When a barcode comes up empty
A scan miss used to be a dead end. Now it's the start of a lookup chain: identify the product, research the nutrition, cite the source — and the next scan just works.
Every tracker has the same worst moment: you scan a barcode and get nothing. Empty state. “Not found.” You’re standing in your kitchen holding a protein bar, and the app that promised to make this easy just shrugged.
Most apps handle this by making it your problem — “add a custom food!” — which means typing eleven numbers off a nutrition label into your phone while your dinner gets cold.
We think a miss is our problem. Here’s what happens now when a scan comes up empty in Macroline.
The chain
Step one: check our own shelves. Your scan first hits our catalog — 400,000+ foods, most with barcodes, stored under every padding variant scanners actually emit (12-digit UPC, 13-digit EAN, the works). Most scans end here.
Step two: ask OpenFoodFacts. A community database with global coverage. If the product’s there and the data passes our ingest checks — physically plausible macros, sane serving sizes — we cache it and hand it back, labeled as what it is: community data, estimated tier.
Step three: figure out what the thing IS. This is the new part. On a true miss, we ask a barcode registry to at least identify the product — brand and name. Knowing “that’s a Joyride Lemonade Stand Smacks Candy, 3.5 oz” turns an unanswerable question into a researchable one.
Step four: research it properly. With a real product name in hand, our research pipeline goes to work on the nutrition — and whatever it finds gets written into the catalog with the barcode attached. That last detail matters most: the next person who scans that UPC gets an instant hit. Every miss makes the catalog better.
What we won’t do
If the registry has never heard of the code, we stop. We could ask an AI to guess nutrition from a bare number — but a guess dressed up as data is exactly the thing we built Macroline to end. A clean “not found” plus a manual-entry path beats a confident fabrication every time.
And everything the chain produces is tiered and cited like all our data. Research output isn’t stamped “authoritative” — it’s labeled as researched, sources shown, upgradeable when an official source confirms it.
A dead end became a flywheel. Keep scanning weird stuff — you’re building the database.