If you want to accurately measure the cost of sending a package (or passenger, or data packet), for any reasonable purpose, you can't use the "just one more" test. You have to say, "what if I increased the load by a million units? Okay, the unit cost should be regarded as a millionth of that."
Here's a great application of the concept when Netflix tried to use the "just one more" model: [1]
>The green marketing gurus at Netflix go even further, arguing that the mail is going to be delivered to your house anyway, so the environmental cost of delivering one of their DVDs is effectively zero. ... Here, the Green Lantern feels Netflix may be overplaying its hand just a little: Eventually, the addition of new mail into the system adds up, requiring more trucks, greater strain on the mail-sorting system, and so on. Since we can’t identify the impact of one extra piece of mail, we’re better off averaging the cost of delivering the mail over each item.
[1] https://slate.com/technology/2008/08/is-your-netflix-queue-d...
It was discussed on HN too: https://news.ycombinator.com/item?id=414173
At any rate, your food gets shipped from same place and ends up with you. Isn't the co2 going to be basically the same if that's to a grocery store first or to a big cooking joint? Assuming the locations are roughly similar.
"Of course, shipping meal kits to millions of households causes emissions. But these kits are delivered alongside other mail on normal routes, and the researchers found that this last stage of distribution accounted for 11% of grocery store meal emissions but only 4% for meal kits."