Distinct counts for the five categorical columns are below, per monthly table. The tables share no join key, so I did not compute a combined figure across months. The rendered result (step-2) shows the numbers. | Column | Jan | Feb | Mar | Apr | May | |---|---|---|---|---|---| | Company Name | 114 | 96 | 98 | 94 | 24 | | Location | 11 | 13 | 17 | 39 | 16 | | Sector | 42 | 39 | 54 | 58 | 19 | | Stage | 12 | 11 | 11 | 18 | 6 | | Investor | 107 | 86 | 90 | 54 | 22 | - **Company Name:** it is nearly unique per row in January to April. January has 115 rows and 114 distinct names, and April has 95 rows and 94, so each has one repeated name. May has only 24 distinct names across 42 rows. This fits the card's note that some May rows are misaligned or blank. - **Sector:** this is the highest-cardinality true category, at 39 to 58 values in Jan–Apr and 19 in May. Sector labels are probably inconsistent and fine-grained. - **Location:** this rises from 11 values in January to 39 in April, so April covers many more cities. - **Stage:** this is the smallest vocabulary, at 11 to 18 values in Jan–Apr and 6 in May. - **Investor:** January has 107 distinct values across 115 rows, so it is close to free text. April drops to 54 across 95 rows. I only profiled these five columns. Founders, What it does and Amount are free text or text-typed numbers, so I did not treat them as categorical.