**The most surprising finding is that this funding dataset's dollar total is driven by a handful of mega-deals, and the largest "deal" looks like a data error.** - **Typical deal vs. total.** The median deal is only about **$1.7M** (median $1.70M over the 2,065 parseable amounts, $1.725M over the 2,066 amounts present). Yet the amounts add up to roughly **$38.06B**. - **Concentration.** Only **73 deals of $100M or more** make up about **63.6%** of all dollars. That is roughly 3.5% of the 2,065 deals. Averages or totals by city, sector or year will mostly reflect a few Flipkart, Paytm and Udaan rounds rather than typical startup funding. - **Suspicious top entry.** The single largest record is **Rapido Bike Taxi, Series B, 27/08/2019, $3.9 billion**. It is more than the 2017 Flipkart round of $2.5B, and alone it is **about 10.2%** of all dollars in the dataset. A $3.9B Series B for a bike-taxi startup is implausible. From my own recollection, not from this data, Rapido's Series B was about $3.9 million, so this looks like a 1,000× unit or entry error. I have not verified this against the dataset. If it is an error, it inflates the total and every aggregate built on it. - **Missing and unparseable amounts.** Only **2,066 of 3,044 rows (about 68%)** have an amount, so about 978 deals have none. One row (Drums Food, 21/07/2016) came back as non-numeric, which is why my last summary query returned NaN for the total and max. Any "total funding" figure covers only about two-thirds of the deals. **Takeaway:** for benchmarks such as ticket size by round or sector, use the median rather than the mean. Check the very large amounts, starting with Rapido's, before trusting any dollar totals. The three queries above back these numbers: the top-deals list, the concentration summary (total, median, top share, 100M+ share), and the row-coverage summary.