MarcoLearn
How accurate is USDA FoodData Central?
USDA FoodData Central is accurate as a laboratory and label-based reference for many foods — it is the standard dataset consumer trackers rely on. Your personal accuracy still depends on choosing the right food and honest portions. Marco uses FoodData Central (plus published brand/restaurant facts) and asks when your description is ambiguous.
What “accurate” means here
FoodData Central entries come from analytical data, calculated values, or manufacturer labels. That is far more trustworthy than a generative model inventing macros — but two different yogurts can still differ, so the match matters.
Where errors creep in for humans
Wrong item selected, vague “oil,” restaurant customizations, and eyeballed cups. The database can be right while the log is wrong if the description is sloppy.
- Prefer brand + product when packaged
- Include weighed grams when you have them
- Answer clarifying questions when Marco asks
Marco’s role
You should not have to learn FoodData Central IDs. Marco maps language → candidate foods → a logged entry, and refuses the “confident wrong guess” pattern common in general AI chats.
Bottom line
FoodData Central is a solid backbone. Pair it with clear texts and you get usable day-to-day tracking — not lab-certified metabolic measurement, not chatbot fiction either.
Frequently asked questions
Is USDA FoodData Central trustworthy?
Yes — it is the primary public U.S. nutrient reference used across research and consumer tools. Match quality and portions still determine your personal log accuracy.
Does Marco only use USDA data?
Marco uses FoodData Central and published brand or restaurant nutrition facts when those are the better match for what you ate.
Why did my burrito not match exactly?
Custom orders vary. Name the chain and modifications; Marco will clarify when needed instead of inventing a perfect number.
Ready to track by text?
Join the waitlist for early access. No app to download — once you’re in, you text Marco from the messaging app you already use.