How it’s made:

Each transaction from the last 3 months is gathered from the relevant xyz.core.transactions table and the transaction fee (in the chains native asset) in all cases is converted to floating point notation (compensating for decimals). In all cases we calculate the USD equivalent value of the fees at the the time (hour) of the transaction.

The first tricky case is osmosis as this chain allows fees to be paid in assets other than osmo. For osmo we use a REGEX to isolate amount and the currencies on chain address. We can than use the dim.labels and dim.prices tables find the currencies relevant symbols and USD equivalent prices at the time of sale.

The second tricky case is flow, as there is no pre-processed fee data in transactions table. However, we can find the flow fee value in the last event of every transaction result, in the first field among the values:

• transaction_result:events[ARRAY_SIZE(transaction_result:events)-1]:value:fields[0]

• ethereum and binance smart chain alone make up over 95.5% of all fees paid last month, with ethereum guzzling up over 80.6%, over 115M USD.
• The Layer 2 chains Polygon, Optimism and Arbitrum make up about 4%, with Polygon around 2% (3 M USD)
• Osmosis and Solana are in the 100k range (605k and 172k USD)
• Near, Algorand and Avalanche are in the 10k range (60k, 15k and 10k USD)
• Gnosis and Flow used 1.5k and 632 USD each.

• The daily tally of the fees shows increased fee expenditures and (activity) following the start of the FTX revelations on Ethereum, Polygon, Optimism, Arbitrum, Solana, Avalanche and Gnosis.
• Interestingly, Osmosis fees trippled after October 26 and stayed stable.
• Fees on BSC, Near, Algorand and Flow were stable (except a brief increase on allgorand on Oct 24-25)
• The major correlated spike in activity on all chains happen after CZs tweet announcing a potential FTX takeover and SBF deleting his ‘Assets are fine‘ tweet
• Find details of the story and the source of the event breakdown in CoinDesk original
• Data in the chart is missing for some chains after Nov.11 because the crosschain.core.fact_hourly_prices table is missing data in the period