**The most surprising finding is that several sites report more entrances than page views, which should be impossible.** Every entrance is the first page view of a session, so entrances can never exceed page views. The site-level summary I ran says otherwise: - **gis.brla.gov:** 5,219,822 entrances against 897,623 page views, about 5.8 times as many entrances as views. Its maximum single-page entrances (1,680,247) is also higher than its maximum single-page views (110,198). - **budget.brla.gov:** 171,623 entrances against 126,126 page views, about 1.4 times as many. - **checkbook.brla.gov:** the two are nearly equal (181,053 entrances vs 181,345 views), and the site has only 3 rows. Its bounce rate and exit rate are both exactly 1.0 on every row, so it looks like a different unit or a placeholder. This matters because the dataset is meant to guide which entry pages to optimize and which pages to fix for bounce or exit rates. For these sites, entrance counts and probably the rates derived from them can't be trusted at face value. Other oddities in the same summary point to inconsistent measurement across sites: - **city.brla.gov** has just 404 rows but 720.8M page views, about 3 times the 237.2M from www.brla.gov, which has 24,597 rows. It averages about 1.78M views per page row and has a median of 2,088. The legacy-looking `city` site apparently dwarfs the main site, which cuts against the idea that traffic has shifted to newer sites. - **311.brla.gov** totals exactly 700,000 page views over 35 rows, an average of exactly 20,000.0. That is suspiciously round for organic data and suggests a cap, an estimate, or a manually entered figure. - **Rate scales differ by site.** gisdata.brla.gov has a maximum bounce rate of 2.0 and a maximum exit rate of 1.0, while most other sites reach 100. Those two sites seem to be on a 0–1 scale rather than a percentage, so averaging the rates across sites would be misleading. I verified this only at the site level; I didn't check individual page rows. The next step would be to pull the specific gis.brla.gov pages where entrances exceed page views and see whether they come from a particular year or ingestion batch.