Reformulations, quiet recipe revisions and data corrections, caught by re-photographing
labels and comparing them against supplier data. Updated monthly with the shelf.
August 2026
A carb figure so low we refused to publish it, until the pack proved it
A lower-carb loaf arrived in supplier data claiming 1.6 g of carbohydrate per 100 g, a number one tenth of ordinary bread and low enough to look like a data error. Under our rules a single-source value that implausible is blocked from publication. Then the pack photograph came back: the printed panel says 1.6 g. The block lifted, the value now shows as verified, and the loaf tops the lowest-carb list on merit. See the product page →
August 2026
53 single-source values failed the plausibility screen
Feeds and public databases carry typos: a burger thin listed at 0.39 mg of sodium (the real figure is about a thousand times higher), loaves with impossible energy totals, wraps with vanishing fat. Every single-source value outside the plausible range for bread is now blocked from publication until a second source or the pack label settles it. 53 values are currently held back this way.
August 2026
33 discounter bakery lines join, and 20 of them have no barcode anywhere
The discount-supermarket bakery range now has full entries, read from high-resolution panel images: from a 210-kJ-per-slice lower-carb loaf with a 5-star rating to sourdoughs carrying the highest sodium in the category. A wrinkle worth recording: 20 of these products publish no barcode on any database we can find, so their pages carry a “no GTIN published” marker instead of pretending. See the product page →
August 2026
Two loaves, one panel, and this time it’s real
A brand’s white sandwich and white toast loaves print byte-identical nutrition panels. In peanut butter the same pattern turned out to be a feed bug; here the packs themselves agree: same recipe, different slice thickness. Worth knowing if you buy by the numbers: the thick-cut loaf differs only in serve size. See the product page →
August 2026
105 label-vs-feed disagreements under review
Across the category, 105 individual values currently disagree between the pack label and supplier data. Each product affected carries an “under review” marker on the exact value in question, and nothing is published for that value until it resolves. That’s the deal: no number without a status.