Logging takeaway food, without the guesswork.
Takeaway is where calorie tracking usually breaks. There is no barcode, no nutrition label, and the portion that arrives rarely matches the standard recipe a menu number describes. Most people respond in one of two ways: skip logging the meal entirely, or copy a database entry that may describe a different restaurant’s version. Both quietly bend the week’s numbers.
The three-component method
Any takeaway box breaks into components you can estimate one at a time:
- The protein — a chicken breast, a portion of beef, two falafel. Size it against your palm; most single-protein portions land between one and two palms.
- The carb — rice, noodles, fries, bread. This is where portions run largest; delivery rice portions are commonly double what you’d cook at home.
- The sauce and fat — the invisible calories. Creamy, glossy or fried means oil and butter you can’t see. When in doubt, round up here and nowhere else.
Log those three honestly and you’re within a sensible range — the accuracy that matters for a weekly average, without pretending precision that no one has.
Why a photo beats the menu number
A menu calorie count describes the recipe as specified. Your box contains what one cook made on one evening — the gap between the two is mostly portion variance. A photo of the actual meal captures the portions you actually received: scan the open box with Cal AI: Calorie Scanner and each component comes back listed with its own estimate, ready to correct in one tap if the rice is clearly a mountain.
The habits that make it stick
- Log before the first bite. After a meal, portions shrink in memory — recall research shows people consistently underestimate what they ate.
- Don’t skip the “bad” boxes. A logged kebab is data; an unlogged kebab is a mystery that erodes the week. The log is a measuring tool, not a judge.
- Keep a house-favorites shortcut. Your usual Friday order barely changes — log it well once, then reuse it.
Three free AI scans — calories, protein, carbs and fat from one photo.