My idea was to use server logs too but AWStats showed more than 100% higher number of unique visitors and more than 18 times higher number in page views (both compared to Plausible numbers) so I excluded it from the study as I thought it’s very inaccurate.