**The most counterintuitive finding is that the "total funding" figure is driven by a few outliers, and the single biggest deal is probably a data error.** **1. A tiny share of deals holds most of the money.** - Of the 3,044 rows, 2,065 have a usable USD amount, totalling about **$38.06B** (step 12). The other 979 rows (32%) have no usable amount. - The **median deal is only $1.7M**, yet 73 deals of $100M or more make up **63.6% of all dollars**. That is about 3.5% of valued deals. - The average deal is therefore far above the typical one. Any "typical ticket size" built from totals or means would be badly misleading. **2. The largest deal is suspicious.** - The top record is **Rapido Bike Taxi, Series B, 27/08/2019, $3.9 billion**. That is about **10.2% of all funding in the dataset** from one row. - It is larger than Flipkart's biggest deal ($2.5B, Aug 2017), and larger than the $1.4B Flipkart and Paytm rounds (step 0). A ride-hailing startup at Series B raising more than Flipkart is implausible. - This looks like a unit or decimal entry error. I cannot confirm that from the data alone, so it is an inference. The row nonetheless sits at the top of every ranking. **3. The mega-deals are dominated by repeat names and a loosely defined round type.** - Flipkart (also entered as "Flipkart.com") and Paytm appear several times in the top 10. Most of the biggest rounds are labelled "Private Equity", and one Paytm round is just "Funding Round". - These names recur under different spellings, and the round labels are inconsistent, so counts by startup or round type are unreliable without cleaning. **4. Another record has an unusable amount.** One row (Drums Food, 21/07/2016) has a non-finite amount value, and I excluded it from the totals. **Takeaway:** Funding totals and averages in this dataset mostly reflect a handful of huge, partly dubious records. Median ticket size, deal counts, and trimmed or capped sums will describe the market better than raw sums. Removing the Rapido row alone would cut the total by about a tenth. All figures come from the aggregate queries in steps 0 and 12 (`raw.startup_funding`, amounts cast from text).