Across the 62,496 rows, the categorical columns have these distinct-value counts (from the two profiling queries above): - **WEBSITE** has 11 distinct values, so it is the low-cardinality classifier for grouping. - **YEAR** has 11 distinct values. It is stored as a BIGINT but works as a category or time axis, covering 2014–2024 per the card. - **PAGE PATH** has 37,410 distinct values. - **PAGE URL** has 38,916 distinct values. That is more than PAGE PATH, so the same path appears under more than one URL. This is likely because the URLs differ by website or domain. - **_batch_id**, **_source_file** and **_source_system** each have just 1 distinct value. They are constant ingestion metadata and carry no analytical signal. PAGE PATH and PAGE URL are near-unique, at roughly 60% of the row count. They behave like page identifiers rather than useful grouping categories. For breakdowns, WEBSITE and YEAR are the practical dimensions.