How to use an app to scan food ingredients, and when it fails
An app to scan food ingredients does one of two things. It reads a barcode and fetches a product record held elsewhere, or it reads the ingredient text printed on the back of the packet. Yuka, the most used of them in the United States, reporting 28 million users there on its own press page (checked 10 August 2026), works the first way.
Which mechanism an app uses decides how it fails, and the two fail in opposite directions. This page explains both, sets five apps against each other on what they read, and marks where LabelPeek — our own tool, and the least finished of the five — is not the answer.
What an app can actually read off a packet
Two things on a packet can be read mechanically. The barcode is a pointer to a record stored somewhere else, usually on a server the app owns. The ingredient list is the statement itself, the one the manufacturer is responsible for, and it is printed on every packaged food sold in the US.
Barcode lookup
The camera reads a number and the app fetches whatever record matches it. Fast, needs no typing, and is only ever as good as the entry somebody else already made.
Reading the printed list
The text on the back is matched entry by entry against a published reference. Nothing needs to exist in a database beforehand, because the label is the input.
The camera is not the difference
Both mechanisms can point a lens at a packet. What separates them is what the picture becomes: a number to look up, or words to match against a list.
Only the first is available to a US shopper in a shop today. Turning printed text into words needs optical character recognition, which the LabelPeek checkers do not run — the camera is reserved for the app, and the checkers take pasted text and the camera belongs to software that has not shipped.
Why the two mechanisms fail in opposite ways
A barcode app fails silently. When the record is missing the scan returns nothing, and nothing looks the same whether the product is genuinely unremarkable or simply absent from the database. A text checker has no coverage gap at all, because it reads whatever you hand it, and it cannot tell you which brand you are holding.
4.6 million
Four and a half million is a large number and still a finite one. Regional brands, store labels and packaging reformulated last quarter are where the gaps cluster, and a US basket carries more of those than a European one. The gap is recoverable — someone adds the record — but not while you are standing in the aisle.
A second failure hides inside a successful scan. An app can recognise the product and still not recognise one of the entries in its ingredient list, and printing a clean verdict over that unrecognised entry turns a hole in the data into a statement about the food. Anyone whose real question is about the additive line rather than the product should start at the additive scanner page, where the same three-state logic is set out against a printed reference.
Five apps that scan food ingredients, and what each one reads
Five apps answer this question in the US market: Yuka, Open Food Facts, Bobby Approved, Fig and LabelPeek. Four of them read barcodes and one reads pasted text. The table records how each one takes in a label, what data stands behind the answer, how it is sold, and the case where it returns nothing useful.
| App | How it takes in a label | What stands behind the answer | Cost model | Returns nothing when |
|---|---|---|---|---|
| Yuka | Barcode, food and cosmetics | Own product database and published weighting | Free to scan, optional Premium | The barcode is not in the database |
| Open Food Facts | Barcode, food | Volunteer-contributed open database, ODbL 1.0 | Free, non-profit, no paid tier | Nobody has added the product yet |
| Bobby Approved | Barcode, groceries | One curator’s published ingredient standards | Free to download (US App Store, checked 11 August 2026) | The barcode is not in the database |
| Fig | Barcode, packaged groceries | Product data filtered by the restriction you set | Free tier plus paid upgrade | No dietary filter has been configured |
| LabelPeek | Ingredient text you paste | 74 markers printed on the page, out of a 78-entry reference | Web checkers free, app unreleased | You have no label text to paste |
Cost is given as a model rather than a figure on purpose: tiers move without notice, and a dollar amount copied into an article ages inside a quarter. A wider field, including scanners that did not fit these five rows, sits in the roundup of food scanner apps, and every head-to-head on this site is indexed on the app comparison hub.
Yuka
Yuka reads barcodes across food and cosmetics in one app and returns a score out of 100 with a colour band. Its press page reports 28 million users in the United States and 85 million worldwide, which makes it the default answer to this question in most American households. Scanning is free and neither tier carries advertising; Premium, at from $10 a year in the US, unlocks search without scanning, offline use, unlimited history and dietary filters. The score is Yuka’s own weighting rather than any authority’s, so it answers how a product ranks rather than what sits in it. If the ranking is the part you are questioning, four Yuka alternatives compared puts that rule beside three others.
Open Food Facts
Open Food Facts is a barcode scanner sitting on a collaborative database rather than a verdict engine. Its homepage counter showed 4.6 million products on 10 August 2026, all volunteer contributed, and the whole dataset can be downloaded under the Open Database License 1.0 — the only one of the five you can inspect in bulk before deciding to trust it. It is free, non-profit and has no paid tier, and a product it does not know is one you can add yourself from the aisle. Coverage is the trade-off, and US private-label items are where it thins out. Cosmetics live in a sibling project, Open Beauty Facts, rather than in the same app.
Bobby Approved
Bobby Approved is the FlavCity grocery scanner built by Bobby Parrish, and it needs disambiguating before anything else: an unrelated subscription-tracking app also called Bobby shares the name in search results. It scans a barcode and returns an approve or avoid answer against one person’s published ingredient standards, covering categories such as synthetic dyes, certain sweeteners, preservatives and refined oils. That is the design rather than a defect — one consistent standard, applied the same way every time, which is exactly what some shoppers want and too blunt for others. Refined oils are the category people most often arrive for, and scanning for seed oils covers it on its own terms.
Fig
Fig inverts the question. Instead of one universal verdict, you declare the restriction and the scan answers that and nothing else. It is built around dietary protocols — Low FODMAP, low histamine, alpha-gal, gluten-free and vegan sit among the 25 diet attributes Fig lists on its own products page (checked 11 August 2026) — and its web catalogue lists over 363,250 products searchable without a login, the figure Fig gives on its own site (checked 10 August 2026), which is unusual in this category. Competitor reviews report a free tier of roughly five scans a month with a paid upgrade beyond it; that is a reported figure rather than a published one, so check the current store listing. Fig is the sharpest of the five when the question is a named restriction, and the quietest when you have not set one.
LabelPeek
LabelPeek is the text-reading option, and it has no app in any store as of 10 August 2026 — the web checkers are what exists. Paste a grocery ingredient list into the ultra-processed food checker and every entry is matched against a hand-written reference of 78 entries, 74 of which are markers of industrial processing and 25 of which carry a documented regulatory restriction with the source document named. The output has three states rather than two: matched, recognised but not on the list, and not recognised at all. Nothing is uploaded, nothing is stored, no account is asked for.
When no app helps at all
No app helps when there is no label. Loose produce, bulk bins, the bakery shelf, the deli slicer and anything cooked to order carry neither a barcode nor a printed ingredient list, so both mechanisms stall in the same place — the lookup finds no record, and the text checker has no text to read.
Restaurants sit in the same gap. Large US chains post calorie figures on menu boards, and a calorie figure is a nutrition disclosure rather than an ingredient list. The move that works there is a question to the person serving you, which no phone improves on.
A scanner reads packaging. The parts of a shop with no packaging — the fruit, the fish counter, the bakery — are the parts it has nothing to say about, and they are not the least interesting parts of the shop.
Portion size is the other blind spot shared by every app on this page. The same barcode returns the same answer whether you eat a spoonful or finish the box, and no scan can see how much of the packet ended up on the plate.
How to find out which app works in your shop
Whether an app works is a question about your shops, not about its rating. One trip settles it. Take the ten things you buy every week rather than ten interesting things, and count how many come back empty — that number decides more than any review, because it is measured in the aisles you actually stand in.
Step 1
Scan your own basket
Ten regular items, not ten curiosities. Count how many return no record at all. A basket of national brands usually misses rarely; a basket of store labels is where the number gets interesting.
Step 2
Read the reason, not the colour
For three of the ten, look for the ingredient behind the verdict. An answer that names the entry and the rule behind it survives being questioned by somebody else in the kitchen.
Step 3
Try one packet with no useful barcode
An imported item, a deli tub, a regional brand. Paste its printed ingredient text into a text-based checker and compare how far each mechanism got on the same packet.
Where LabelPeek is the wrong choice
LabelPeek is the wrong choice in the aisle. No app has been released, there is no camera and no barcode lookup behind the web checkers, so shopping with it would mean typing an ingredient list into a phone one-handed. A released barcode scanner wins that job outright, and pretending otherwise would not survive one trip to the shop.
LabelPeek web checkers
Paste an ingredient list and every entry is matched against a reference printed in full on the same page, with three possible states instead of a yes or no. Runs in the browser, no account, nothing uploaded.
A released barcode scanner
Point the camera at the code and an answer arrives in about a second, drawn from a database covering millions of packaged products.
It is also the wrong tool for counting calories or macros, which it does not attempt, and for allergen decisions, where a purpose-built dietary filter such as Fig’s is designed for the job and a general additive reference is not. The narrow case LabelPeek fits is a label you can already read, checked against a list you can also read.
Nothing here is medical advice, and nothing here is a verdict on a named brand’s product. The page describes what five apps read and where each one stops; whether a particular food suits you is a question for your own circumstances, and for a professional when health is involved.