Calorie Counting Accuracy

Food · §243

The number your calorie-tracking app shows is wrong — typically by 20–30%, sometimes more. Labels round low, restaurants understate, nuts deliver about a quarter less than they claim, and your memory of what you ate is the worst input of the lot. Tracking still helps people lose weight — but not for the reason most users think.

Know · As-needed Evidence Moderate Chapter Food

None of this turns calorie tracking into a precise tool — it isn't one and never was. What it can be is a calibration tool, useful as a relative signal once you stop treating the daily total as bookkeeping. The reward isn't a sharper number; it's the end of the recurring "why isn't it working" panic and a habit that survives past month three. The catch: counting is one of the harder behaviour changes, and it's a bad idea if you have any eating-disorder history.

Five error sources stack on top of each other, and they almost always lean the same way — toward "you ate more than the app says." None of them is bad faith. They are the cost of compressing a real meal into a number.

Labels are allowed to lie. The Food and Drug Administration's compliance policy gives manufacturers a 20% margin in either direction on stated nutrient values, and audits show packaged foods reliably trend high — the 200-calorie granola bar that legally contains up to 240, every bar FDA 2013 Urban et al. 2010.

Restaurants understate. A team at Tufts measured 269 dishes from chain sit-down restaurants against the calorie counts printed on the menu. Stated values were on average 18% lower than what the food actually contained; some single dishes were off by more than 200 calories Urban et al. 2011.

Nuts give back less than they claim. The standard math for converting carbs, fat, and protein into calories — the Atwater system — assumes your body digests everything. Whole nuts don't cooperate. Their fat is locked inside cell walls your chewing can't fully break. Almonds were measured at 129 calories per 28 grams against a label-calculated 173 — about a quarter less than what's printed Novotny et al. 2012. Walnuts, pistachios, and cashews behave the same way Capuano et al. 2018.

Cooking moves the number. Cooking softens starch and unwinds protein, both of which make food easier to digest — and therefore higher in calories your body actually keeps. Cooked egg protein is 91% absorbed; raw, only 51% Evenepoel et al. 1998. Heat-processed starches release roughly 10–40% more energy than raw ones, depending on the food Carmody et al. 2011. The label on a bag of rice doesn't tell you whether you'll boil it, roast it, or eat it cold the next day.

Memory is the worst input. The gold-standard way to measure what someone actually eats is to dose them with water labelled with rare isotopes and trace what comes out — doubly-labeled water. Studies that pair the isotope measurement against self-reported intake consistently find people underreport by 20–40%. In one of the original studies, obese subjects convinced they had broken metabolisms turned out to be underreporting by 47% on average Lichtman et al. 1992. The OPEN study replicated this at 32% across nearly 500 free-living adults Subar et al. 2003. And the bigger the meal, the bigger the proportional miss Wansink and Chandon 2006.

Stack them and the typical pattern is: log 1800, eat 2100–2400 — depending on whether your day was label-heavy, restaurant-heavy, or nut-and-recall-heavy. The errors don't average out. They line up.

How big the gap actually is

Each error source above is documented separately. The five together set the floor — and the floor is much higher than the apps imply.

Mandatory calorie labels on restaurant menus haven't fixed the gap on that side either. Even when chains post numbers, the numbers stay biased low, and customers themselves still underestimate fast-food meals by about 175 calories on average — labels visible or not Block et al. 2013. And tracking still helps people lose weight: a review of self-monitoring trials found tracking improved weight-loss outcomes by 1–3 kg over not tracking Burke et al. 2011. The benefit comes from food awareness, not from numerical precision. That distinction is the whole game.

Tracking when the number lies

The tracker who beats the error floor doesn't track harder. They track differently. The accuracy literacy collapses into a small set of habits.

Calibrate to your own maintenance. Eat normally for two weeks, weigh yourself daily, log everything. The logged number at which your weight stays flat is your personal-maintenance value. Aim a deficit from that, not from the algorithmic estimate any app spits out.
Use a kitchen scale. Cups and spoons compound the error. A $15–30 scale is the single biggest accuracy upgrade per dollar you can buy.
Match raw to raw, cooked to cooked. "100g rice" means very different things depending on which one your database entry assumes. Most large logging errors are this one.
Plan logged intake about 200–300 below your real target. The systematic underestimate is real; buffer for it instead of being surprised by it.
Track patterns more than totals. Where calories come from — restaurant meals, oils, handfuls of nuts — matters more than the exact daily sum. These are the high-error categories; cutting one of them shifts the whole curve.
Recalibrate every few months. Your habits drift, your body weight drifts; the personal coefficient between logged and actual drifts with them.

What most apps don't tell you

The one belief that does the most damage is also the one the interface trains hardest: that the logged number is the eaten number. Three follow-ons fall out of it.

"I'm eating 1500 calories and not losing weight — my metabolism is broken." Almost never true. The isotope studies have falsified the broken-metabolism story at this magnitude repeatedly; the people most convinced their metabolism was the problem were the ones underreporting the most Lichtman et al. 1992. What's broken is the count, not the body.

"I just need to track more carefully." Past a point, no. The floor is the label tolerances, the food matrix, the cooking, the restaurant variability — not your diligence. A user weighing every gram still can't fix the fact that the almonds in front of them deliver a quarter fewer calories than the label says, and they can't make a restaurant chef use exactly the same amount of oil twice.

"Restaurant calorie counts on the menu are reliable now." They aren't, and a chain printing the same number on every menu doesn't change what's actually on the plate that day in that location. Posting the number satisfied a law; it didn't change the food Urban et al. 2011 Block et al. 2013.

When tracking becomes the problem

The accuracy floor isn't the only reason calorie counting fails. The way it fails matters.

The plateau panic. The most common failure mode: weeks of disciplined logging, weight stalls, trust collapses, diet abandoned. The real cause is the systematic underestimate finally catching up — the "1800 kcal/day" was 2200 the whole time — plus normal weekly water-weight noise. Recalibration would fix it; the user fixes it by quitting.

The precision spiral. Tracking stops being a tool and becomes the goal. Eating becomes the thing you do to feed the spreadsheet. The link between obsessive calorie tracking and disordered eating behaviour is well-documented, especially in adolescents and young women Patton et al. 2017.

The wrong-direction buffer. A reader who has half-heard the accuracy story sometimes corrects the wrong way — eating 200 calories under the logged target instead of over it, then 200 more on top of that, ending up at 1200 actual when they thought they were at 1800. Hunger, fatigue, and binge cycles follow.

The satiety blind spot. A 1500-calorie day of ultra-processed food and a 1500-calorie day of whole food are not the same day. In a tightly controlled inpatient trial, people offered ultra-processed meals ate roughly 500 calories more per day than people offered whole-food meals matched for calories, fat, sugar, and protein — and reported similar palatability Hall et al. 2019. The tracker who hits a calorie target with the wrong inputs runs out of willpower and calls it a discipline problem. It isn't.

What you gain when you stop trusting the count

The first week feels like demotion. The number you used to treat as a budget becomes a noisy index. You weigh your food anyway, you log it anyway, but the daily total stops being the verdict on whether you "stuck to the diet." Body weight is the verdict, taken as a weekly average over noise.

The first month: the recurring stalled-diet panic stops triggering. When weight doesn't move for ten days, the response is "recalibrate" instead of "my metabolism is broken." Friends and partners stop hearing the loop. The conversation about food at dinner thins out, because the math isn't a moral question anymore.

The first year: a tracking habit that's actually still going. Most users abandon within months — the friction of pretending the count is bookkeeping wears them out. The user who knows the count is a calibration signal logs faster, second-guesses less, and stays in the practice long enough for the underlying behavioural awareness to compound. Burke et al.'s 1–3 kg outcome advantage for self-monitoring shows up reliably for people who actually keep tracking Burke et al. 2011; the people who quit get nothing.

This is not a transformation pitch. The accuracy story doesn't make you lean. It makes the tool sustainable, which is what most weight-management failures actually need.

A few adjacent rabbit holes worth knowing about:

Macronutrient targets — protein, carbs, fat — are a separate question from total energy, with their own accuracy story.
Alternatives to counting — the plate method, hand-portion estimation, food-group rules — get most of the awareness benefit with none of the false-precision baggage.
The other half of the math — even if you knew your real intake exactly, the calorie budget (the TDEE estimate every app gives you) has its own ±10–15% error. Both sides of the equation are noisy.
Metabolic adaptation — the body lowers its energy use during sustained calorie restriction. Real, but smaller than the tracking-error gap.
Ultra-processed food — the Hall et al. 2019 finding that food type changes intake by 500 kcal/day at matched calories is itself a load-bearing piece of nutrition literacy.

Related in the handbook

— Same trap as a glucose sensor: the absolute number is shaky, but the trend is useful. Track for calibration, then stop bookkeeping it.
— When the calorie math and the scale disagree, a body-composition scan shows what's actually changing — fat, muscle, or water.
— Like a sleep score, the calorie count is wrong in the details — track for the trend, not the exact figure.
— If counting calories is unreliable, restricting your eating window is a way to cut them without tracking every bite.
— Calorie counts are unreliable partly because ultra-processed foods are built to slip past your fullness signal.

Substance and claimed effects

Calorie counting — the practice of tracking dietary energy intake against a numeric daily target — is widely sold as a precise weight-management tool. The substance under examination is the practice itself, and the claim being audited is its implicit precision: that "1500 kcal logged" approximates "1500 kcal eaten and metabolized." The literature on tracking accuracy is robust and converges on a different picture. Five compounding error sources — Food and Drug Administration label tolerances FDA 2013, restaurant menu inaccuracy Urban et al. 2011, Atwater-factor bias for food-matrix foods like nuts Novotny et al. 2012, cooking effects on macronutrient bioavailability Carmody et al. 2011, and dominant self-report misestimation Lichtman et al. 1992 Subar et al. 2003 — produce a real-world error floor that routinely exceeds ±20–30% of true intake on a typical day. This entry covers the accuracy of calorie counting as a behaviour, the consequences for intake-target adherence and weight outcomes, what tracking still buys when treated as a relative tool, and the failure modes — both the stalled-diet pattern and the eating-disorder risk of false precision. Meta dimensions touched: short-term health (modest behavioural benefit from food awareness), cumulative body composition (indirect, small), mood (mixed; structure versus obsession), effort (substantial), evidence (strong on accuracy facts), controversy (active dispute about whether the tool is net beneficial).

Evidence by addressing question

Mechanism

The mechanism of inaccuracy is arithmetic: a tracked daily total is the sum of many noisy per-food estimates, each with an independent error source, and the errors compound because they share a systematic direction (most are biased low). Five canonical sources:

Label tolerances. The FDA Compliance Policy Guide Sec. 562.700 allows manufacturers a ±20% tolerance on stated nutrient values; in practice manufacturers calibrate to the lower-bound side FDA 2013. A bar labeled 200 kcal can legally contain up to 240 kcal, and reliably trends high in audit samples Urban et al. 2010.
Atwater-factor bias for food-matrix foods. The 4-9-4 kcal/g factors for carbohydrate, fat, and protein assume complete digestion. Whole nuts, seeds, and high-fibre foods deliver less metabolizable energy because intact cell walls block lipid release. Novotny et al. measured almonds at 129 kcal/28 g against an Atwater-calculated 173 kcal/28 g — about 25% lower true energy than the label Novotny et al. 2012. Walnuts, pistachios, and cashews show the same pattern Capuano et al. 2018.
Cooking effects on bioavailability. Cooking gelatinizes starch and denatures protein, raising metabolizable energy by roughly 10–40% for the same dry mass, depending on the food Carmody et al. 2011. Cooked egg protein is 91% absorbed versus 51% for raw Evenepoel et al. 1998. Labels typically reflect raw or "as packaged" macronutrients; the database entry the user picks rarely distinguishes states.
Restaurant menu inaccuracy. Urban et al. measured 269 items from American sit-down chains and found stated calories averaged 18% below directly measured values, with individual items off by >200 kcal in either direction Urban et al. 2011. The reduced-energy packaged-food category shows similar but smaller discrepancies Urban et al. 2010.
Self-report error. Dominant source. Doubly-labeled-water (DLW) studies, the gold standard for energy expenditure, consistently document 20–40% underreporting of intake versus measured expenditure in weight-stable subjects Schoeller 1995. Lichtman et al. found self-reported diet-resistant obese subjects underreported intake by 47% on average against DLW Lichtman et al. 1992. The OPEN study replicated at ~32% mean underreport across 484 free-living adults Subar et al. 2003. Wansink and Chandon showed estimation error scales with meal size — bigger portions produce bigger proportional underestimates Wansink and Chandon 2006.

A representative compound pattern: a free-living adult logging 1800 kcal/day is plausibly consuming 2100–2400 actual kcal, depending on which of the above dominate their food sources (label-heavy versus restaurant-heavy versus nut-heavy versus pure self-recall).

Evidence

Each error source above is documented with replicated primary literature. The label-tolerance regime is regulatory rather than empirical, but audit studies confirm manufacturers exploit the tolerance window: Urban et al. tested 39 commercially prepared reduced-energy foods and found measured energy averaged 8% higher than label, with packaged-meal items running 18% higher Urban et al. 2010. The restaurant 18% underestimate replicates across independent sampling; consumers themselves underestimate fast-food meals by ~175 kcal on average even when calorie labels are posted Block et al. 2013. The nut and food-matrix evidence is multi-laboratory and converges on 20–30% reductions for whole tree nuts Novotny et al. 2012 Capuano et al. 2018. Cooking-effect evidence began with the Carmody mice work and extended to human macronutrient bioavailability studies Carmody et al. 2011 Evenepoel et al. 1998. Self-report DLW validation is unambiguous and decades-deep Schoeller 1995 Subar et al. 2003.

Importantly, tracking's behavioural benefit survives the accuracy floor. Burke et al.'s systematic review of self-monitoring in weight loss interventions found tracking improved weight loss outcomes by roughly 1–3 kg over non-tracking controls Burke et al. 2011. The benefit appears to come from food awareness and accountability, not from numerical precision — a tracking-without-believing-the-numbers framework reconciles the accuracy facts with the outcomes literature.

Practice / clinical consensus

Dietitians and obesity-medicine clinicians treat calorie tracking as a relative tool, not an absolute one — useful for noticing dietary patterns and surfacing energy-dense foods, not for hitting a precise budget. Academy of Nutrition and Dietetics weight-management position statements endorse self-monitoring while explicitly acknowledging the precision floor. A common clinical calibration: instruct the patient to track an unchanged-eating baseline for two weeks, identify the logged value that corresponds to stable weight, then create a deficit from that personalized number rather than from population-level TDEE estimates. The personal baseline absorbs the systematic error sources.

Protocol

If counting is the chosen tool, the protocol that survives accuracy literacy:

Calibrate to personal maintenance. Track at stable weight for ~2 weeks; the logged value at which weight stays flat is the personal-maintenance number. Create deficits from that, not from an algorithmic estimate.
Weigh, don't eyeball. Volume measurement (cups, tablespoons) compounds estimation error. A kitchen scale eliminates one error source for ~$20 one-time cost.
Match raw versus cooked database entry to actual state. Most large discrepancies come from logging "100 g cooked rice" against a raw-rice entry, or vice versa.
Buffer in the right direction. Given the systematic underestimate, plan logged values about 200–300 kcal below the actual target.
Track patterns, not totals. Where calories come from (restaurant meals, nut handfuls, cooking oils — high-error categories) is more useful for adjustment than the precise daily sum.
Re-calibrate periodically. Database habits drift, body weight drifts; the personal coefficient between logged and actual needs refresh every few months.

Contraindications

The strongest contraindication is prior eating-disorder history. Calorie tracking is associated with onset of disordered eating behaviours and with symptom exacerbation in those with prior anorexia, bulimia, or binge-eating disorder Patton et al. 2017. Adolescents show elevated risk independently of prior history. For these populations, plate-method or hand-portion estimation delivers comparable behavioural awareness without the numerical-precision fixation. A secondary edge case: athletes or physique-focused trainees whose tracking can tip from tool to compulsion — usable but warrants self-monitoring.

Misconceptions

The dominant misconception is that the logged number is the eaten number. Three sub-misconceptions follow:

"I'm eating 1500 kcal and not losing weight, so my metabolism is broken." Almost always a self-report or compounding-error problem rather than metabolic anomaly. DLW studies have systematically falsified the broken-metabolism narrative at this magnitude Lichtman et al. 1992.
"More precise tracking yields better results." Past a point, no. The floor is the irreducible error sources (label tolerance, food matrix, cooking, restaurant variability), not the user's diligence. Diminishing returns are sharp.
"Restaurant calorie labels are reliable now that they're posted by law." Urban 2011 disproved this for sit-down chains, and Block et al. showed consumer estimates remain systematically low even with labels visible Urban et al. 2011 Block et al. 2013.

Failure modes

Plateau panic. Tracker hits the target number but weight stalls → trust in the framework collapses → diet abandoned. The actual problem is the 20–25% systematic underestimate plus normal water-weight noise. Solution is recalibration, not effort intensification.
Precision obsession. Tracking becomes the goal; eating becomes input to the spreadsheet. Strong correlation with disordered eating patterns Patton et al. 2017.
Over-restriction from buffering the wrong way. A user who has read about systematic underestimation but applies the correction in the wrong direction can end up eating ~1200 actual kcal while aiming for 1800.
Satiety blind-spot. A 1500-kcal day of ultra-processed food produces measurably worse satiety than a 1500-kcal day of whole-food intake Hall et al. 2019. Pure calorie tracking obscures the satiety dimension, leading to compliance failure that the user attributes to willpower.

Practicalities

Free apps (MyFitnessPal free tier, Cronometer free tier, Lose It free tier) cover basic tracking. Premium tiers run $50–80/year.
Kitchen scale: ~$15–30 one-time. Single largest accuracy upgrade per dollar.
Database accuracy varies wildly. User-submitted entries (most MyFitnessPal entries) are systematically less reliable than verified-database entries (Cronometer's verified entries, USDA FoodData Central). Branded packaged foods are most reliable; home recipes least.
Time cost is the dominant burden — even disciplined users report 5–15 minutes/day, and adherence drops sharply past month three.

Stakes

Acted on the false-precision belief over a realistic mid-decade story: months of frustrated tracking that "stops working," repeated diet-app cycling (MyFitnessPal → Noom → Cronometer → back to MyFitnessPal), eroded trust in any structured eating framework, and eventually a quiet conclusion that calorie math doesn't apply to one's metabolism. The mystery isn't a metabolism; it's the error floor masquerading as the user's failure.

Payoff

Acted on the accuracy reality: tracking becomes a low-frustration calibration tool. The number on the app is treated as a noisy index, not as truth. Body weight and energy feedback override the app when they disagree. Compliance improves because expectations match reality, and the recurring plateau-panic cycle stops triggering.

Out-of-scope (handled by sibling entries)

Macronutrient tracking specifically (protein, carbohydrate, fat targets distinct from total energy)
Continuous glucose monitor–guided eating
Intuitive eating, hand-portion estimation, the plate method (alternative frameworks)
TDEE estimation accuracy — even with perfect intake measurement, the budget itself has ±10–15% error
Metabolic adaptation under sustained energy deficit
Ultra-processed food's specific satiety effects beyond calorie matching Hall et al. 2019

Credibility range

Optimist case

Calorie tracking, error floor and all, remains the most evidence-backed behavioural tool for weight management. Burke et al.'s systematic review establishes 1–3 kg of outcome improvement versus non-tracking controls Burke et al. 2011. The accuracy criticisms don't undermine the outcomes: the benefit comes from food awareness, identification of high-density inputs, and daily accountability — not from arithmetic precision. A user who learns the error floor and adjusts buffers (either by personal-maintenance calibration or by planning logged intake below target) reliably hits weight goals. The right inference from the accuracy literature is not "abandon counting" but "treat the count as a relative index, not bookkeeping truth." The tool's failure modes (eating-disorder risk, obsession) are real but applicable to a minority of users who could be screened out; for the majority, tracking is a high-leverage tool whose precision claims have been misframed by app marketing.

Skeptic case

The compounding error makes calorie tracking a placebo of precision. With a 20–40% real-world floor, a user "targeting 1800 kcal" is functionally cycling through a 1500–2400 kcal actual-intake band day to day; any weight change is more attributable to the awareness effect — which alternative frameworks (plate method, hand portions, food-group rules) produce equally cheaply — than to the math. The cognitive load, daily friction, and obsession risk produce measurable eating-disorder elevation, especially in adolescents and females Patton et al. 2017. And the tracking framework hides the satiety dimension that Hall et al. showed is actually driving overconsumption Hall et al. 2019. The correct intervention is not "track more accurately" but "stop tracking and eat mostly whole foods."

Author's call

Both positions are partly correct. Tracking is a useful tool when used as a relative-not-absolute calibrator, and the accuracy literature is best read as a deflation of false precision rather than a refutation of the practice. The numbers are not lies; they are noisy estimates that work as personal-baseline indices. The literacy this entry teaches converts tracking from a frustration generator into a reasonable tool with calibrated expectations. For populations at eating-disorder risk, the skeptic case dominates and the entry's eating-disorder-history contraindication applies. For others, tracking with realistic expectations beats both naïve tracking and category-only methods on adherence and outcomes. Controversy on the meta-question of whether to track at all is real and unresolved — hence the moderate controversy score — but the accuracy facts themselves are well-settled.

Stakeholder and incentive map

App and wearable industry. MyFitnessPal, Cronometer, Lose It, Noom, Apple Health — revenue from premium subscriptions and aggregated food-database licensing. Incentive: encourage tracking as a sticky habit; minimal incentive to surface accuracy limitations.
Food manufacturers. Beneficiaries of the FDA ±20% tolerance window; systematic rounding-down of stated calories improves perceived value-density at point of purchase.
Restaurant chains. Mandatory calorie posting (US ACA Section 4205) has produced compliance, not accuracy — chains report a single corporate number that doesn't reflect prep variability across locations.
Dietitians and obesity-medicine clinicians. Mixed. Tracking is part of established protocols, but practitioners increasingly direct patients toward category-based or plate-method approaches.
Eating-disorder treatment community. Strong skeptic position; many ED programs forbid tracking apps for active patients and discourage them in recovery.
Intuitive-eating movement. Counter-counter camp; argues tracking is intrinsically harmful regardless of accuracy.
Diet-culture influencer space. Heavy promoter of precise tracking; sells confidence in the count as a service.

Population variability

Gut microbiome. Jumpertz et al. and Turnbaugh et al. document 5–10% individual variability in caloric harvest from identical diets, related to microbiome composition Jumpertz et al. 2011 Turnbaugh et al. 2006. Two users with identical logged intake genuinely absorb different amounts of energy.
High-nut and high-fibre diets. The food-matrix effect Novotny et al. 2012 means their actual intake is systematically lower than tracked. Plant-forward eaters can have meaningful tracking-versus-reality gaps in one direction.
Restaurant-heavy diets. Urban 2011 errors compound. A user eating out 4–5 times a week has the largest real-versus-tracked discrepancy.
Home-cooked, weighed, single-ingredient diets. Lowest error floor; disciplined users can plausibly land within ±10%.
Sex differences. Females are overrepresented in eating-disorder risk from tracking Patton et al. 2017.
Older adults. Self-report error tends to be larger; daily-logging friction higher.

Knowledge gaps

Personalized calorie absorption. No clinical-grade test exists for an individual's true caloric harvest rate. Doubly-labeled water measures expenditure, not absorption per se. A cheap individual-calibration tool would convert the accuracy floor from population-level error to person-level offset.
Long-term tracking outcomes. Most behavioural trials are <12 months; beyond that, drift and discontinuation dominate measurable signal.
Modality-specific eating-disorder risk. Calorie tracking, macro tracking, and mindful eating differ in risk profile, but the comparative epidemiology is sparse.
Cooking variability quantification. Most cooking-effect studies use single foods in controlled prep. Real multi-ingredient meal variability is poorly characterized.
Ultra-processed satiety decoupling. Hall et al. 2019 showed 500 kcal/day excess intake at matched macro and palatability when food was ultra-processed; the implication that the calorie framework fundamentally undercounts what matters (satiety, food matrix, processing) hasn't been integrated into either tracking apps or clinical guidance Hall et al. 2019.

Scope call. The brief named "calorie counting accuracy" with consequences across intake-target adherence, weight outcomes, satiety, and the practical limits of tracking. All four are covered. The article treats calorie counting as the substance and accuracy as the lens; meta scores reflect the practice as informed by accuracy literacy, not the bare practice or the bare literacy in isolation.

Action call. Action is know rather than do. The entry teaches a calibration of an existing practice, not a new behaviour to start. A user not currently counting calories should not start counting on the basis of this entry; one already counting walks away with a better-calibrated relationship to the numbers. The do framing would have collided with the eating-disorder contraindication.

Rating difficulties.

Mood (1). Genuinely two-sided. Realistic expectations reduce plateau-panic anxiety for moderate trackers; obsessive use raises eating-disorder risk. Settled at 1 (small positive for general population) with the contraindication carrying the downside.
Controversy (2). The accuracy facts themselves are settled science. The controversy lives one layer up: whether tracking-with-known-error-floor is net beneficial or net harmful. Active debate but not foundational; 2 felt right.
Effort burden (3). Accuracy literacy modestly reduces effort by removing precision pressure, but daily logging plus weighing is still substantial. Not bumped to 4 because the entry's framing makes the practice less demanding than naïve tracking.
Beauty (cumulative) (1). Borderline 0 vs 1. Kept at 1 because sustained weight management does feed long-term body composition, and the chain — counting → sustained → weight stable → body comp — is real even if attenuated.

Future-link candidates (not yet in catalogue).

Plate method / hand-portion estimation — the alternative framework. Direct sibling.
TDEE estimation accuracy — the other half of the calorie-math equation; flagged in out-of-scope.
Metabolic adaptation under deficit — the legitimate small effect users sometimes blame for tracking error.
Ultra-processed food and satiety — Hall 2019 inpatient finding deserves its own entry; touched twice in this one.
Macronutrient targets — separate accuracy story.
Self-monitoring and eating-disorder risk — the contraindication side could anchor its own entry given the size of the affected population.

Separate-entry candidate surfaced during writing. The doubly-labeled water methodology itself is interesting and load-bearing across multiple entries (this one, any future TDEE entry, exercise-energy-cost entries). A short methodology entry would let downstream entries cite by reference rather than re-explain.

Hard decisions. Considered separating "label tolerances" and "self-report error" into two evidence subsections; collapsed into one to keep the stacking-error mechanic visible. Considered an audience block for adolescents in the contraindications section; decided the warning callout was sufficient and the audience block would have over-segmented a small note.