How to use the FLW Database

The database on Food Losses and Waste is a novel endeavour by FAO under Sustainable Development Goal (SDG) 12.3.1. to share available knowledge and help focus efforts where they can have the most impact in order to meet the SDG target. This is a living dataset, based on an extensive and continuing literature review, so always check back for additions and improvements. We know it will be a collaborative effort to improve the data – so if you see an area where you have data and would like it included please send it to us!

The database is open for users to interact with the data. Want to know where tomatoes are lost? How about in your region? Here you can find all the options for refining your queries to the dataset.

The database is made of two main parts:

  • A selection pane on the left:
    The selection pane allows you to refine the search on a number of filters
  • The data visualisations at the centre of the page where you can chose how to visualise the data of the following options:

    Graph of Loss %: showing all the selected data by year 

    HeatMap of Available Data: showing where and for what commodities there is information, thus providing an overview of available knowledge

    Boxplot by Stage: providing a snapshot of the level and variety of losses by showing the range of losses by stage of the value chain

    Data: a table comparing the levels of losses and providing the actual percentages with a set of metadata information

Data visualisations

1. Graph of Loss %

It has all the selected data points according to the filters. When opening the database and before making any selection, the graph shows the content of the whole database.

Each point represents a specific research/study. Loss percentages are plotted by year. A color code indicates the different Methods of Data Collection.

When pointing the cursor on a point, a metadata box appears reporting the related year, actual loss percentage, commodity, country, method of data collection, data source, reference and link to the original source.



2. HeatMap of Available Data

It summarises the extent of available data by geographic area (region or country depending on the filters) and items (commodity groups or individual commodities depending on the filters) for the whole food chain.

The color changes with the amount of available information. The graph therefore illustrates data scarcity or abundance.



3. Boxplot by Stage

It is a display of variation in a data set, that can be used to identify the stage at which losses occur the most and therefore target intervention strategies.

In this database, it compares the levels of food loss at the different stages in the value chain. It has a box to show where 50% of the data falls into, the mid-value of the data shown by a line and extreme points out of the box for the percentage loss at every stage in the supply chain. On other words, the data is divided into quarters where the whiskers indicate variability outside the third quarter of the data and the first quarter.

The end of the whiskers represents the minimum and maximum values of the distribution (potential outliers). An outlier is an observation point that is distant from other observations, and thus falls outside of the overall trend that is presented in the data.

In box plots, an outlier is a number, which is greater than the end of the third quarter by more than 1.5 times the distance between the first and third quarter of the data.

Refining the dataset – Filtering Options

There are several options available to the users in order to refine the information and zoom into the results that are key for the user’s needs. The filters can be used independently of each other in most cases, so that a user could look for losses across all commodities for a geographic aggregation or vice versa – specific commodities or commodity groups across all geographies.

Year Range:

Select the period of interest. This is the year that the study was undertaken.  Years range from 1945 to Present.


Select the broad geographic area of interest. The geographic aggregates follow the M49 geographic standard of the UN Statistics Division. Several alternative geographical aggregates such as countries in which FAO has operational capacity or the 2018 World Bank Income Classifications are also provided.

Aggregation Options:

Select the sub-groups within the Aggregation heading previously filtered.


Individual Countries can be selected within the sub-groups in the Aggregation Options. The list of countries follows the M49 standard.

For example, under the ‘SDG Region’ (Aggregation), and the ‘Latin America and the Caribbean’  (MDG=M49) (Aggregation Options), only the countries in this region can be selected with the Country filter (Aruba, Anguilla, Argentina, etc.). If the data exists for the selected country(ies), the data points will be filtered and presented in the graphs.

Basket Items

The data can be filtered for six commodity groups. The options in this filter are:

  • Cereals & Pulses
  • Fruits & Vegetables
  • Roots, Tubers & Oil crops
  • Animal Products
  • Fish Products (to be added)
  • Other (stimulants, spices, etc.)

These are the same commodity groups used to compile the SDG indicator 12.3.1 (a) Food Loss Index and that countries will use to select their top ten commodities. The selection in this category restricts the choice of food commodities in the Commodity (CPC 2.1) filter, much like the geographic aggregation.

Commodity (CPC 2.1)

Individual commodities can be selected with this filter within the sub-group Basket Items. The list of items follows the international Central Product Classification version 2.1 (CPC 2.1) used in FAOSTAT. These are the international standard for definitions, concepts and classifications for food and commodities.

Value Chain Stage(s)

In an effort to improve the knowledge on where losses and waste is occurring within the commodity supply chains, the data points were tagged with the stage of the value chain that was delineated in the source study. They follow the main stages considered within the SDG (point of production (farm, slaughter, landing); storage; transport; wholesale; processing); however, for some countries and commodities additional stages or activities have been included as they are critical loss points that were the object of the study.

Method of Data Collection

The options for this filter summarize the main loss measurement/estimation methods found in the literature or point at the main sources of the data. Knowing the method used in each study helps understanding its data scope or representativeness.

Options on the methods are:

  • The Expert Opinion method consists in the advice, belief or judgment given by an expert on the losses.
  • The Rapid Assessment data collection method can be described as assessing a product via a visual damage scale and then the estimation of sample weight loss by using a simple equation. In the database, rapid assessments include collecting information from a variety of stakeholders and sources, such as semi-structured interviews, reviews of existing estimates and development of the broad structure and flow of a commodity within a country. They can focus on a specific group or area of the supply chain and identify critical loss points. No direct measurement or systematic surveying across the wider population is done. 
  • Case Studies examine food losses of specific persons or groups over a period. Case studies are often limited to a small set of participants, which may or may not be representative of the overall population. 
  • Crop Cutting includes direct measurement of a pre-sampled area of production and follows the local harvesting practices. This is a cost-intensive measurement that can benchmark interview-based surveys.
  • Field Trials estimate losses via random assignment after field studies. They are experiments undertaken at an extension farm or semi-controlled for aspects related to the outcomes of an experiment, but non-laboratory based. They often produce loss factors for farming/harvest/catch/slaughter practices.
  • Laboratory Trials evaluate losses via random assignment after trials conducted not in the field. Laboratory trials are done in a fully controlled environment, often with small sample sizes and likely do not mimic the conditions that would occur under normal circumstances of practice. 
  • Modelled estimates refer to estimation of losses via statistical models.
  • Survey/Declarative are estimates where a defined survey and sample have collected loss data.

Options on the sources, whose data are compiled with the same method, are:

  • APHLIS estimates refer to those calculated in the African Postharvest Losses Information System (APHLIS), which is a combination of modeling and other data collection efforts for Sub-Saharan Africa in Grains. 
  • FAO sources report estimates from other sources outside the annual agricultural production questionnaires such as country reports and publications.
  • Food Balance Sheets/Ag. Production Questionnaires data points come from official sources within country governments (e.g. National Statistics Offices, Ministries of Agriculture, etc.). These data points are collected with production and other utilizations of the commodities on an annual basis.
  • National Statistics Yearbook loss data come from official published sources within country but may not have been reported in the annual agricultural production questionnaires.
  • National Accounts loss data or estimates come from official sources and systems within country and have been used to compile the national accounts but may not have been reported in the annual agricultural production questionnaires.
  • The World Resources Institute FLW Protocol provides loss estimates, however the protocol does not prescribe a method of data collection. The different reporting entities may therefore collect or estimate data using a variety of methods. The entities measuring losses and reporting through the Protocol may have comparable results over time.

On/Off Options

Only Top 10 SDG Baskets

The focus of the SDG 12.3.1.a is to improve the data in the top 10 commodities selected for the country to track within the 2030 agenda. Clicking on this option will filter the data to the default top ten commodities by country.

However, FAO is collecting data where it is available and countries may report above and beyond the top 10 commodities and studies, so all data in the database is available.

Aggregate to Baskets (Heat Map)

In order to aggregate the commodity level data to reflect the heat maps in the State of Food and Agriculture. If this box is not checked then, the crops and countries can be filtered for a localized dataset on where there may be any number of data points across the time selected. 

Back to the Database

Share this page