Analytics that used to cost tens of thousands a year is now free, installs in five minutes, and is on a very large fraction of the web. Almost nobody who pasted that snippet in treated it as a transaction, and it is one.
What you receive is measurement you could not previously afford. What you provide is a record of who visits your site, what they read, where they came from, and where they went, delivered to a company whose main business is knowing where attention is. That is not sinister and I am not suggesting anyone has misused it. It is simply a trade, and a trade evaluated by nobody is not a trade anyone got a good price on.
The larger problem is not who holds the data. It is what having the data does to the person looking at it.
Analytics measures what is easy to count. Page views, visits, time on page, bounce rate. Those exist because they can be recorded from a browser, not because anybody decided they described something worth having. They are proxies chosen for availability.
Then the proxy becomes the target, and everything downstream deforms around it.
Optimise for page views and you split an article across six pages. Optimise for time on page and you write something slower, or you make the answer harder to find, and the number improves. Optimise for return visits and you start sending emails nobody asked for. In every case the measurement rises and the thing being measured, which was supposed to be whether the site is any good, gets worse.
I made almost this exact complaint about download counts a few weeks ago, and the resemblance is the point. A download is not a listener. Time on page is not attention. Both are numbers standing in for something nobody can observe, and both get treated as the thing itself because they are what the dashboard shows.
A useful discipline, and one I try to hold myself to, is to decide what you actually want before you look at the figures. Write it down first. Then find out whether anything on the dashboard relates to it, and be prepared for the answer to be that nothing does.
Measuring the wrong thing accurately is worse than measuring nothing, because it feels like knowledge and it moves decisions.