A useful estimate is one you can inspect. Start with what you know about the meal, then adjust the parts the app cannot see. That is the whole idea — and it is worth saying plainly that an estimate is a starting point, not a measurement.
What an estimate can and cannot infer
A photo carries shape, colour and context. From that, an estimate can propose what the food probably is and roughly how much of it there is. What it cannot do is weigh your plate, see the oil in the pan, read the sauce, or know what the kitchen left out.
- Can usually infer: the identity of visible items, an approximate portion, and a plausible nutrition profile for those items.
- Cannot infer: exact weight, cooking fat, dressing and sauces, sugar in a drink, hidden ingredients, and anything off-camera.
- Cannot confirm: that a barcode product is the exact variant in your hand, or that a restaurant dish matches a generic entry.
Three details worth checking first
Three useful places to start are portion, ingredients and preparation.
- PortionAsk: is this the amount on my plate, or a default serving? Portion error scales everything else, so fix it first. A measured portion or package serving information can help you check the quantity.
- IngredientsAsk: did the app include the thing I know was there, and did it add the thing I know was not? Oil, butter, dressing, cheese and sugar are the usual discrepancies. Add what is missing; remove what is not there.
- PreparationAsk: was it grilled, fried, baked, steamed or boiled? Did it come with a sauce? Preparation and added ingredients can change the nutrition values.
A worked review of one mixed meal
Take a plate of chicken, rice and vegetables with a sauce on the side. Here is what a review actually looks like — not a demonstration of a perfect result, but the ordinary version.
Reviewing a mixed chicken-and-rice plate
- What I know
- The sauce came in a separate bowl, so only a spoonful went on the plate.
- The rice was weighed dry before cooking, so the amount is known.
- The chicken was grilled, not fried.
- What I am estimating
- Whether the estimate applied the sauce to the whole plate or left it separate.
- Whether the chicken portion assumed a standard serving weight.
- What that means for the entry
- The fix is small: move the sauce to its own line, mark the cooking method, and adjust the chicken portion to what is actually on the plate. The total changes; the entry becomes something you would trust six months later because you know where it came from.
How to correct the entry
Makro shows you the items it matched, the portion it assumed, and the macros for each one, before anything is saved. From that review screen you can change an item’s name, change the quantity, and change the nutrition values directly.
- Open the meal and look at each line rather than the total.
- Fix the item that is wrong, not the total. The total recalculates.
- Change the quantity or the nutrition values to match what you actually know.
- Save it. The corrected entry is what lands in your day, and the correction is what makes the pattern worth reading.
When a label or a measured portion is better
Use a more specific source when you have one, while checking its units and assumptions.
- Packaged food: compare the label, serving size and amount eaten. Labels are a useful reference, not a measurement of your individual portion.
- Anything you weighed: use the number. Make sure the unit and raw or cooked state match the entry.
- A recipe you made yourself: the ingredients list is better than a photo of the result.
- A restaurant dish: the restaurant’s own published nutrition information, checked for the dish you ordered, beats a generic match.
How uncertainty carries into the daily view
Corrections made at review time change the daily view, because the daily view is built from the entries you saved — not from the first guess. Keep incomplete logs and uncertain portions in mind when reading the totals.
Where an entry stays uncertain, the honest thing is to leave it uncertain. A total that mixes two confident numbers and one guess looks more certain than it is, and that is how a log starts to feel like it is judging you.
Where to report a confusing estimate
If an estimate was obviously wrong, tell us. Include what you ate, what the app suggested, and what you corrected it to. That is far more useful than a rating, because it describes a case someone else can reproduce.
Common questions
More about this topic.
How accurate is Makro’s food recognition?
We do not publish an accuracy figure, and we would rather explain why. An accuracy number needs a published method, a defined sample and a clear evaluation process. What we can tell you plainly is what the estimate can and cannot see: portion, hidden oil, sauces and off-camera ingredients are the weak points, and those are exactly the parts the review step is for.
Does editing an estimate change my daily total?
Yes. The day is calculated from the entries you saved after review, so a correction changes the total it produces. Editing is the intended workflow, not a workaround for a broken product.
Can I trust a barcode scan?
A scan returns a match when the product is in the food database, and no match when it is not. Check the variant and the serving size on the result — that is where the errors tend to be.
Is a restaurant meal worth logging if I am not sure of the ingredients?
Yes, with the unknowns marked. A meal you logged and corrected is more useful than a meal you skipped because you did not want to guess. Our guide on restaurant meals covers how to do it.
About this article. General education from the Makro team. Product information checked November 4, 2026. This article is not medical advice or an individual nutrition plan. Product guidance from the Makro team. Food values remain estimates to review.
