Keeping score - will a single number help you change your behavior?

Jun 19, 2025 9:14am

Nutrition, health and food scores have been used to nudge, educate and inform for decades. Artificial intelligence as however opened up new avenues of using scores as a behavior change or modification tool. This article dives into the type of scores currently on the market and the state of the science of the their effectiveness on behavior change.

 

Nutrition, health and wellness are top priorities for many individuals across the globe. Currently, the nutrition industry is already estimated at $2Trillion with new solutions being launched on a daily basis. Rapid advances in generative AI, Natural language processing and Machine learning, are fuelling the growth of new solutions that leverage AI to meet the complex needs and preferences of individuals by combining numerous data points via an app or a platform. These solutions powered by Artificial intelligence (AI), promise to quickly curate, synthesize, analyze and then recommend what, why and how to “X” only then to realize…. that they actually have to do the work. This is where behaviour change becomes a challenge and is not a quick fix, even with the best intentions.

Just about everyone and their mate has a smartphone, which makes accessing information, knowledge and services even faster thanks to the web and computing power.  Plus, with individuals increasingly becoming overwhelmed with too much data and dashboards, more and more individuals are demanding simplicity and transparency in order to act on the personalised advice. One way of achieving this without having to think, decipher, or doom scroll is one single number. This is where our love for scores comes into play.

Scores are most often a numerical or alphabetical single measure that is composed of different data points and weightings synthesized into one single number frequently used as a behavior change technique. In the context of health, a score can be based on personal, biological, psychological (mood), physical activity, sleep, or nutritional intake to name a few. As the input from the individual changes through dietary tracking or doing exercise, the score changes, and it tells them how to get into a more “optimal” zone through specific actions.

But are these single scores sufficient to change consumer behavior? Is the science strong for this approach? In this article, we dive deeper into the role of using scores to drive behavior change.

 

The role of scores in Personal health and prevention

Scoring systems or scores, are grounded in behavioral science or statistics. The purpose of the scores is to:

• Enhance knowledge and inform
• Creating opportunity
• Boosting motivation

Ultimately the role of scores is to bridge the intention-action gaps by making abstract health goals tangible and providing instant feedback.

 

 

The role of scores in Health and Digital health

Health scores have been in use for decades if not centuries, generally to calculate someone’s risk of developing a chronic condition such as heart disease or Diabetes.

For instance, a Dr could tell their patient that they have a 1 in 5 chance of having a stroke if they continue smoking or drinking or do not reduce their body weight.

The purpose of a score is to quickly grasp how high an individual’s risk is without going into detail about the science or even how the score was calculated. People can gauge how it applies to them and where they fall and act on the advice or recommendation.

In precision health, advances in technology across different omic technologies, polygenic risk scores have been used to determine someone’s risk of developing a chronic condition. In this case, if you have inherited a number of high-risk gene variations that increase your predisposition of developing a specific condition, your score would be higher. If you have only inherited a few, your polygenic risk score would be lower. However, in reality, the assumption has been that if you know that you have a higher risk that your behavior would change accordingly. However, this is not as simple. A recent systematic review found that there is no consistent or strong evidence that receiving something like a polygenic risk score has an impact on behaviour (Driver 2022).

 

Scores are most often a numerical or alphabetical single measure that is composed of different data points and weightings synthesized into one single number frequently used as a behavior change technique. 

The role of scores in Personalised nutrition and wellness

In the digital nutrition and health world, there are several scores or ways of scoring food, ingredients that have an impact on health and therefore have been increasingly used as a behavior change tool.

The journey of scores in personal health and nutrition has been very interesting to follow, which is why we will outline a timeline of using scores as a behaviour change tool according to our Qina data. Based on our market data, it is clear that the number of companies using barcode scanning and meal logging is on the rise, owing to a rise in consumer interest. Industry leaders such as Oura, Myfitnesspal and Whoop are increasingly leaning into nutrition in order to understand what consumers are actually consuming. They achieve this by incorporating the latest in tech (Computer vision) that makes it convenient to track, recognise and log food items.

 

There have been a range of different scores developed some of which we will go into below as well as how they have been developed.

 

Types of scores currently in the market:

• Nutritional signposting
• Metabolic score
• Gut score
• Biological age score
• Lifescore
• Habit score
• UPF (Ultra processed food) score

 

Healthy eating swap scores

In 2012, apps became more popular and widespread which lead to the launch of several apps that used a decision tree approach to guide consumers to better options. These included apps such as Foodswitch launched in the UK and Australia which were government funded and helped consumers to switch from their current products to healthier alternatives which could include products higher in fiber or lower in salt.

A small study showed that the apps were effective in reducing salt intake.

 

Meal/recipe scores

By 2015 Personalised nutrition companies started to develop their own scores creating their own ai apps that could combine different data sets and calculate a score in seconds. The AI was trained on product SKU’s, nutritional information as well as clinical and public health guidelines in order to match the personal and preferences of the user to the other data sets. All you had to do is set up your profile, stipulate your preferences and avoidances and then start scanning the barcode to check if a food is suitable for you or family member and suggest any alternatives. Companies that provides these smarteating apps include Smartwithfood, Spoonguru and the FoodHealth company. In most cases the apps were used to guide shoppers in retail or online stores.

 

Metabolic health score

By 2016 the score become more sophisticated, personalised nutrition companies such as Day two (now ceased), created a score that guided you towards the foods that would best match your microbiome profile. This was followed by the launch of Habit in 2017 that provided a score based on your metabolic flexibility. In order to improve your metabolic flexibility you needed to consume specific foods and exercise.

By 2019 companies such as Levels, January AI and Signos and Veri (acquired by Oura) launched their metabolic health solutions that consist of wearing a CGM  and using an app that created a metabolic score based on your CGM data. In order to improve your score you could scan the barcode of products and see whether the food matches your metabolic profile. In addition recipes are also scored to predict how they would essentially influence your blood sugar levels.

 

Nutritional signposting

Before AI was really thing, scores were used to guide food behaviors through the use of score on food packing such as the Nutriscore which was launched in 2017. This front of labelling was meant to improve public health and guide shoppers towards more healthier options. The score was based on the nutritional information by assessing the nutrient composition per 100g as it relates to the public health guidelines. The score was easy, colourful and meant to be a convenient way of assessing the healthfulness of a product. From a scientific perspective, the Nutriscore led to consumers choosing healthier cereals in a Dutch cohort whilst (Van den Akker, 2021) and a recent systematic review concluded the same, that the Nutriscore led to improved purchasing behaviors (Andreani et al 2025). However, the Nutriscore also caused a lot of disagreements and there is still a lot of political debate about the Nutriscore, with Nestle recently announcing that it is discontinuing the score. However, the Nutriscore it is not the focus of this article. Nutritional signposting has been used as a tool to improve public health and would be considered at the lower level of personalisation as the scoring does not provide insight into how the Nutriscore in this case would fit with an individuals’ preferences, needs, cultural habits etc.

 

Predictive scores

By 2021, the personal health company Zoe launched its AI app and platform based on a battery of tests such as bloods, continuous glucose levels and a microbiome test as well as a dietary assessment. After a couple of weeks, a customer’s results would be revealed and they would receive their personalised metabolic and microbiome score. This meant that every food, ingredient, product or recipe would get a score and users could make a strategic personal decision whether to eat a specific food before consumption and decide whether it was worth the impact on their blood sugar levels. These predictive scores are also frequently referred to as digital twins.

 

Planetary scores

By 2022, consumer awareness about the environment, climate change, global warming and the societal impact of their choices was wearing heavy on the minds of consumers. Consumers choices are now not only influenced by cost, taste and convenience but also health and sustainability. Companies such as Greenchoice as well as Smartwithfood launched an eco score to help consumers towards choosing foods with lower carbon footprint and many other parameters important to consumers such as animal cruelty, no child labour and fair pay.

Despite the importance, the research actually showed that when an Eco score is shown together with a nutritional score, this did not influence shopping behavior. Research showed that Eco-Score ratings (A–E) significantly influence purchasing decisions, especially among environmentally conscious consumers. In addition, products with Eco-Score A experience higher purchase intent, while those labelled “E” face rejection, even if priced competitively. However, the impact of the score is influenced by pre-existing environmental concern. This means that individuals with low engagement or concerns may ignore sustainability scores altogether.

 

Biological age scores

By 2022 the Longevity industry starts heating up and we started seeing an explosion of  companies claiming to determine an individual's biological age, by using their proprietary algorithms. These algorithms calculate whether your real age matches your biological age. These biological age score use a variety of data points such as blood biomarkers, lifestyle data, epigenetic data, sleep and nutrition. Companies leading the charge include Mudho, Agerate and FOXO. While fascinating and potentially motivating, experts have expressed growing concern over the accuracy of these scores considering that they are calculated differently and the science on which they are based is shaky. Following that, there is limited research showing a clear link to knowing one’s biological age and subsequent healthy behaviors, however lotions, potions and peptides appear to be in plenty supply!

 

 

Food/Health score

By 2023 things got turned up a notch again by consumers increasingly focusing on clean(er) eating and the Food as medicine movement. Consumers got excited about the prospect of using foods to heal or treat their conditions (link to blog), yet surveys show that actually, many consumers do not understand what the term food as medicine actually entails. Nevertheless, innovators to the rescue, and companies such as Foodsmart, Instacart, Bitewell (now Foodhealth company) as well as Sifter served up AI-powered apps to match personal, health and preferences to physical food products.

I would love to say that we have plenty of evidence to show that using these AI apps actually leads to better food choices and dietary behaviors but the truth is we don’t yet know. As none have done any research specifically comparing their app to the real world.

 

UPF score

By 2024, the MAHA movement is gaining traction, GLP-1 use is skyrocketing and transparency is high on the agenda across the globe and the topic of ultra processed food dominates the headlines, government meetings and inside closed door at food companies. Obesity and Diabetes rates continue to rise with clear links to UPF’s (Hall et al 2019), and it is clear that public health approaches through product scoring has (to date) little impact. Consumers are more demanding, they want transparency and they are actively looking for healthier but still affordable alternatives. But the debate on what constitutes “processed” and “ultra-processed” rages on….. a topic we won’t go into here either. Companies such as Zoe, Yuka (56 million users in 2023), Wisecode and Gococo have developed scores to make it easy to understand how processed a food is, again how these were developed differs.

Yuka combined food related information into a single score which is calculated out of 100 by combining the Nutritional information (ie nutritional breakdown and nutrient ratio’s), the risk of the nutrient to health based on the available evidence base as well as a product’s organic certification (Source Yuka website), while Zoe looked at “hyperpalatibilty” and “energy intake rate”

So far, it is too early to tell whether the use of these scores actually leads to a reduction in the intake of UPF’s or even better higher healthy eating index overall (however you define that).

 

Life score

Health is determined by many factors including of course social determinants. This is why it is important to consider things such as access to fresh fruit and vegetables, how strong a habit is, as well as Quality of life or wellbeing, not just hard health outcome measures. This systems thinking towards health has led to the development of a Lifescore which takes several factors into consideration. Lifesum, a Swedish company has developed their Lifescore by using a number of datapoints collected from their users. Earlier this year, Insidetracker recently launched a Habit score which calculates how a lifestyle habit contributes (positively or negatively) to your overall health. However, the impact of tracking the scores on behaviour change is too early and we are eagerly awaiting for a few publications to fill the gap in our knowledge.

 

Reflections

It is clear that the advent of AI has opened the opportunity to combine large datasets as massive computing power to food information which was historically hidden. Heightened consumer interest and dwindling trust in commonly consumer products have meant that the industry has had to react and reformulate. Just recently Kraft Heinz announced that it would be removing preservatives from their products. This demonstrates that this trend is pushed by consumers, and we are only at the beginning.

In other news IRX FOODSCORE recently announced that they have received a patent for their food scoring system.

Considering the plethora of scoring solutions we have on the market, it would be interesting to see whether consumers actually care about a patent, or whether app use is purely driven by the UX and completeness of the product database.

Ultimately, the lessons over the years have been that it is difficult to create a single score that please everyone and that stands the test of time. It remains important to develop different tools that can educate, inform and inspire large groups of the population who have different food and health literacy levels. However, consumers are voting with their wallet and want to know whether a food, meal, ingredient, supplement is suitable for THEM. That is the elephant in the room and the challenge the industry needs to prove.

 

It is clear that the advent of AI has opened the opportunity to combine large datasets as massive computing power to food information which was historically hidden. Heightened consumer interest and dwindling trust in commonly consumer products have meant that the industry has had to react and reformulate.

 

Current challenges and gaps

Scores have their place in health, in healthcare and in prevention as demonstrated in this article, however we must challenge these at every step and think about how we promote the use of them carefully. The challenge of using simple scores lies in balancing simplicity with accuracy, ensuring ethical AI integration, and addressing diverse user needs, especially those with lower levels of digital literacy.

 

The questions we should constantly ask include:

• Who developed the score?
• What data, gold-standard, population are they based on?
• Who approved the final score?
• How often is the score updated?
• How relevant are the scores based on location, race, ethnicity, culture?
• Who has influence over the score? Eg does a solution provide a better score for a food product if there are kickbacks? How would you know?
• How inclusive is the food database the score is based on?
• What are the proven health outcomes on following the score?

 

Future outlook

While I don’t see folks scanning products in store here in Portugal, perhaps they are doing at home or in other countries. Using a digital AI app score should be informative and an easy experience, but this should not detract users, or make users lazy to use common sense. In my views, using scores at the lower levels of personalisation for example recipes, meals plans, products, supplements to find better alternatives or to discover new products may be a better and more digestible (excuse the pun) way without causing too much psychological harm.

The future will most likely include more hybrid scoring models, in terms of real-time biological scoring, as well as AI driven (mood) assessments coupled with real-time messaging and feedback that is relevant to the individual and can impact public health.

 

Bottom line

We have seen a shift away from nutrition scores as they were too simplistic, not transparent, limited and quite frankly, still misunderstood. Scores based on biological data look very simple but as they are based on more complex data, need validation. Despite this, it is clear that as humans, we like simple nice round numbers, just consider that Yuka boasts 3.6million scans per day, meaning that food and health scores are not going away anytime soon. However, the evidence shows the potential of scores to change behavior, however we do need companies to be transparent and do the scientific work to demonstrate that this is the case, especially when including biological data.

So next time you get a score, stop and ask yourself, does this really reflect how I think and feel today? Is it enough for me to want to do better tomorrow?

My Name is Mariette Abrahams, CEO & Founder of Qina, Thought-leader in Personalized nutrition, health & wellness, Speaker and Entrepreneur. To learn more about how we help companies create better products through science, data and research, get in touch.

 

 

References:

1. Driver MN, Kuo SI, Dick DM. Returning complex genetic risk information to promote better health-related behaviors: a commentary of the literature and suggested next steps. Transl Behav Med. 2023 Feb 28;13(2):115-119. doi: 10.1093/tbm/ibac071. PMID: 36125098; PMCID: PMC9972341.
2. Karolien van den Akker, Dimona Bartelet, Lotte Brouwer, Syrah Luijpers, Tobias Nap, Remco Havermans, The impact of the nutri-score on food choice: A choice experiment in a Dutch supermarket, Appetite, Volume 168, 2022, 105664, ISSN 0195-6663, https://doi.org/10.1016/j.appet.2021.105664.
3. Andreani, G., Sogari, G., Wongprawmas, R., Menozzi, D., & Mora, C. (2025). Nutri-Score and Eco-Score Labeling: A Systematic Review of Their Impact on Consumer Understanding, Attitudes, and Behaviors. Food Reviews International, 1–25. https://doi.org/10.1080/87559129.2025.2460052
4. https://world.einnews.com/pr_news/793287155/ai-nutrition-leader-expands-trademarks
5. Hall KD, Ayuketah A, Brychta R, Cai H, Cassimatis T, Chen KY, Chung ST, Costa E, Courville A, Darcey V, Fletcher LA, Forde CG, Gharib AM, Guo J, Howard R, Joseph PV, McGehee S, Ouwerkerk R, Raisinger K, Rozga I, Stagliano M, Walter M, Walter PJ, Yang S, Zhou M.Cell Metabolism.2019 May 10. pii: S1550-4131(19)30248-7. doi: 10.1016/j.cmet.2019.05.008. PMID: 31105044.