Exactly, the "correct solution" would be my first attempt as it uses the (clustered) index scan to count pages/records, it's likely to be the fastest option whereas the first method should not even be an option. This should be immediate for someone who understands how to develop in virtually any RDBMS.

This is also, more broadly, something I've been thinking about recently. It feels (to me) like industry knowledge is getting lost somewhere, because you see people trying to reinvent the wheel in a lot of technical areas. One of them is data: there are proven approaches to most problems faced by the average organization, especially when it comes to designing and managing a data platform. Yet, you see people rolling their own approaches/designs/methodologies to basic things like ingestion, ETL and data modelling when the traditional approaches would suffice and would take half the time to implement and one third of the effort to maintain. It's great that people try to innovate, but what I see regularly is more like people trying to solve a solved problem without bothering to educate themselves on how it was solved so far.