The risk assessment is the reasoning step of EUDR due diligence: Article 9 collects the information, Article 10 obliges you to assess it, and you may only place the product when that assessment concludes no or negligible risk. Most guidance stops at quoting the article. This guide turns it into a procedure: what you evaluate, what satellite screening contributes, when mitigation kicks in, and what the documented output looks like.
What Article 10 asks you to weigh
The regulation lists assessment criteria; grouped functionally they are:
- Origin risk: the benchmark tier of the production country, presence of forests and of deforestation or degradation in the area, and prevalence of production-driven clearing.
- Evidence risk: reliability and verifiability of the data you hold, concerns such as corruption, document fraud or weak enforcement in the origin, and any substantiated concerns raised about your supply.
- Chain risk: complexity of the supply chain, how many hands the product passed through, and the risk of mixing with material of unknown or non-compliant origin.
- Rights risk: presence of indigenous peoples and claims to the land, and consultation where relevant, feeding the legality limb.
The structure to notice: country tier is one input among many. A plot can fail in a low-risk country and pass in a standard-risk one; the tier sets procedure and inspection odds, the plot evidence decides compliance.
What screening contributes, concretely
For the deforestation-free limb, the operative question per plot is empirical: is there evidence of forest conversion after 31 December 2020 inside this geometry? Modern screening answers it by convergence of evidence, the approach of FAO's open WHISP methodology: multiple independent datasets (tree-cover baselines, annual loss layers, near-real-time alerts, commodity and plantation maps) are read together per plot. Agreement across layers produces a confident verdict; disagreement produces "needs review", not a coin flip. That matters because single-layer screening fails in predictable ways: shade coffee and mature oil palm read as forest, replanting cycles read as loss.
This is the layer plotvera generates for you: per-plot indicator readouts, a verdict per commodity question, and the dataset names and run dates attached, so the screening run itself is citable evidence in the file. A verdict is an input to your assessment; the conclusion, weighing screening against the other criteria, remains yours.
The decision rule
- All evidence converges on clear: record the conclusion of negligible risk and proceed to filing.
- Something flags: you may not place the product on that evidence. Move to Article 11 mitigation: obtain more or better data (a boundary polygon instead of a point, planting records, prior imagery), commission independent verification, engage the supplier, or exclude the plot or supplier from the consignment.
- Mitigation cannot get you to negligible: the product does not go on the EU market. That outcome, documented, is the system working.
Special cases
- Low-risk sourcing: Article 13 exempts you from the formal assessment and mitigation steps when all plots sit in low-risk countries and you have assessed mixing and circumvention risk; any contrary information revokes it. Details in the simplified due diligence guide.
- Referenced statements: where upstream DDS numbers cover part of your product, your assessment covers the remainder plus the plausibility of what you rely on; large companies must ascertain upstream due diligence actually happened.
What the documented output looks like
An auditable assessment for one consignment is short: the consignment and plot file identifiers; the country tier at filing date; the screening run reference with verdict summary; notes on chain complexity and mixing; flags raised and how each was resolved or excluded; the named person's conclusion and date. Ten lines that take minutes when the data is organised, and that your five-year file preserves. Article 12 adds a maintenance duty: review the due diligence system itself at least annually and keep that review on record too.