Nobody has to approve this
The usual barrier to a better analytics course is not pedagogy. It is a licence quote, a security review, a lab-image change request, and a semester spent waiting on all three. None of it applies when there is nothing to install.
You want it for spring. Procurement wants a quote, IT wants a security review, and the lab wants a change window. Each of those is measured in months, and spring is measured in weeks.
The queue simply does not exist
Everything runs in a standard browser. No client to deploy, no seat count to negotiate, and no version for a student to have the wrong one of.
A student on a five-year-old school Chromebook opens the same tools as one on a new MacBook. That is not a compatibility promise, it is just what a web page is.
The three questions you will be asked, answered
“What does it cost the students?”
Nothing to install and no software to buy. They need an account, and a course code links them to your section. There is no per-seat licence to pass along in a course fee.
“Is it accessible?”
Keyboard navigation, skip links, screen-reader labelling, visible focus states and reduced motion are built in rather than roadmapped, and the position is written down where you can point at it.
“What happens to student data?”
Coursework activity, quiz attempts and progress. No proctoring, no webcam, no keystroke logging, no browsing outside the platform. The trust page states the retention position plainly, so you forward a link instead of drafting an email.
The full cost of adopting it
⚠️ The fair objection
"If it is free and needs no approval, what is the catch, and will it exist next year?"
A reasonable thing to ask of anything you build a syllabus around. The tools and the course infrastructure are free to teach with; the paid tier covers the premium pieces such as the certificate track, and that is what funds the rest. Your course content and question banks are exportable, items travel out as QTI, and student analyses export as Python. If you ever leave, you are not leaving your work behind — which is the only version of this answer worth trusting.