How the Weight Impact Score Works

A calorie total answers how much energy a meal holds, not what it is likely to do to your blood sugar. Two meals with identical calorie totals can produce very different blood sugar and insulin responses, as the insulin index (Holt, Miller and Petocz, 1997) and the wearable blood sugar sensors (CGMs) in Zeevi et al. (Cell, 2015) both show. The Weight Impact Score adds a second number you can act on at the table: 1 to 10, where a lower number means a lower estimated blood sugar and insulin impact.

This is not a calorie count. It estimates blood sugar and insulin impact, not raw energy content, so it answers a different question: what a meal is likely to do, not only what it contains.

Whether the carbohydrate share of what you eat, and not only its calorie total, changes how body fat handles energy is still an open research question, and the human evidence is early. In an exploratory analysis from a weight-maintenance feeding trial, fat-tissue samples from 27 adults who had eaten 60% carbohydrate for 10 to 15 weeks showed lower mitochondrial respiration in the lab than samples from people eating 40% or 20% carbohydrate (Bikman et al., 2022). A 2011 review of mostly cell and animal research by Bikman and Summers describes ceramides, fats that build up under cellular stress, as signals that shift cell metabolism in ways relevant to diabetes. Neither paper measured blood sugar or insulin after a meal, and neither author has reviewed or endorsed the score.

What moves a Weight Impact Score?

When you scan a meal, RealFoods estimates its score from the foods on the plate. The score builds on the research behind the glycemic index (Jenkins et al., 1981) and the insulin index (Holt, Miller and Petocz, 1997). It is RealFoods' own modelling, and none of the study authors built it, reviewed it, or endorse it. What tends to move a score:

• Refined carbs and sugar push it up. White bread, pasta, juice, sweets and other fast-digesting carbs are what send blood sugar soaring and call in a lot of insulin.
• Protein, fibre and whole foods keep it low. Eggs, fish, meat, nuts, vegetables and whole fruit sit at the low end.
• Less-refined carbs score lower than refined ones. Whole fruit and vegetables, with their fibre intact, score lower than juice or white flour.
• Portion size matters. The same food scores higher at a larger serving.

What does the research show?

The research here is short, controlled feeding studies that add protein, fat, nuts or fibre to a carbohydrate meal and measure blood sugar and insulin for a few hours. Most are small and cover a single meal. Their results are fairly consistent for blood sugar and mixed for insulin, and we report both.

• Protein. A 2024 meta-analysis of 154 feeding comparisons found that adding protein to a carbohydrate meal lowered the blood sugar response and raised the insulin response; per gram of protein per gram of carbohydrate, blood sugar fell about 50% in people without diabetes and 10% in type 2 diabetes, while insulin rose 56% and 76% (Wolever et al., 2024). Single trials show the same split: 25 g of protein lowered the blood sugar response to a sugar drink in 17 adults with type 2 diabetes, except egg white, while raising insulin to 190% to 360% of the sugar-drink-only response (Gannon et al., 1988); 50 g of tuna protein did the same in older adults (Meng et al., 2017); and whey with a sugar drink lowered blood sugar in younger and older men while raising insulin more than either alone (Oberoi et al., 2022), as did whey or casein in adults with prediabetes (Hoefle et al., 2015, funder listed as Mondelez International). Not every study agrees: 10 to 50 g of protein did not change the blood sugar response to glucose in 7 healthy adults (Westphal et al., 1990), and protein added to a sugar drink left blood sugar unchanged in 30 healthy adults (Ramne et al., 2023). Protein also calls in glucagon, the hormone that counterbalances insulin, while sugar on its own pushes it down: glucagon rose after every protein meal and fell after glucose alone (Gannon et al., 1988), rose with the protein dose (Westphal et al., 1990), and roughly doubled after whey or casein drinks (Hoefle et al., 2015).
• Fat. In 10 healthy men, butter eaten with potato lowered the blood sugar response by about 40% and slightly raised insulin (Gannon et al., 1993), but in 7 men with type 2 diabetes it did not lower blood sugar (Gannon et al., 1993). In older adults, 11 g of fat raised insulin without changing blood sugar (Meng et al., 2017).
• Nuts. 20 g of almonds before a sugar drink lowered blood sugar and insulin in 60 Asian Indian adults with prediabetes (Gulati et al., 2023). A 28 g serving of almonds with a bagel meal lowered the blood sugar rise by about 30% in adults with type 2 diabetes but not in those without (Cohen and Johnston, 2011). Whole almonds at breakfast lowered blood sugar over the day and after lunch in 14 adults with impaired glucose tolerance, with a higher insulin response at lunch (Mori et al., 2011), and peanut butter at breakfast lowered the blood sugar response to the next meal, not to the breakfast, in 15 women with obesity (Reis et al., 2013). A walnut breakfast lowered the two-hour insulin response in 28 adults with overweight or obesity (Rock et al., 2017). Several nut trials were funded by nut industry groups, noted in the sources.
• Fibre and legumes. Breakfasts with about 5 g of soluble fibre lowered the blood sugar rise in 19 adults with type 2 diabetes, with no difference in insulin (de Carvalho et al., 2017). Hummus on its own produced much smaller blood sugar and insulin responses than the same carbohydrate from white bread (Augustin et al., 2016).
• Insulin index. At equal 240 kcal portions, insulin responses tracked blood sugar overall, but protein-rich foods and bakery products raised insulin more than their blood sugar response would predict (Holt et al., 1997). A 2024 compilation lists insulin index values for 629 foods and drinks from 80 studies (Hendricks et al., 2024).

That split is one of the score's known limits: the blood sugar evidence is fairly consistent, the insulin evidence is mixed, and the score has not been checked against measured insulin.

What does a Weight Impact Score look like for real meals?

Here are scores for real, everyday meals:

Ready to stop guessing what to eat?

What does the Weight Impact Score not do?

We want to be direct: the score is a modelled estimate that has not been clinically validated against measured insulin responses, so read it as a pattern across your week, not a lab result. It is not a medical device and we do not claim it replaces clinical judgement.

• It doesn't replace a doctor's advice.
• It doesn't account for complex medical conditions, extreme allergies, or specific medication interactions.
• It doesn't measure micronutrients (vitamins, minerals). You can have a very low score and still need to eat your vegetables for long-term health.

How is this different from a glucose monitor?

A glucose monitor shows your blood sugar; insulin is likely what decides whether you keep burning fat. The Weight Impact Score estimates that second thing, how much insulin a meal is likely to call in, which a glucose reading does not show.

A monitor also measures one person on one day, while the score is an estimate you get before you eat. Zeevi et al. (Cell, 2015) found that people's glucose responses to identical meals varied widely, so the score estimates a typical response, not yours, and it has not been validated against measured insulin.

How confident is a Weight Impact Score?

Confidence depends on how much the app knows about the meal and how closely your body matches a typical response. These are the limits we know about:

• Input quality. A barcode scan of a packaged food with a full nutrition label gives the most complete inputs. A photo or a typed description means the app has to infer ingredients and portion size, so those scores carry more uncertainty.
• Your own response. Zeevi et al. (Cell, 2015) found that people's blood sugar responses to identical meals varied widely. The score estimates a typical response, not yours specifically.
• Context the app does not see. Sleep, stress, physical activity, illness and medication all change how your body handles a meal, and none of them are inputs to the score.
• No clinical validation yet. The score has not been tested against measured insulin responses in a clinical study.

The practical reading: use the score to compare options and to spot patterns across a week. Treat a difference of a point as a tie, and look at the drivers the app lists for the meal rather than the single number. If you take medication that affects blood sugar, talk to your doctor first; your measured blood sugar and their advice come before this score.

Common questions

What is 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.

Is there an app that gives food a Weight Impact Score?

Yes. RealFoods, an iPhone app, gives any meal a Weight Impact Score from a photo, text description or barcode scan. RealFoods is free to download. Barcode scans, product search and manual meals include the full score. Premium (AI photo and text meal scoring) has a 7-day free trial, then $9.99 a month or $59.99 a year.

Is the Weight Impact Score the same as calories?

No. Calories measure the energy in food; the Weight Impact Score estimates how strongly a meal is likely to raise blood sugar and call in insulin. Two meals with identical calorie counts can have very different scores, because refined carbs and sugar push a score up while protein, fibre and whole foods keep it low.

What makes a Weight Impact Score go up or down?

Refined carbs and sugar push a meal's score up, and protein, fibre and whole foods keep it low. Research on the glycemic index (Jenkins et al., 1981) and the insulin index of foods (Holt et al., 1997) is the background, but the score itself is RealFoods' own modelling.

Curious to read more? See how the score compares to the problem with calorie counting, or explore our frequently asked questions.