Methodology

Where the data comes from

Katofalk is built entirely on the 2026-04-30 USDA FoodData Central Branded Foods — a public dataset the USDA compiles from ingredient and nutrition labels that manufacturers submit for products sold in the United States. We do not run our own lab testing, and we do not scrape retailer sites. Every flag on this site comes from matching text patterns (like "Red 40" or "FD&C Red No. 40") against the ingredient list a manufacturer filed.

The current release covers 464,005 US products from 22,841 manufacturers, deduplicated to the latest filed version of each UPC barcode.

What "label last updated" means

Manufacturers do not re-file a label every time they change a recipe. Some products in this dataset were last updated years ago; others were updated this month. We show the filing date on every product page, and flag anything filed before 2024 as potentially stale. A product marked as containing a dye may already have been reformulated on the shelf — always check the current package if it matters for you.

How the reformulation tracker works

We compare the October 2021 and April 2026 USDA releases side by side, matched by UPC barcode. When a product's ingredient text listed one of the FDA's six phase-out dyes in 2021 but no longer does in 2026, it appears on the tracker with the date of the newer filing. This catches real, filed reformulations — not press-release pledges, and not every reformulation, since it depends on a manufacturer actually re-filing the label.

What "dye-free" and a clean badge mean

A clean or dye-free result means we did not find any of the specific additives we track in the listed ingredients. It is not a certification, and it does not mean the product has no additives at all — only that it has none of the ones this site checks for.

Limits

• Pattern matching on ingredient text can miss unusual spellings or catch false positives in rare cases.
• Some manufacturers file incomplete or generic ingredient text.
• This site is not a medical, legal, or safety authority. It reports what is on file, not whether an ingredient is harmful.
• Regulatory status text is written to be accurate as of the current USDA release date, not live-updated daily.

Data source: USDA FoodData Central Download Datasets.