There's a growing trend to only report C2-C4 (ethane, propane, butane) as NGLs. C5+ (pentanes plus, natural gasoline, naphtha, whatever you want to call it) often goes in company financials and reported as crude oil. It is for the most part (usually fetches a price 85%+ of WTI), so this isn't some sneaky accounting trick.
What's making this very confusing is that the majors and some large independents own midstream and downstream infrastructure, so double counting of supply is very real.
1. Oil company (or the upstream business unit of an integrated major) reports c5+ in their crude oil production. Cool. No rules against that, and is consistent with what the liquid actually is.
2. Midstream company (or midstream business unit of an integrated major) separates the heaviest of the c5+ at the processing plant and blends it with crude oil. This gets sent down the crude oil pipeline instead of the Y-grade pipeline. Cool, this is consistent with what the upstream company is doing. But EIA's accounting of hydrocarbons often counts it again (especially if the midstream company has an incentive to do so, depends on the GPT contract), especially on these weekly EIA updates. So now we have upstream and midstream counting the same c5+ for supply purposes.
3. Y-grade (mostly c2-c4, but still enough c5+ to throw balances out of whack) gets sent to the gulf coast in dedicated pipelines, individual components go through fractionation. C5+ is then separated and blended into crude oil during downstream processing. This also gets counted as crude supply, depending on how it's separated. This is never balances, volatile hydrocarbons are unpredictable and change drastically with small changes in temperature.
4. Crude oil transportation infrastructure. The lightest ends of stabilized crude oil (this is also c5+) can still flash to vapor, then condense again with small changes in temperature. This is often removed, especially before export. More "crude oil" for the purposes of EIA's supply. Here's more confusion: even heavy crude oil contains a lot of c2-c4 that gets thrown in with propane, etc for supply purposes.
It's also worth discussing the differences between the weekly EIA estimates, which are just estimates. Best to take a 4 or 6 week rolling overage, which ends up closer to the balanced supply and disposition a few months later.
I'm still confused with all of this, and so is everyone else. The only way to do this more precisely is track the each specific hydrocarbon chain (c2-c30+) and adjust for cracking, and that's not possible. See that big jump in 2023? That's partly a result of the changes DeCarolis was talking about. I have no idea how much of that 1 million bbl/d jump is a result of double-counting, but I do know there's some.