living experiment
The supermarket,
as a graph
355,192 products from the Spanish supermarket, the ones whose labels are declared on OpenFoodFacts, turned into a single graph: brands, categories, additives and allergens, connected by appearing together on the same label. Touch anything and watch what moves with it. Rename the nodes and the graph stays the same: parts that break down at the same time, symptoms that show up together in a medical history, accounting entries that always move hand in hand.
Each additive carries its EFSA regulatory status under a strict rule: grey by default, green only with an explicit evaluation and no flagged risk, red if it is banned in the EU. This is not nutrition or safety advice; it is what the labels and the rulings say, shown as is.
The graph, in figures
The aggregated map you navigate above has 3,296 nodes (2,177 categories, 596 additives, 500 brands and 23 allergens) joined by 9,515 connections, summarising the 355,192 products. These are the most asked-about figures, taken from the same files that feed the dashboard.
The most used additives in the Spanish supermarket
| Additive | Products declaring it |
|---|---|
| E330 - Citric acid | 17,142 |
| E322 - Lecithins | 8,829 |
| E202 - Potassium sorbate | 7,556 |
| E300 - Ascorbic acid | 6,099 |
| E471 - Mono- and diglycerides of fatty acids | 5,512 |
| E500 - Sodium carbonates | 5,152 |
| E322i - Lecithin | 4,989 |
| E407 - Carrageenan | 4,644 |
| E415 - Xanthan gum | 4,517 |
| E412 - Guar gum | 4,373 |
The brands with the most distinct additives in the catalogue
Only brands with at least 20 products enter, a declared threshold that avoids rewarding tiny samples. Many distinct additives is not a grade: a brand with 10,000 products touches more additives than one with 30.
| Brand | Products | Distinct additives | Additive-free |
|---|---|---|---|
| hacendado | 10,741 | 250 | 76.3% |
| carrefour | 8,356 | 223 | 59.6% |
| dia | 3,228 | 197 | 78.4% |
| consum | 1,939 | 159 | 74.6% |
| auchan | 2,746 | 155 | 86.1% |
| lidl | 2,221 | 150 | 83.7% |
| nestle | 1,885 | 147 | 76.6% |
| alteza | 1,910 | 136 | 87.5% |
The most declared allergens
milk (14,038 products) · gluten (11,946 products) · soybeans (8,227 products) · nuts (4,381 products) · eggs (4,296 products) · fish (2,107 products) · sulphur dioxide and sulphites (2,036 products) · sesame seeds (1,456 products).
Categories where 100% of products declare zero additives
Olive oils (3,592 products) · Extra-virgin olive oils (2,874 products) · Virgin olive oils (2,529 products) · Tunas in oil (1,058 products) · Fresh eggs (1,033 products).
All of this is what the labels declare on OpenFoodFacts, a collaborative database: a badly read label can inflate a count, and counting additives is not a safety or nutrition judgement. The EFSA rulings for each additive are linked in the dashboard, additive by additive.
how we do it
- Aggregated view: what you navigate is the map of connections, not the 355,000 products one by one.
- Declared data: OpenFoodFacts is collaborative; a badly read label can inflate a count. The caveats of each figure are written next to the figure.
- Sources: OpenFoodFacts contributors (ODbL) · Wikipedia entries with their attribution · EFSA rulings linked on every additive.
- Exploratory, no backtest: this demo predicts nothing and is not examined against the future; it is a map to look at and touch. The demos that promise prediction carry their backtest; here the value is the exploration.
Can you use this? Yes. The text and graphics of this page are published under a CC BY 4.0 licence: use them, share them or analyse them freely, citing the source, Team Banzai (team-banzai.com). The graph and the counts derive from OpenFoodFacts and remain under the ODbL licence, which requires attribution and sharing derived databases under the same licence; EFSA rulings keep the terms of their original sources.
And in your company?
The same graph works for seeing which spare parts are always ordered together in a warehouse, for understanding which procedures in a public administration share steps and could be unified, or for grouping customers that resemble each other without you deciding beforehand what they have in common.
If any of these experiments reminds you of a problem of yours, a product range that grows by inertia, spare parts always ordered together with nobody taking advantage of it, aisles laid out by habit rather than by what people buy, get in touch and we'll talk it over. No fuss: we'll tell you whether it can be done or not.