Your article is informative, but some additional historical context is helpful, and there's at least one stop on the train that’s missing- the flat file databases that ran on PCs during the mid 80s/early 90s.

But before that, during the late 50s, 60s and 70s, if you wanted a new report on the data stored in the proprietary ISAM database on your mainframe, you filed a request that got sent down to "Information Systems." A Systems Analyst looked at the request and then a programmer would go implement the report. A week later the report arrived via interoffice mail. You eagerly opened the report, examined the dot-matrix output (on alternating lines of green and white paper), and immediately decided that you wanted additional data. Rinse and repeat. Wait another week. You now have your new report in the form you want, but it took two weeks (or longer). The "good" news is that now that the report is designed, someone could simply add it to a menu screen somewhere and give you access, thus enabling you to "order" your report whenever you wanted, in which case you'd get it several hours later (if not the next day).

One of the key goals of SQL came out of the desire to allow business users to make their own queries, greatly reducing the latency. As it took-off in the 80s, SQL was not without competition. There were several ISAM-like databases which dominated the mid 80s-early 90s which provided easier ways to retrieve data. Paradox had QBE (which was a pretty cool concept that I could totally see being re-introduced), dBase and its litany of clones had a full-fledged programming language, and the most amazing one from that time was an office product called "QA", which included a database that had Natural Language Processing. You could tell it things like, "Give me all the stock accounts that perform more than 100 transactions a year".