**The most counterintuitive finding is that entrances exceed page views on a large share of pages, which should be impossible.** Every entrance (a session starting on a page) is itself a page view, so `ENTRANCES` should never exceed `PAGE VIEWS` on a row. Across the dataset it does so on about 24,648 of 62,496 page rows (about 39%), summing the per-site counts from the checks I ran. - **Widespread:** The problem is not confined to a few bad rows. - www.brla.gov has 11,830 of 24,597 rows affected. - brgov.com has 9,211 of 17,595. - data.brla.gov has 2,958 of 14,379. - **Extreme at the site level:** gis.brla.gov has 5.22M entrances against only 0.90M page views, about 5.8 entrances per view. my.brla.gov and budget.brla.gov each come to about 1.36 entrances per view. - **Plausible elsewhere:** The remaining sites have entrances well below page views, between 0.19 and 0.96 per view. checkbook.brla.gov is the exception at about 1.0, but it has only 3 rows. gisdata.brla.gov has just 2 affected rows. This probably reflects how the analytics export was built rather than real behaviour. Possible causes are page-level metrics aggregated over different filters or time windows, sampling, or entrances counted on a different basis than views. I can't tell which from the data alone. Treat entrance-based and bounce/exit-based rankings with caution. Entrance counts, bounce rate and exit rate may not be comparable with page views, particularly for gis, my and budget. Two related oddities support this: - **Skewed traffic:** city.brla.gov has only 404 rows but 720.8M page views, about three times the 237.2M across www.brla.gov's 24,597 rows. - **Suspect rate values:** checkbook.brla.gov has bounce and exit rates of exactly 1.0 on all 3 rows. gisdata.brla.gov never exceeds 2% bounce, while most other sites reach 100%. These look like a different unit or scale, such as fractions rather than percentages. The numbers come from the per-site comparison of entrances to page views that I ran earlier in this analysis.