Environmental impact

What does it cost the planet to put this food on your plate? We measure three things you can hold in your head: the climate cost in carbon dioxide, the land it takes to grow, and the water it drinks up. The numbers come from published life-cycle research, with honest ranges so you can see how much wiggle room each estimate has.

This measures the cost of producing the food. What happens after it leaves the farm, like trucking, refrigeration, cooking, and packaging waste, is not yet in the number. The ranges shown are how much real producers actually vary from one farm or fishery to the next.

Three indicators2,425 food entries in the LCA catalogueHonest uncertainty ranges

Four ways to score

One food, two foods side by side, a full day of eating, or every metric at once.

Score a food or meal

Pick foods from our catalogue, set serving sizes, and read the climate, land, and water cost. Switch between per serving, per calorie, per 100 g, and per gram of protein to see the picture from different angles.

Compare two foods

Stack two foods side by side. The right tool for “beef or lentils?” and “rice or quinoa?”. The uncertainty ranges keep you from reading too much into a single point estimate.

Score a full day

Walk through your day one meal at a time. The wizard adds up the footprint across breakfast, lunch, dinner, and whatever else happened, so you see the day as one number.

See it next to the rest

All scores lines up environmental impact next to healthy eating, health impact, Food Compass, star ratings, and eating style. One food list, six different questions, all in one place.

Three things we measure

Life-cycle research can measure many things, but only some have strong enough per-food evidence to publish honestly. We start with the three you probably already think about: climate, land, and water.

Climate

kg of CO₂ equivalent

The greenhouse gases released to grow, raise, and harvest the food, added up as if they were all carbon dioxide. Beef is around 10 kg per 100 g you eat. Lentils are around 0.4 kg.

Land

square metres for one year

How much farmland it takes to produce the food, weighted by how long the land is held. Beef sits near 9 square-metre-years per 100 g. A head of lettuce is closer to half a square metre.

Water

cubic metres of scarce water

The freshwater drawn from rivers and aquifers to grow the food, weighted by how scarce that water is locally. Almonds in California carry a much higher water cost than the same almonds in Spain.

Not yet shown. Air quality, water pollution, ozone-layer damage, fossil-fuel and mineral scarcity, and chemical toxicity all matter, and life-cycle science has methods for them. We do not show numbers for those yet because the per-food data we trust is not there. Pesticide residues in your food, regenerative-farming credit, and what happens after your meal hits the bin are different questions that need different tools.

How the number is built

Five things happen behind the scenes when you score a food.

Pick the right reference number

Every food has a published climate, land, and water figure based on averages from many real farms and producers. When the matching is uncertain, we fall back to the average for the food group, which is a less precise but still defensible number.

Show the range, not just a single number

Real farms vary a lot. A litre of milk from one dairy can carry many times the climate cost of a litre from another. We show a low, a central, and a high estimate so you see how much that variability matters. These are envelopes from published farm-level data, not statistical confidence intervals.

Compare on a fair basis

You can ask the same question four ways: per serving, per 100 calories, per 100 grams, or per gram of protein. Per-calorie is the default because it stops a cucumber from looking artificially cheap next to a bowl of pasta.

Make water local

Climate is the same wherever the gases come from, but water scarcity is not. You can score a food with global average water-scarcity weights or pick a country so the water number reflects where the food is actually grown.

Match smart, never invent

When you score a food from the Canadian catalogue, we use a language model to choose the best life-cycle entry from a shortlist of real candidates, not to make one up. If the model is not confident enough, we fall back to the group average and flag the match openly.

Break composite dishes apart

Pizza, stew, and casserole are not a single ingredient. When you score one, we break it into the ingredients that make it up, score each, and combine them by mass. The breakdown is shown so you can see what is driving the total.

Three ways to read every result

The numbers do not change. The explanation does. Pick the view that fits why you are looking.

Everyday

Climate, land, and water in plain language, with a quick read on whether this food sits low, medium, or high for its group. No formulas, no acronyms.

Researcher

Full life-cycle method, how the food was matched, the data quality rating behind the matched entry, and a parallel set of values from a second method as a cross-check.

Policy

Population-level framing for procurement, taxation, and labelling decisions. Includes an optional dollar value of the climate impact using the Government of Canada's published social cost of carbon.

What this is not

Not the whole life of the food. Only the production stage. What happens after, like driving, refrigeration, cooking, and disposal, is left out for now.
Not a statistical margin of error. The low and high bounds reflect how much real producers vary, drawn from published farm-level meta-analyses. They are honest envelopes, not confidence intervals.
Not a toxicity score. The life-cycle methods for toxicity are still considered provisional in the underlying research, so we keep them out rather than show a number we cannot defend.
Not a pesticide-residue check. What you might be eating from your food is a different question from what was released growing it. We do not answer the residue question yet.
Not regenerative agriculture. The data behind the numbers comes from conventional farming. If you are eating something grown regeneratively, the climate and land numbers may be too high for your specific case.
Group averages when matching is uncertain. When the tool cannot pin down a specific life-cycle entry for your food, it falls back to the average for its group. Within a group, real foods can vary widely. Skim milk and aged cheddar belong to the same group but their climate footprints differ by roughly ten times.

Where the foods come from

Every food you score is drawn from the same catalogue used by every other tool here. That is 5,691 foods from Canada's national food file, plus 1,028 West African staples from the FAO regional table. The matcher resolves either source against the life-cycle catalogue the same way.

Other regional food tables can plug in later through the same setup.

Where the life-cycle numbers come from

Climate and land numbers come from a peer-reviewed meta-analysis of thousands of real farms (Poore & Nemecek, Science, 2018), combined with France's national life-cycle catalogue AGRIBALYSE, which holds 2,425 commodity-level entries. Water uses a separate published source that tracks scarcity by region.

Each catalogue entry carries a data-quality rating, and most entries meet a good-enough bar. A handful of entries are known to have published errors in the source data, and we flag those clearly when they appear in your result.

Scan a packaged product

Take a photo of the Nutrition Facts panel and ingredient list of a packaged product. The app reads the label, suggests what is in it, you confirm, and the result feeds the environmental scorer along with the other five lenses.

See it next to the rest

Environmental impact is one of six measures on all scores. Sustainability decisions rarely live alone. Diet quality, healthy-life minutes, product-level ratings, and a Food Guide read travel with the environmental view on the same panel.

Where the science comes from

Method

The life-cycle assessment method is ReCiPe 2016, developed by Huijbregts and colleagues for the Dutch national institute RIVM and published in the International Journal of Life Cycle Assessment in 2017.
The per-food climate and land numbers come from Poore & Nemecek's 2018 meta-analysis in Science, the largest of its kind.

Data and matching

AGRIBALYSE is the French government's national life-cycle catalogue, maintained by ADEME and updated to version 3.2 in 2024.
The matching layer that connects food databases to life-cycle entries is informed by recent work on retrieval-then-rank methods by Zhou and colleagues (2025) and earlier interlinking research by Furrer and colleagues (2024).

Ready to see the footprint?

A single food, a homemade meal, or a whole day of eating. Same pipeline, same honesty about what we know and what we do not.