**The most surprising finding is that the "one row per company" grain doesn't hold, and the repeats suggest the April amounts are not comparable to the earlier months.** **1. Companies recur across months, and April is always the biggest figure.** - 14 company names appear more than once, and the 19 blank-name rows from the May file are one extra group (step-1). - Pine Labs appears in all four months: $20M (Jan), $150M (Feb), $50M (Mar), then **$1.2 billion** (Apr). - For every company with a disclosed April amount and an earlier amount, April is the larger figure: - DealShare: $45M in Feb, $624.8M in Apr. - Chargebee: $250M in Feb, $468.2M in Apr. - Livspace: $180M in Feb, $431.6 in Apr. - Licious: $150M in Mar, $488.7M in Apr. - Classplus: $70M in Mar, $129.5M in Apr. - Mamaearth: $52M in Jan, $110.8M in Apr. - Xpressbees: $300M in Feb, $577.6M in Apr. - The stage barely changes between those rows (e.g. DealShare is "Series E" both times, Xpressbees "Series F" both times). A company that stays in the same stage but raises 5–14× more is implausible as a series of separate rounds. - A likely explanation is that April rows are cumulative totals, a different currency conversion, or a different convention. The data can't confirm which, but summing or ranking amounts across the monthly files would be misleading. **2. The April amounts are malformed text.** - `Amount` is stored as text, and April mixes separators in ways that change the value: `$468.200,000`, `$624.800,000`, `$110.800,000`, `$129.500,000`, and the bare `$431.6`. - Securden is listed twice in April for the same round, once as `$11.700,000` and once as `$11,700,000`. NowPurchase is listed twice in January with identical $2.4M Seed rows. - Treating these as numbers needs careful cleaning, or the April values come out wrong by orders of magnitude. **3. The amounts are also often missing.** - Across the amount values I profiled (step-0), 35 rows are "Undisclosed" and 25 are blank. - The May file contains 19 rows with a blank company name and no data at all, plus one stray `6400000` / `$Series A` row where the stage and amount look misaligned (steps 4–5). This matches the card's note that some May rows are misaligned or blank. **Bottom line:** the dataset looks like a clean list of one round per company per month, but it is really a set of overlapping monthly snapshots. The same companies recur with inconsistent, much larger April figures, and many amounts are blank, undisclosed or badly formatted. Any total or ranking built on `Amount` without deduplication and reformatting would be dominated by artefacts such as Pine Labs' $1.2B.