Food loss and waste (FLW) are a surprisingly large global problem, with consequences for climate, land use, food security, and the resources embedded in food that is never eaten. Yet researching the problem—especially understanding where, why, and how much food is wasted—often requires expensive, labour-intensive primary data collection.
This post explains the FLW problem and its consequences; explains why data is so important for this problem and why we don’t yet have enough of it; and suggests a solution (that we are currently developing) to the data problem – a data-sharing platform, specifically for this field.
Daphi is a not-for-profit initiative working to bring FLW to zero. One of our vectors of actions is to increase both quality and quantity of the research about the topic. At the end of this post, we talk about a service we would like to make available to support FLW prevention, as discussed in the post itself.
I used an LLM to which I prompted the topic, the goal of this document, some notes on style, external resources, and some materials I already had written. I then fact-checked the entire document, merged data from different sources that the LLM failed to merge, and made some phrasing and style adjustments. I then sent it to Roy Schulman and Edo Arad for review (thank you guys!) and made final edits.
All mistakes are mine; and all opinions I form based on these sources are mine and may not represent their authors’ opinions.
A large share of the food produced for human consumption never gets eaten.
According to the UN Environment Programme's (UNEP) Food Waste Index 2024, around 1.05 billion tonnes of food were wasted in 2022 at the household, food service, and retail stages alone—around 19% of the food available to consumers. This is in addition to an estimated 13% of food being lost earlier in the supply chain, between post-harvest and retail.
Hence, the total annual global food waste amounts to about 1.75 billion tonnes[1], and the proportion of food that is wasted is about a third of the entire global food production each year.
The consequences are not limited to the value of the food itself.
Food that is produced and then wasted also represents wasted agricultural land, water, energy, labour, transportation, refrigeration, and other inputs. UNEP estimates that FLW are associated with 8–10% of global greenhouse-gas emissions (GHG) and around 28% of the world's agricultural land being used to produce food that is ultimately lost or wasted.
The scale of the problem has also attracted attention from researchers and organizations working on climate and global priorities. For example, Project Drawdown models reducing FLW by 25-50% as the eighth most impactful intervention from over 50 interventions rated as “highly recommended” (in addition to some other, less-recommended interventions)[2], emphasizing the importance of taking actions in this area.
In other words, there seems to be a lot of potential value in understanding this problem better.
When dealing with FLW, there is also a food security[3] dimension to the discussion that is worth mentioning. According to the WHO, as of 2026, 2.1 billion people experienced moderate or severe food insecurity.
Some argue (see references section) that reducing FLW should increase food security, although it does not mean that reducing food waste automatically translates into greater food security—the causal chain is more complex.
There are at least two broad ways of responding to food waste.
One is food rescue: recover food that would otherwise be discarded, and redirect it to another use—for example, through food banks or redistribution organizations.
The other is prevention: understand why food becomes waste in the first place and change the processes, incentives, technologies, or behaviours that cause it.
While food rescue only reduces damage caused by the waste itself, prevention includes the impact food rescue generates plus a likely increase in food security, a decrease in pollution, a decrease in net-negative impact land use (allowing for more effective uses of this land), and more (see references section for more information).
Moreover, these approaches are at least partially mutually exclusive (see references section). Hence, if we wish to increase FLW prevention, a particularly difficult research question arises:
What interventions actually prevent food from becoming waste or lost, at what cost, and in which contexts?
Answering that question requires much more than knowing the total amount of FLW. To answer this question, it will be helpful to know, for example:
These kinds of questions move us from “the scale of food waste is large” to “we should implement intervention X in context Y because the evidence suggests it will achieve Z.”
The difficulty is that answering many of these questions requires collecting primary data.
UNEP's methodology explicitly recognizes this problem, as demonstrated in Table 27: Comparison of measurement methodologies in the retail sector, and in Annex 2: Table of datapoints. On the latter, one can observe the relatively small number of datapoints per survey and short surveying duration, resulting from the difficulty and cost of the measurement operation.
This is mainly because waste compositional analysis (WCA) is considered to be the most accurate method for surveying FLW, but on the downside, it’s the most labour-intensive and costly method (see UNEP report as well as other examples, all in the references section).
In brief, in the existing research, WCA is done by examining, weighing and classifying each garbage item manually, for each garbage bag. To the best of my knowledge (and if I’m wrong, I’d be happy to be corrected), and as of the time of writing, there is no available method that produces the same (or better) accuracy with lower costs.
This is not simply an inconvenience for individual researchers. It creates a potential bottleneck for the field as a whole. If answering a research question requires recruiting participants, arranging waste collection, transporting samples, manually sorting them, weighing and classifying the material, and then repeating the process sufficiently often to obtain representative data, a relatively small research question can turn into a substantial field operation.
Another way to deliver WCA data is by training the staff of the business in question. This must mean that the staff have less time to do their actual job, hence requiring extreme cooperation from the business owner, and all the research papers that used this method, that I know of, did so for a limited period (and I believe that maybe some compensation was also given by the researchers).
That creates a trade-off between quality, resolution and number of datapoints. One can either have many low-quality datapoints or a few high-quality datapoints. Or a few high-resolution datapoints versus as many as needed low-resolution datapoints.
This leads to a situation where some potentially useful questions remain under-researched—not because they are unimportant, but because the marginal cost of obtaining the necessary evidence is extremely high.
This is what I believe, based on these examples, my knowledge and experience in this field, and conversations I have had with both academia and industry experts.
Daphi is developing a service that would make FLW data easier for researchers and others to access and use. Rather than every research team independently paying for the same kinds of fieldwork, the system will allow to:
We do not yet know which needs (whether discussed here or not) are most pressing for those working on FLW, or whether there is sufficient demand for such a service at all.
That is why we are starting with a call for expressions of research interest and data needs.
We would particularly like to hear from
Research proposals are not required. We are only interested in understanding your data needs (and whether our data could help), recurring problems in gathering such data, and research questions for which these data could be valuable. And, if such a service could help, it would be very helpful for us to hear about any requirements that you might have for the system.
Daphi is a not-for-profit organization. However, the data-access service we are exploring is expected to be paid for by most users[4]. The reason is that collecting, cleaning, maintaining, and providing high-quality data has real costs. A paid model will allow the service to become sustainable, while keeping the prices well below today’s data collection costs.
We are therefore interested in hearing both from people who might become users and from people who might want to collaborate on generating, improving, or distributing data. Also, we welcome researchers or other individuals and organizations who already have good-quality datasets to share them through the platform[5].
If you are in research, industry, government, or consulting, or have any other connection to the field, and have a FLW research question that more or better data could help answer—or you work with people who do—we would be very interested in hearing from you.
Regarding this post, the opportunity mentioned, or any other idea, thought, connection, collaboration, or donation, we would love to hear from you: [email protected].
All the facts stated are drawn from these sources, or from sources mentioned by them. To allow better navigation in the in-depth sources, I highly recommend reading the sources listed under General Information first.
WHO: The State of Food Security and Nutrition in the World 2026. https://www.who.int/publications/m/item/the-state-of-food-security-and-nutrition-in-the-world-2026 ; https://doi.org/10.4060/cd8306en
Sources suggesting a plausible connection between FLW and food (in)security:
Besides the information given in the general information sources, see then methodology sections in the following paper as examples of the discussed problem:
[1] According to Project Draw Down, this estimate, which was correct for 2022, is still considered valid for recent years as well. See Sources section for links and more reading materials.
[2] Sorted by CO2-eq.
[3] Food security: sufficient, nutritious, and safe nourishment that is physically and economically accessible at all times. Other definitions exist, but they differ in nuances, not essence.
[4] I don’t want price to be an obstacle for advancing the field. Those who may lack the financial means are kindly requested to email us.
[5] Details will be discussed via email.
