Food waste is a huge issue. Combined, Americans and Canadians throw out almost $200 billion of food each year, equal to hundreds upon hundreds per person.

We do this for many reasons. Food is plentiful and relatively cheap, and yet at the same time our need for value means we buy in bulk for a discount even if we actually lose money by throwing out much of it. The media shape our standards to only want “perfect”-looking food, meaning lots of perfectly-fine food goes unsold at the grocery store. Inconsistent and often-flawed guidelines confuse when food is still edible, with “best by” dates benefiting manufacturers more than customers.

I wondered if technology could put a dent in food waste, potentially saving households money. What if, for example, a scanner in a fridge could tell you what you had and when it might potentially go bad? On top of that, it could take the pain out of using odds-and-ends by suggesting recipes to use up aging ingredients. It could be a version of Tetris, where every week people would metaphorically try to combine ingredients together to form a meal, akin to completing a line that’d vanish.

Turns out I wasn’t the first to think about this concept. Several apps out there will do this, even integrating your grocery receipt to help make creating an inventory easier. I was feeling glad they existed until I asked myself why none of them seemed to have taken off. If an app could easily save you money and lessen environmental impact, why wouldn’t you?

The answer appears to be that the system isn’t that easy at all (or easy enough). How do you ensure that food count is accurate? What if food ages differently than the app expects? Are the recipes the kind you wanna make? And if it’s less work to just throw out the food rather than fiddle with an app (because let’s face it you’ve already spent the money so it’s a sunk cost) why open yourself up to the anxiety of constantly trying to efficiently use everything?

That’s when it dawned on me that centering the problem around the fridge was a bad assumption. I had to go back in the process to a different location: the grocery store. That’s where we buy the food and anyone who does the household shopping knows how frustrating the experience can be. Let’s go through a typical scenario.

Imagine I’m a parent buying for my spouse and two kids. I’m time-strapped, so I’ll likely only make one or two trips a week. I might plan out some meals ahead of time, but usually I’ll throw in whatever’s on special and buy enough of everything to have flexibility around my meal plan. I don’t want to risk buying too little as that would force another trip to the store, and besides many items become cheaper when I purchase multiples. I checkout with approximately a week’s worth of food—I’ve been doing this a while so it’s not like I’m completely in the dark—and toss in a few splurge items since I’m proud that I’ve finished on-time and slightly under-budget.

At home, lots of reasons might throw my grocery purchase awry. I discover vegetables from last week that I forgot to cook and need to urgently use. The sour cream I’d previously bought has started growing green stuff on it, which means I’ll have to rejig one meal idea. I didn’t buy enough of another ingredient so that means even more change, but I also have a family-size package of something else that we can’t get through without eating it every day this week. Then the kids want pizza and I’m tired enough to give in to not having to cook.

Obviously, I’m piling it on a bit but it explains how our fridges end up a mish-mash of everything. What needed to change was the grocery store experience. Instead of buying goods in lump sums, it’d be better to buy them as they’d be used. I’m thinking of how Toyota famously invented just-in-time production, which reduced waste by receiving goods only as needed. Buying just what you need isn’t a new thing (weekly supermarket trips broke the tradition of going to the market daily) but it got lost in the advent of the big box store. Now, the equation is changing again: why even step into a store when Amazon has taught us how to get everything delivered?

A few solutions have popped up, figuring out pieces of the puzzle. The most obvious might be something like Blue Apron, where you can select from meals and have it appear on your doorstep up to four times a week with properly-portioned ingredients. But at nearly $150 ($200 Canadian) a week for half a week’s meals, many families can’t afford its services. Some grocery stores offer next-day home-delivery but steep minimums and expensive delivery costs keep their services from getting widely adopted.

With Amazon’s purchase of Whole Foods, a true alternative is shaping up. Amazon now has prime real estate footprint across the continent and the know-how to deliver items approaching real-time. With groceries arriving as easy as Uber Eats, all it takes is an app that provides meal suggestions to guide customers to ordering just enough each day. (And while you’re at it why not catch a few cooking shows on Prime Video to boot?)

Scott Galloway of L2 has hypothesized how Amazon will swallow up retail and shared why the company is at war with brands. Brands, Galloway notes, have been able to charge a premium for slightly above-average quality because branding signals trustworthiness to consumers and saves them the work of sorting through the market. Amazon’s algorithms disrupt this by ranking brands that either are newly popular or are high-quality at good value—and neither describes traditional brands.

Traditional supermarkets may make deals to highlight products at end-caps and shelf levels, but Amazon tosses out that and any other factors that raise costs. Amazon’s much happier to serve its customers private labels if that means decent quality at affordable prices. If you believe the gospel of Amazon, that should lead to better quality products at lower prices delivered conveniently for customers.

What I like about this thought experiment is taking it even further. Once you move into real-time grocery shopping, you can get rid of even more outdated ideas. Why should I have to buy anything prepackaged in an amount I may not need? For example, take the sour cream example from earlier: maybe I just want 50 grams of it for pierogis delivered in a reusable container and not a whole tub. That guarantees me fresh sour cream at a convenient time with little chance of spoilage. This moves us to a just-in-time system like the one Toyota introduced decades ago, something that wasn’t available until you had a company at the scale and with the impatience of Amazon.

One could argue that this automation would continue disintegrating the amount of time people spend in social environments. I’d argue the opposite. No one wants to wait in a grocery checkout line or fiddle with a broken cart wheel or wait to leave the parking lot—and the interactions at most big chains is passable at best. The time savings could be funneled toward visiting local farmers’ markets where the social interactions return more value. If I’m a fan of fine cheeses, I could spend more time visiting cheese shops now, deepening relationships with the owners and staff.

As I close this off, I want to highlight that, obviously, this doesn’t solve the whole equation. This doesn’t factor in restaurant waste at all, and it doesn’t guarantee that people still won’t throw out food. (My friend Karon notes that a lot of food waste comes from a lack of education around using the odds-and-ends to make stock or soup, for instance.)

Also, there’s the question of what happens to the many retail jobs that would be affected by this change. These are problems I will continue to think about, but this is a case where change has such a positive effect for everyone involved that it’ll be worth investigating solutions for these questions.