The appeal and the pitfall of JSON is just that it's less broken down than tables. Suppose you built a "contact database," but all it was was the scanned image

The appeal and the pitfall of JSON is just that it's less broken down than tables. Suppose you built a "contact database," but all it was was the scanned images of people's business cards. You could flip through pages of them, but you can't do searches. But the input is so easy, just scan and save. And the output is likewise easy, just embed the images in a web page. 1. Running optical character recognition would be the first step in breaking down these cards into parsable information. 2. Converting the text into JSON and storing it in MongoDB would be taking it to the next level. 3. Breaking down the JSON into a set of tables would turn it into the most flexible and useful form --- but it's also the most work. You don't have to take all your data all the way to tables. For example, I leave my website log files as text. Grep and awk are usually all I need. I don't usually import them into tables --- although I have a few times for easier analysis! Although you don't have to turn everything into tables, I would say there are several applications out there where the developer should have gone all the way and not stopped at JSON.