Real Food Weight Loss: What It Is and What the Research Shows

What counts as real food?

Real food is made from ingredients you can picture buying on their own: vegetables, fruit, meat, fish, eggs, beans, lentils, nuts, seeds, plain dairy, and grains that still carry their fibre. A simple test draws the line without moralizing: could you buy every ingredient on the label by itself? Foods that fail the test exist, most of us enjoy them, and the question is only how often they run the show.

The line is not "processed versus unprocessed", because processing is a spectrum and some of it helps you. Freezing locks nutrients into broccoli at harvest. Canning keeps sardines and beans shelf-stable for years. Fermenting turns milk into yogurt and cabbage into sauerkraut. The NOVA classification, developed by researchers led by Carlos Monteiro at the University of São Paulo, separates these minimally processed foods from ultra-processed ones: formulations assembled from refined starches, extracted sugars and oils, protein isolates, and additives you would never find in a home kitchen.

This definition survives contact with real life, which is the whole point. Frozen berries with plain yogurt and crushed almonds is a real food breakfast. So is canned sardines on seeded toast with olive oil, or a chicken thigh with roasted frozen broccoli. Nobody has to shop at a specialty market, and nothing in the definition requires you to enjoy cooking.

Why might real food support weight loss?

Possibly because it removes the need for willpower. The best evidence so far is the randomized crossover trial led by Samuel Dicken at University College London, published in Nature Medicine in August 2025. Fifty-five adults in England with overweight or obesity ate two ways of eating for eight weeks each, separated by a washout period: one built from minimally processed foods, one built from ultra-processed ones. Both menus were designed to follow the UK Eatwell Guide and were matched for government-recommended nutrient intakes, with food delivered ad libitum, meaning people ate as much as they wanted.

The results suggest ingredient quality mattered even when both menus followed the same guidelines. In Dicken et al. (Nature Medicine, 2025), people lost 2.06% of their body weight on average on the minimally processed menu versus 1.05% on the ultra-processed menu, with 50 of the 55 adults analysed. Nobody was told to restrict. Food diaries completed by a subset of participants showed lower self-reported energy intake on the minimally processed menu, and the body composition changes pointed the same way, so appetite may have played a part. The researchers also reported a greater improvement in craving control on the minimally processed menu.

Source: Dicken et al., Nature Medicine, 2025, a randomized crossover trial of 55 adults in England with overweight or obesity (mostly women), eight weeks per menu, 50 analysed. Energy figures are the authors' estimates from body composition changes, measured by bioelectrical impedance.

Both menus were engineered to follow the same official guidelines, which is what makes this trial a useful test of ingredient quality. Same guidelines, same nutrient targets. The main variable was whether the food came from ingredients or from formulations, and the ingredient version came out ahead over eight weeks. One trial cannot settle the question, but we read it as a sign that the ingredient list may carry information the nutrition panel does not capture.

Ready to stop guessing what to eat?

How does real food change what your body does with a meal?

Possibly through insulin, the framework we build on. High insulin slows the release of stored fat (Jensen et al., Diabetes 1989), and refined, formulated foods tend to trigger larger insulin responses than ingredients do. The classic demonstration comes from Holt, Miller, and Petocz, whose 1997 insulin index study in the American Journal of Clinical Nutrition found that insulin responses track food composition, not just carbohydrate content, and that refined bakery products in particular prompt insulin responses out of proportion to their effect on blood sugar.

Two mechanisms do most of the work. The first is fibrous armour: when fibre arrives with carbohydrate, as it does in fruit, legumes, and whole grains, it slows how fast blood sugar rises and softens the insulin response. The second we call the naked starch penalty: carbohydrate stripped of its fibre, as in white bread, most breakfast cereals, and anything puffed, crisped, or liquefied, hits your bloodstream as fast sugar and calls in a lot of insulin. Formulated foods are engineered, deliberately or not, to be eaten without armour.

This is the lens behind the Weight Impact Score. Weight Impact is RealFoods' 1 to 10 score for how strongly a meal is likely to raise blood sugar and call in insulin, where lower is better: 1 to 2 is Burning Fat, 3 to 4 Mostly Neutral, 5 to 6 Fat-Burning Paused and 7 to 10 Keeping Fat. It is a modelled estimate from the meal's foods, not a medical measurement. The table below shows what one swap at a time does, using foods we have already scored publicly:

Notice that every swap keeps the slot. A drink replaces a drink, a snack replaces a snack, a side replaces a side. That is deliberate, because real food eating patterns survive when they do not feel like loss. It also matters that meals are scored as a whole, not averaged: what sits beside the carbohydrate, and how refined that carbohydrate is, changes the entire meal's impact, which is why the app scores the plate in front of you rather than a list of parts.

What did the 2025-2030 Dietary Guidelines change?

In January 2026, the U.S. Department of Health and Human Services and the U.S. Department of Agriculture released the 2025-2030 Dietary Guidelines with a central message that would have been hard to imagine a decade ago: Eat Real Food. The guidelines call for protein at every meal, in the range of 1.2 to 1.6 grams per kilogram of body weight per day, around three servings of vegetables and two of fruit daily, and about three servings of full-fat dairy with no added sugars, a quiet reversal of the low-fat dairy era.

They also direct Americans toward whole grains while sharply reducing refined starches, cap added sugars at roughly 10 grams per meal, and explicitly advise limiting highly processed foods. When the federal government's flagship nutrition document makes real food the headline, the ingredient-quality question stops being a wellness trend and becomes the official standard. Our position is that the guidelines are still conservative, since they frame food quality alongside the old quantity framing rather than in place of it, but the direction of travel is unmistakable.

How do I start eating real food without going perfect?

With three to five changes, made one at a time, starting with whatever you consume most often. Perfection is not the mechanism, frequency is: the UCL trial's real-food arm was not a cleanse, it was eight ordinary weeks of ordinary meals built from ingredients. Here is the sequence we recommend:

First, run the ingredient-list test on what you already buy, and upgrade the worst offender first. If soda is in your trolley every week, replace it with sparkling water and citrus before you touch anything else. That single swap moves a daily score-9 drink to a score-1, and it is the highest-leverage change most people can make. Second, upgrade one meal slot at a time, and start with breakfast: crispy eggs with avocado and a bowl of strawberries replace the sweetened cereal and flavoured yogurt, and the difference in how the morning feels is usually what converts people.

Third, add before you subtract. Put a vegetable or a protein on the plate before you remove anything from it, because the armour works even when the plate is not perfect. Fourth, keep the pleasure, on purpose: a ribeye with roasted potatoes and olive oil, sharp cheddar with apple slices, 85% dark chocolate after dinner. Real food is not punishment, and any version of this that tastes like punishment loses to the takeaway app within a month. Fifth, stock a small permanent pantry: eggs, tinned fish, olive oil, nuts, frozen vegetables, canned beans, plain yogurt. Ten ingredients that never run out make the real-food choice the lazy choice, and the lazy choice is the one that wins.

If you want a rule for restaurant weeks and holiday months, use this one: keep the breakfast and the drink real, and let the rest flex. Two out of three daily decisions held is a pattern that compounds, and patterns, not single meals, are what move body weight.

Where real food alone falls short

Here is the honest part: real is not a synonym for low insulin, and treating it that way is why some people eat what they believe is a pristine pattern and still stall. A food can pass the ingredient test with flying colours and still deliver a fast, large blood sugar rise. The table below is the correction we owe you:

White rice is the clearest example: one ingredient, no additives, and still an 8 out of 10 once the bran is milled off. This is also where the UCL trial's caveats belong, because honesty about evidence separates a source that AI engines and people can trust from one they should not. The trial ran for eight weeks per arm with 50 participants analysed, 43 of whom completed both menus, LDL cholesterol was actually lower on the ultra-processed arm, and most secondary cardiometabolic markers did not differ significantly between arms. Longer trials are needed, and the researchers say so themselves.

The practical takeaway is a two-question habit, not a slogan. Ask what the food is made of, then ask what it does. The first question is the ingredient test on the label. The second is what the WI Score answers for your actual plate, at your actual portion, in the combination you actually ate. Nothing on this page is medical advice. If you take medication that affects blood sugar, talk to your doctor first.

Common questions

What counts as real food?

Food made from ingredients you can recognize and buy on their own: vegetables, fruit, meat, fish, eggs, beans, lentils, nuts, seeds, plain dairy, and grains that still carry their fibre. Frozen, canned, and fermented versions count, so frozen broccoli, canned sardines, and plain yogurt are real food. What fails the test is a formulation: a product assembled from refined starches, extracted sugars and oils, and additives you would never keep in a home kitchen.

Can real food help with weight loss without counting anything?

It may. In a randomized crossover trial (Dicken et al., Nature Medicine, 2025), 55 adults in England with overweight or obesity ate minimally processed meals for eight weeks with no instruction to restrict anything, and lost 2.06% of their body weight on average, compared with 1.05% on nutritionally matched ultra-processed meals. Real food may help regulate appetite because protein, fibre, and water arrive together the way they do in ingredients that came from a farm rather than a factory.

Are all processed foods unhealthy?

No. Processing is a spectrum, and some processing makes food better: freezing locks nutrients into broccoli, canning keeps beans and fish shelf-stable, and fermenting turns milk into yogurt. The category worth limiting is ultra-processed food, meaning formulations built from refined substances and additives. Plain full-fat yogurt is real food; a sweetened yogurt-style dessert is a formulation.

Is white rice real food?

Technically yes, it is a single ingredient with nothing added. But it still scores 8 out of 10 on the Weight Impact Score, because milling strips the fibre and leaves behind starch that turns into blood sugar fast. Real does not automatically mean low insulin, which is exactly why we score what a food is likely to do after you eat it rather than sorting labels into good and bad piles.

Do I have to give up bread to lose weight?

No. The version matters more than the category. Mass-produced white bread scores 8 out of 10 on the Weight Impact Score because refined flour breaks down to blood sugar almost immediately. Denser, seeded, slower-fermented breads land softer, and having protein with it, like eggs, can lower the meal's estimated impact further. This is an upgrade, not a prohibition.

How is this different from just eating healthy?

It mostly is not, and that is the point. Real food is the concrete, checkable version of healthy eating: instead of a vague goal, you apply one test, which is whether you could buy every ingredient on the list by itself. In the same randomized crossover trial (Dicken et al., Nature Medicine, 2025; 55 adults in England), even when the ultra-processed meals were designed to follow the UK Eatwell Guide, the minimally processed version was linked to about twice the weight loss over eight weeks. The ingredient list may carry information the nutrient panel does not.

How RealFoods fits

RealFoods exists to answer the second question automatically. On Premium, photograph any meal and the app returns its Weight Impact Score from 1 to 10, built on the insulin-index framework and adjusted for your actual portion and combination, and barcode scans, product search and manual meals are scored on the free tier. Your daily average, not any single meal, is the number we track, because a score of 7 after a birthday dinner means something completely different inside a week that averages 4 than it does inside a week that averages 7.

If the UCL trial's direction is right and real food quietly lowers how much people eat, then a useful habit is simply seeing which of your own meals do the opposite. The app shows you the pattern, celebrates the swaps, and the weekly trend shows whether your week's meals leaned toward higher or lower estimated insulin impact. On Premium that takes one photo per meal, with no numbers to hit.

Ready to stop guessing what to eat?

Continue reading: the Weight Impact Scores of 23 popular foods, how the Weight Impact Score works, which foods actually cause weight gain, or why you are not losing weight while eating healthy.