🛠️ nutrition something something

I've been poking around on a project related to groceries, nutrition and prices.

☙

"Food" is a mature industry. The OG industry.

We can support a global population of billions, and it takes less and less of our time and resources to do it*. All while offering an ever increasing amount of choice.

...and yet, malnutrition is still a problem: from over-consumption, mis-consumption, and under-consumption; affecting the power and the wealthy. We've been at this for thousands of years!

Most people don't know what "eating healthy" is (ie. nutritional literacy is low), and the scientists still disagree what "eating really healthy is" (ie. there are still a lot of important open questions).

Even ignoring financial costs, "eating healthy" takes a lot of time and effort (and perhaps more than it used to in "simpler times").

Part of the issue is that consumers are fighting a battle they don't realize they're participating in. Food is a mature industry, with it's own interests, and it's had a long time to optimize how it interacts with consumers.

Contemporary consumers are offered a mind-boggling amount of choice. Walking into a modern supermarket presents a continuous torrent of information. I dare say that consumers are purposefully dazzled into a stupor. Each product viciously competing for your attention and your wallet.

...and with all this information, I find myself standing in the dairy section, asking: "What should I be putting in my coffee? Is this almond-milk-half-and-half "healthier" than this full-fat-milk-based-creamer? Which is more nutritious? Which one is actually more expensive?" (Repeat every few metres).

Not to mention the bigger question: "what's a meal plan for an entire day of meals that is reasonably healthy and cheap?" Never mind, these hot-pockets sure look good.

"Eating healthy" takes a lot of effort and time: learning about nutrition, planning meals, comparing prices (which feeds back into planning meals), checking what you already have, doing the actual shopping, transporting, putting groceries away, cooking, cleaning. ...oh, and yes, the "eating" step.

...or, if you don't make the effort, you can just: hope someone else does it for you, pay a premium (say, to not price compare), or end up eating "unhealthily" for years (overeat, under-eat, miss out on nutrients).

☙

With all that said, the "thematic goal" of this "project" is: help people eat healthier for cheaper.

And right now, I'm exploring this problem in 3 parts:

• "wikigrocery" - a price comparison tool for groceries (camelcamelcamel for food)
• "bougie scoring" - a value to quickly compare products on price and nutrition ("$/g" is useless)
• "healthy frugal optimizer" - a program that calculates cheapest way to meet all nutritional requirements in a given area (the results will surprise you!)

A Price Comparison Tool for Groceries

If I need to buy a computer part (say, some RAM) PCPartPicker shows me alternatives, prices at different stores, and a history of those prices:

I want that for groceries.

Standing in the grocery store, I want to know: "is this tofu pack worth it?"

• ...no, the other tofu pack is better value
• ...no, it's on sale frequently, wait a few days if you can
• ...no, it's 15% cheaper across the street

Part of the catch is that individual groceries are relatively cheap (unlike RAM), but bought regularly. Is it really worth the effort to shave $1 off your bag of milk? This time, probably not. But regularly? Yes (...if it was convenient enough to do so). How much effort time/effort would you be willing to spend to save 20% off your yearly grocery expenses? For many people "comparison shopping" is already worth it and they do it manually (they are money poor and/or time rich), but for many, it would only be relevant it was super convenient or offer something more.

The other catch is, that, until recently, the data was very hard to get.

Some thoughts on how to get the data....

• internet scraping
  • COVID helped push pretty much all major supermarkets online
• crowdsourcing
  • in-store capturing
    • have volunteers manually enter prices, or make it easy to take a photo (new digital price tags make this much easier)
    • (for a while, I was doing this myself, and it's pretty tractable; at least for the range of products that one buys for themself)
  • receipt scanning
    • Checkout51 and may other apps used to pay you to send in your receipts; now with LLMs, parsing them is relatively trivial (you just have to map each store's labels to actual products)
• flyer scraping
  • Flipp has prices on the web, and if its on the web, it's possible to scrape (but this is just "advertised deals"... I want everything)
• direct partnerships
  • every store has their prices in some software system (they now use them to update digital price tags); would they ever be incentivized to publish their prices on a price comparison site?

A bunch of hobbyists and some startups have started to do this:

• https://grocerytracker.ca
• https://www.sakisv.net/2024/08/tracking-supermarket-prices-playwright/
• https://www.eezly.com/

To do this at proper scale isn't trivial. There are 7,000 to 15,000 grocery stores in Canada (depending on what you count as a store), so about 3 to 4 million globally. With billions of products. And frequently changing prices.

On my end, I've been...

• experimenting with manual entry (spreadsheets, airtable)
• experimenting with receipt scanning (promising!)
• experimenting with AI aided photo capture from stores (promising!)
• (above resulted in coming up with a decent data model to capture all the information)
• experimenting with first steps on how to display the information
• planning to experiment with scraping from the stores I care about

The future of grocery shopping is likely to shift more and more towards online ordering and delivery (we saw this happen during COVID and it's remained popular in hyper-urban places like Hong Kong). If that comes to be, supermarkets will lose much of their "closest store to you" monopoly, and will need to compete on price more or find other ways to benefit consumers.

And then there's AI agents. If the future involves buying groceries at all, then you will likely delegate your food decisions to an agent - finally, an entity that will have the knowledge of nutrition, all prices, and the "patience" to optimize every penny for you. Such an agent would need access to the price information. So maybe, let's create the foundations for that digital marketplace?

(More on this soon...)