#2 would probably be the main draw for me

I currently have about 1.5 million unique users per day that perform client-side operations that result in modifications to their personal "tree" of data. So performance is definitely an issue. Average user modifies their information about 80 times a day and the data I'm gathering from them has enabled me to do some optimizations that preclude the need to worry about this issue in the short term. In about a month, the people offering to buy this service from me predict we'll need better methods since volume is roughly doubling per week.

Ask me a week ago and I'd be more stressed out about how I was going to solve this since it was just a personal project. I also wasn't sure how I was going to make rent

Now I can hire people and I have an unbelievably huge offer that I'm considering taking for acquisition of this service.

If I don't accept, it'll be because retiring is less important to me than remaining an active industry-shaping force in a field I am passionate about, and I'll be hiring several full time people to take this workload off my hands.

Wow, that escalated quickly! 

Good stuff!  I'm very glad to hear it made such an incredibly positive turn.

And yes, with that many users and data permutations you'll want to seriously rethink your data structures and representation. I wish you all the best and hope it'll continue growing steadily!

In case anyone is interested (probably not) I ended up declining the offer, hiring some staff to expand production. Then while we were shopping for an office space, the original offerers came back with a 180% hike to their offer (almost tripled).

I caved. As fun as the project was and as much as I was loving having correspondence with big people in the industry that I've admired for years, it's time to provide for my family.. and extended family.. and friends.. for multiple lifetimes!

For anyone reading this. You can place json or xml files just fine into your resource folder next to your html and png files ON THE WEBSERVER. Then you load it using xhr. Set cache timeout to something sensible. User data goes to local data storage "COOKIES". This is basic stuff. There is no: xml vs json. XML has some serious schema definition stuff and xpath and namespaces. If you sure you do not need this or want to reinvent this in json, just go ahead.