What did a large two-year trial compare, and what did it find?
Four ways of dividing the same reduced-calorie day. In a trial by Sacks and colleagues, published in the New England Journal of Medicine in 2009 and read on 17 September 2026, 811 overweight adults were assigned to diets aiming at 20, 15 and 65 percent of energy from fat, protein and carbohydrate; 20, 25 and 55; 40, 15 and 45; or 40, 25 and 35. The diets used similar foods, participants in every group were offered group and individual sessions, and the trial ran for two years.
At six months, every group had lost about 6 kg on average, 7 percent of starting weight, and the groups began to regain after a year. At two years the differences between splits were small and not statistically significant. The authors' conclusion: "Reduced-calorie diets result in clinically meaningful weight loss regardless of which macronutrients they emphasize."
| Comparison at two years | Shares compared | Average weight lost |
|---|---|---|
| Protein | 15 percent against 25 percent | 3.0 kg against 3.6 kg |
| Fat | 20 percent against 40 percent | 3.3 kg in both groups |
| Carbohydrate | 65 percent against 35 percent | 2.9 kg against 3.4 kg |
Bottom line: in this trial, four reduced-calorie diets with different splits ended with similar average weight loss at two years.
Two more findings sit in the same abstract. Hunger, satiety and satisfaction with the diet were similar across the four groups. And attendance at the group sessions was strongly associated with weight loss, about 0.2 kg per session attended. That is an association, not proof that the sessions caused the loss, and it is one trial; and its averages describe groups of people, not what any one split will do for you.
If the split matters less, what is it for?
For living with the calorie target you chose. A split decides what a day looks like on the plate: pasta and fruit, or eggs and cheese, or something between. In the trial, weight lost was similar whichever of the four splits people were given, all of them reduced in calories; the person eating may care a great deal, and a day that suits you is one you may find easier to repeat. That is a reason to pick a split, not a promise about what it does.
Training is the other reason to set one. Setting a protein target first and letting carbohydrate and fat follow is one way to do it during a deficit; that choice rests on sports-nutrition positions about keeping lean mass, not on weight lost, and the protein-first article goes through what those positions say. For everyone else, a calorie target alone is a complete target, and the macros are simply shown; the simplest macro tracker article compares grams and percentages.
What a split cannot do is replace the calorie total. Two days with the same shares and different totals are different days; in this trial, diets with different shares, all reduced in calories, ended in similar places for weight. The day total against your target is the figure a record can hold you to; the split is how you chose to reach it.
How does IZeat record the split you choose?
IZeat is the calorie tracker inside ChatGPT, Claude or any other assistant. Tell the assistant you already use what you ate instead of searching for foods and entering them one by one. A target is set the same way, in the conversation, as calories and, if you want, protein, carbohydrate and fat in grams; any one of the four can be set alone.
If you think in percentages, the grams are a short calculation: a 1 800 kcal target split 30 percent protein, 40 percent carbohydrate and 30 percent fat is 135 g of protein and 180 g of carbohydrate at 4 kcal per gram, and 60 g of fat at 9, the factors of the European labelling regulation, read 14 September 2026. If you ask, the assistant can do that arithmetic as your general assistant, kept apart from IZeat's reply; IZeat does not calculate targets, and stores only the grams you confirm. For a target starting Monday 21 September, the reply states it as stored: "Target since 2026-09-21: 1800 kcal · P 135g · C 180g · F 60g."
From that day, each meal's receipt and each read of the day give the macros eaten and, for every figure with a target, what is left: grams remaining, or grams over, in the same plain words. When reports are enabled, the day card the assistant can show draws the day's energy as three shares and the macro targets as three bars; a macro you set no target for shows its total and no remainder. Changing the split is a new dated target, as the article on how a target is set and changed explains.
Thursday afternoon, after a lunch of chicken, rice and yoghurt: "What is left for dinner?" IZeat's reply, which begins "IZeat ·", gives the calories left and the grams left of each macro, say 610 kcal, 52 g of protein, 70 g of carbohydrate and 24 g of fat. If you ask what to eat, the assistant may suggest a dinner that fits, in its own voice, kept apart from IZeat's reply; what you then eat is what gets recorded. The limit, plainly: IZeat does not choose the split, does not judge it, and does not know what your body used; it shows the distance to the numbers you picked.
Frequently asked questions
Short answers to the questions people type, each one sourced above.
What is the best macro split for weight loss?
A large two-year trial found no clear winner. In 2009, Sacks and colleagues assigned 811 adults to four reduced-calorie diets with different shares of fat, protein and carbohydrate; after two years the weight lost was similar across all four, and the authors concluded that such diets work regardless of which macronutrients they emphasize. Which split to pick is not a question this page answers, and IZeat records whichever one you set.
Do macros matter if calories are the same?
For weight, the 2009 trial found smaller differences than the argument around macros suggests: with every diet reduced in calories, 15 against 25 percent protein, 20 against 40 percent fat and 65 against 35 percent carbohydrate all ended within 0.6 kg of each other at two years, none statistically significant. Macros still matter for what you like to eat and, for people who train, for protein; one trial on averages does not say what a split will do for you.
How do I turn percentages into grams?
Multiply the calorie target by each share, then divide by 4 for protein and carbohydrate and by 9 for fat, the factors of the European labelling regulation. A 1 800 kcal target at 30 percent protein, 40 percent carbohydrate and 30 percent fat is 135 g of protein, 180 g of carbohydrate and 60 g of fat. You can ask the assistant to do it; IZeat stores the grams you confirm, dated from the day you say.
Does IZeat recommend a macro split?
No. IZeat stores the targets you choose, calories and any of the three macros in grams, compares each day to them, and never picks, adjusts or judges them. If you ask, the assistant may help you think about a split as your general assistant; that discussion is its own, kept apart from IZeat's figures, and only the numbers you then choose are stored.
Can I set only a calorie target and ignore macros?
Yes. A calorie target alone is a complete target: each day shows the calories eaten and what is left, and the protein, carbohydrate and fat are still recorded on every item and shown as totals, with no remainder since none was set. You can add a macro target later from the day you name, and the earlier days keep being read as they were.
Does the trial mean low-carb diets do not work?
No, and it did not test them. The lowest carbohydrate share in the 2009 trial was 35 percent of energy, so very-low-carbohydrate and ketogenic diets sit outside what it compared. What it shows is narrower: among four reduced-calorie diets between 35 and 65 percent carbohydrate, weight lost at two years was similar. For a diet outside that range, the trial says nothing either way, and IZeat records whichever split you choose.