Where the numbers drift
Self-reported food intake is one of the least reliable measurements in nutrition research. Doubly labeled water studies, which track energy expenditure objectively, repeatedly find that people under-report what they eat, often in the range of roughly 20 to 40 percent, with under-reporting tending to be larger among individuals with higher body weight. Restaurant and packaged-food labels add their own error: regulators in several regions allow a tolerance of around plus or minus 20 percent against the stated value.
Errors compound from several sources at once: forgotten snacks, eyeballed portion sizes, cooking oils that go unlogged, and natural variation in the food itself. Two apples of the same listed weight can differ in actual calories.
Why tracking still works despite the error
Imperfect counting is still useful because the error is often consistent for a given person. If someone under-logs by a similar margin week to week, the trend in their intake estimate still moves in the right direction, and they can adjust based on how body weight responds over several weeks rather than trusting the absolute number.
This is general educational information rather than personalized advice. Counting calories is not appropriate for everyone, particularly those with a history of disordered eating, and a registered dietitian or physician can advise on whether and how to track.