Nobody agrees on how much food America wastes

You have seen the statistic. It's on a poster in a grocery store, or in the first line of an article, or in the pitch deck of a company selling you something. A third of food is wasted. Forty percent of food is wasted. A hundred and thirty-three billion pounds. Two hundred and eighteen pounds per person.

Some of those are real numbers from real federal agencies. They disagree with each other, and the disagreement is not sloppiness. It's the predictable result of asking different questions, and it's worth understanding if you're ever going to evaluate a claim in this category, including ours.

USDA states the situation plainly on its own food waste FAQ page: the United States does not currently have a single baseline estimate of food loss and waste. Instead, two very different measures describe it.

The two federal numbers

In September 2015, USDA and EPA jointly announced the first domestic goal to cut food loss and waste in half by 2030. To measure progress against it, they needed a baseline, and they ended up with two.

EPA's baseline counts disposal. In 2010, EPA estimated 218.9 pounds of food waste per person was sent for disposal, and the 2030 goal aims to halve that, down to 109.4 pounds per person. That's a waste-management measurement. It asks how much food-derived material is arriving at landfills and combustion facilities.

USDA's baseline counts supply. Its Economic Research Service estimated that in the same 2010 baseline year, food that went uneaten at the retail and consumer levels represented 31 percent of the available food supply, roughly 133 billion pounds, worth an estimated $161.6 billion at the time. That's a food-economics measurement. It asks how much of what was available to eat didn't get eaten.

Those are not competing attempts at the same figure. They are answers to two genuinely different questions, and both are useful for the purpose each agency has.

Why the agencies count differently

EPA's mandate explains its metric. EPA notes that more food was sent to landfills than any other single material in everyday trash, and that municipal solid waste landfills are the third-largest source of human-related methane emissions in the United States, at approximately 14.4 percent of them in 2022. If your job involves the waste stream and emissions, the thing you need to know is tonnage arriving at facilities. Food composted at home or fed to animals doesn't show up in that number, because it never entered the stream you're measuring.

USDA's mandate explains its metric too. ERS tracks the food supply, so the meaningful question is what share of available food failed to be consumed, regardless of where it physically ended up. A tomato composted in a backyard is uneaten food by USDA's framing and invisible in EPA's.

Neither definition is wrong. But you can see how quoting one number without saying which question it answers produces a mess, and how easy it is to pick whichever figure sounds most alarming.

It got more complicated in 2021, when EPA updated its baseline to align the food waste portion of the goal with the United Nations' Sustainable Development Goal Target 12.3, which addresses halving per capita food waste at retail and consumer levels while also reducing losses along production and supply chains. Definitional realignment is normal and necessary in this kind of measurement, and it also means figures from different years may not be measuring quite the same thing.

Where it actually happens

The sector split is the part most relevant to anyone standing in their own kitchen. A Congressional Research Service summary of the estimated tonnage generated across the retail, food service, and residential sectors puts roughly 40 percent from households, 40 percent from food service providers, and 20 percent from food retail.

So households are, along with restaurants and cafeterias, one of the two largest contributors in that accounting. Not a rounding error, and not the whole story either.

For global context, CRS notes the United Nations estimate that about a third of all food produced for human consumption is lost or wasted, split into roughly 14 percent lost from harvest through the manufacturing and processing stages, excluding retail, and another 17 percent wasted at the retail and consumer levels. Estimates vary substantially by country, income level, and food type.

The unit problem, which is the one we care about most

Every figure above is denominated in weight or in share of supply. The dollar figure in the USDA baseline is an estimate attached to a poundage.

This matters because weight and money rank food waste completely differently. A pound of ground beef and a pound of iceberg lettuce are the same pound and very different amounts of money. Any measure denominated in tons will be dominated by heavy, watery, cheap items, which means a national tonnage figure tells you very little about where the dollars in your household are going.

That's the reason Savery's savings figure is weighted by dollars rather than pounds, and it's a choice we'd defend in either direction of scrutiny. It also means our number is not comparable to the national figures, and shouldn't be read as a slice of them.

How to read any statistic in this category

Four questions, and they'll get you most of the way with any number anyone quotes at you, ours included.

What is being counted? Food arriving at disposal facilities, or food that went uneaten regardless of destination? Those diverge by a lot.

At what stage? Farm, processing, retail, food service, household? The 2030 goal covers several, and headline numbers often silently blend them.

In what unit? Pounds, share of supply, or dollars? A figure in pounds cannot be converted to a figure in dollars without a pricing assumption, and that assumption is usually not shown.

From what baseline year? The official U.S. baselines are anchored to 2010. That's the correct year for measuring progress toward a 2030 goal, and it also means the headline figures you see are describing a food supply from a decade and a half ago.

Why we're telling you this

It would be commercially convenient for us to pick the scariest available number, print it in a large font, and imply that Savery recovers a proportional share of it. Plenty of products in this category do exactly that.

The problem is that our entire pitch depends on a number being believable. If we're cavalier with the national statistics, there's no reason for you to trust the arithmetic we do on your receipt. So here is one we'll say plainly: the USDA loss data underlying our own calculation is described by ERS itself as preliminary, with its loss assumptions offered as initial estimates for further research and subject to ongoing revision. We'd rather say that out loud than have you discover it.

The honest summary is this. Food waste in the United States is large by every available measure, the measures disagree because they were built for different purposes, households are a substantial share of it, and nobody, including us, can tell you what your specific household wastes without looking at your specific household.

That last part is the only number that was ever going to be actionable anyway.