**The most surprising finding is that the dataset's "funding rounds" don't behave like distinct rounds. The same companies recur in several months with amounts that swing wildly, and April's figures look inflated or badly formatted.** - **Repeat companies.** Only 14 company names appear more than once (step 1). Several show up in different months with the same stage but very different amounts (step 4): - Pine Labs appears in all four of Jan–Apr: $20M, then $150M, then $50M, then $1.2B in April. - DealShare, Series E both times, goes from $45M in February to "$624.800,000" in April. - Xpressbees, Series F both times, goes from $300M to $577.6M. - Chargebee goes from $250M to $468.2M. - Mamaearth goes from $52M to $110.8M. - Livspace goes from $180M to "$431.6". - **April looks different.** In every cross-month pair above that includes April, the April figure is the larger one, by roughly 2x to 14x. I can't tell from this data whether April reports cumulative totals, a different unit or currency, or simply errors. Either way, treating these rows as independent rounds and summing them would badly overstate funding. - **Amount formatting is broken.** `Amount` is stored as text, and some April values use European-style separators ("$624.800,000", "$11.700,000") or have no magnitude at all ("$431.6"). - **Placeholder values are common.** The two most frequent `Amount` values in step 0 are "Undisclosed" (35 rows) and blank (25 rows). - **May is partly corrupt.** 19 May rows have no company name and no amount. The one with a value has Amount "6400000" and Stage "$Series A", which suggests the columns are misaligned. - **Exact duplicates exist.** NowPurchase (Jan, $2.4M, Seed twice) and Securden (Apr, $11.7M, Series A twice, once formatted with a stray period) appear as duplicate rows within a single month. The surprise is that the apparent funding totals are mostly an artifact of inconsistent reporting and entry. Any headline figure such as "total funding raised" or "biggest round" would be driven by April's inflated or mis-formatted rows, not by real activity, so Amount needs cleaning and de-duplication before analysis.