Calorie barcode scanner: point at the label, read the panel

What the barcode actually contains

A common misunderstanding worth clearing up first: the barcode holds no nutrition information whatsoever. Those stripes encode a number and nothing else — 13 digits for EAN-13, used across Europe and most of the world, or 12 for UPC-A in North America.

The number is an identifier. The first few digits indicate the numbering authority that issued it, the next block identifies the manufacturer, the block after that identifies the specific product, and the final digit is a checksum that lets a scanner detect a misread.

Everything you see after scanning comes from matching that identifier against a database. The packet itself contributes only the number. This explains the behaviour that confuses people most: a barcode that scans perfectly and still returns "not found" has been read correctly — that exact product simply is not in the database yet.

Why some products return nothing

Four reasons, in rough order of frequency:

Nobody has added it. Open Food Facts holds around three million products, which sounds exhaustive until you consider how many own-brand and regional lines exist. Small bakeries, local chains and newly launched products are routinely absent.

It is a shop-internal code. Barcodes starting 02 or 20–29 are reserved for in-store use, typically on items weighed at a deli counter. The number encodes a price, and it means something different in every branch, so there is nothing global to look it up against.

Regional formulation. The same brand and product can carry different barcodes and different recipes in different countries. A record may exist for the US version while you are holding the UK one.

The record exists but has no panel. Someone photographed the front of the pack and never filled in the nutrition table. The product is technically in the database with no numbers attached.

In every one of those cases the honest response is to say so, which is what this scanner does. An app that silently substitutes a similar-sounding product is doing something considerably worse than failing.

Making a scan work first time

Barcode decoding is old, well-understood technology and it fails for boring physical reasons:

• Light. The decoder needs contrast between bars and spaces. A dim kitchen is the most common cause of a scan that will not catch.
• Distance. Around 10–20 cm suits most phone cameras. Closer than that and autofocus gives up.
• The quiet zone. Barcodes need clear margin either side. Get the whole symbol plus a little white space in frame, not just the stripes.
• Curvature. Tins and bottles bend the symbol. Angle the camera so the barcode is as flat as it can be, or rotate the tin so the code faces you squarely.
• Shine. Glossy and foil packaging reflects light straight back. Tilt slightly off-perpendicular to kill the glare.

If it still will not read, type the digits in by hand. The lookup is identical and it takes ten seconds.

Reading what comes back

Per 100 g versus per serving

This is the single most important thing on the screen, and the most frequently misread. Food databases store everything per 100 g because that is the only figure that survives a portion change. The number you care about is your portion.

Our scanner defaults to the manufacturer's stated serving where there is one, shows you what that serving weighs, and recalculates live when you change it. If a scanner hands you a per-100 g figure with no portion control, it has left the actual work to you.

The macros

Protein, carbohydrate and fat, in grams. Cross-checking them against the calorie figure is a quick sanity test: protein and carbohydrate are about 4 calories per gram, fat about 9. If the macros imply a very different total from the stated one, the record is probably wrong — and worth reporting.

Nutri-Score and NOVA

Where the database has them, we show both. Nutri-Score is an A-to-E summary weighing energy, saturated fat, sugar and salt against protein, fibre and fruit-and-vegetable content. NOVA is a 1-to-4 processing classification. Neither is a verdict — Nutri-Score compares within a category, so a cheese scoring D is not "worse than" a diet cola scoring C in any way that matters to a person planning dinner.

Accuracy

Barcode scanning is the most accurate route available to a consumer, with two caveats that are worth stating plainly.

The first is that labels themselves are legally permitted to vary. Both US and EU regulations tolerate roughly 20% deviation between declared and actual values for most nutrients. The scanner reproduces the label exactly; the label is an approximation of the food.

The second is reformulation. Manufacturers change recipes without changing barcodes, so a database record can describe last year's formulation. When a scanned figure looks noticeably different from the packet in your hand, believe the packet.

Neither of these is a reason to avoid scanning. They are reasons to treat 187 calories as "about 190" rather than as a measurement — which is the right way to read any calorie figure.

Your first scan is free — open the scanner and point it at whatever is nearest. Unlimited scanning is $19.99 a month or $59.99 a year.