralized web advocats, are "uncensorable" blogs for opposition activists in authoritarian countries.
Sites which sell something can also be built with this technology, but they cannot include customer data. Communication must be established via other means, an example could be tools like Retroshare.
