Calorie counting apps: how they work and where they go wrong

A calorie counter looks like a calculator, and people expect calculator-like precision from it. Inside, it has three different subsystems, each operating with its own margin of error: a food database, a daily expenditure formula, and a diary filled out by the user. The weakest link is the third one, and it is not about the software.

What every tracker is made of

1. Food database. Where calorie counts come from and why they differ between apps is the subject of our article on food databases.
2. Calculating your target. Daily expenditure is calculated based on height, weight, age, and an activity factor. The formulas are broken down in our article on daily calorie requirements.
3. The diary. The place where a person logs what they have eaten. This is where the most data is lost.

The first two parts are reproducible: identical input data yields an identical answer. The third depends on what is logged and how, and it is this part that determines the final accuracy.

What the checks have shown

The accuracy of trackers has been studied through direct comparisons with research databases and methods considered to be the gold standard.

Comparing the databases of two popular apps with a research database yielded different results for different apps: one showed high agreement regarding calories and most nutrients, while the other performed worse on specific food groups and individual nutrients—particularly fat and fiber.

A check of five popular apps found a systematic tendency to underestimate calories and macronutrients relative to a reference database.

A meta-analysis of studies comparing diary apps with traditional methods reached the same conclusion: apps tend to underestimate energy and macronutrient intake.

Note the general trend. All three studies find underestimation, rather than random variation in both directions. This is a sign of systematic error, and it does not cancel out over the long term—no matter how many days you track.

Where the underestimation comes from

It is composed of several factors, and almost all of them are on the user's side, not the software's.

• Unrecorded items. A spoonful while cooking, a candy from a colleague, finishing a child's meal. Each episode is small, but they add up significantly over the day.
• Oil and dressings. A tablespoon of vegetable oil is about 110 kcal, and it rarely makes it into the diary. Less is absorbed during frying than what was poured, but it is not zero.
• Rounding down. "About a hundred grams" for a portion that is actually one hundred and forty.
• Choosing the convenient entry. When there are ten options in the list with different numbers, the hand naturally reaches for the smaller one.
• Drinks. Juice, lattes, and beer are perceived as drinks, not food. How much is actually in them is covered in our article on alcohol calories.

The second half of the calculation: expenditure

The app calculates your target using a formula, which is an estimate, not a measurement. Equations predict basal metabolic rate with an average error of a few percent for a group, but for a specific individual, the discrepancy can be larger. The activity factor is chosen from a list and is subjective: everyone understands "moderate activity" differently.

A separate trap is synchronization with a tracker. The app adds calories "burned" during a workout to the daily limit, and the user eats them. How much this number can be trusted is analyzed in our article on fitness trackers and calorie expenditure: energy expenditure is exactly what trackers estimate the worst.

Two errors then compound in the same direction: underestimating what is eaten plus overestimating what is burned. The calculated deficit ends up being larger than the real one, weight plateaus, and the person concludes they have a broken metabolism, even though it is just a matter of arithmetic.

Why people quit after two weeks

The reason is usually not laziness, but the nature of the task itself. Manual entry requires several actions for every meal: find the product, select the entry, set the weight. In total, this takes minutes a day, and over the long term, such a cost is sustainable for very few people.

Hence, there are two strategies that extend the life of the habit.

Reduce the number of decisions. Recurring meals should be saved once and then added with a single button. You do not need to enter the same breakfast five times a week.

Lower the accuracy requirements. A diary kept approximately, but every day, is more useful than an accurate one kept for three days. On how such habits are formed in general, see our article on food habits.

It is at this point that photo-based estimation becomes appropriate. It is less accurate than weighing, but it requires one action instead of five, and over the course of a month, an approximate record of every meal provides a more accurate picture than an accurate record of the first week followed by nothing.

What "accuracy" means in relation to a diary

It is useful to separate the two tasks that a tracker solves simultaneously.

The first is to find the absolute number: exactly how much was eaten during the day. Here, accuracy is poor, and all the listed errors add up.

The second is to compare days with each other: is today more or less than the average? Here, the same systematic error is present in both days and almost cancels out, so the comparison works much better than the absolute value.

Practically, this means: do not compare your number with someone else's or with a recommendation from an article; compare it with your own previous week and with what the scale shows. Why this is necessary and what it provides is broken down in the article why count calories.

Count it from a photo

Frequently asked questions

How accurate are calorie counting apps?

Checks against research databases find a systematic underestimation of calories and macronutrients. The direction of the error is consistent across different studies, meaning it is a bias rather than random variation, and it does not smooth out over the long term.

Why does the app underestimate what I have eaten?

The main contribution comes not from the software, but from the input: unrecorded small items, cooking oil, rounding weights down, and choosing an entry with a lower number from a list of similar options. Plus beverages, which are not perceived as food.

Should I eat back the calories the app added for a workout?

This figure should be treated with caution: energy expenditure is the least accurate of a tracker's metrics. If you eat the full amount added, the overestimation of expenditure combines with the underestimation of intake, and the calculated deficit turns out to be larger than the actual one.

What is more important — accuracy or consistency?

Consistency. An approximate record every day provides a more accurate picture of the month than an accurate record for the first week followed by nothing. Systematic bias is the same on all days and does not prevent comparing them with each other.

Why doesn't my number match someone else's?

Because the absolute value depends on the data source, the selected entries, and the completeness of the input. It makes sense to compare not with someone else's result, but with your own previous week and weight trends.

Read next

• Where do apps get calorie and macronutrient data and why do the numbers differ — The same cottage cheese gets different calorie counts in two apps, and people usually assume one of them is wrong.
• Fitness trackers and calorie expenditure: why they measure heart rate but estimate calories — On a tracker screen, three numbers look equally reliable: heart rate, steps, and "calories burned.
• Why count calories — Calorie counting does not burn fat.

This article is for general information. It is not medical advice, a diagnosis, or a prescription for treatment or a diet, and it does not replace a consultation with your doctor. If you have a health condition, are pregnant, take medication, or follow a diet prescribed to you, decisions about food belong with your doctor.

Figures from regulations, guidelines and studies are given as they stood when this article was prepared and may since have changed; check them against the primary sources. This article is not advertising, an offer, or individual advice, and neither the author nor the site owner is responsible for decisions taken on the basis of it.