glebora Talk to us
The method
Who it’s for
For coffee importers For cocoa importers Are you an operator? Pricing
Resources
Guides FAQ Glossary About Talk to us

Home › The method

The method

From plot coordinates to a filed statement you can defend.

You import the coffee or cocoa; the filing duty is yours. We carry the whole due diligence behind it (the information, the deforestation assessment, the risk, and the filing), so a legal obligation becomes a short, repeatable routine.

The full obligation

Due diligence is more than a satellite check. We carry all of it.

EUDR due diligence has three parts the regulation is explicit about: collect the information on your supply chain, assess the risk it carries, and act on that risk before you file. Most tools stop at a map. We run the whole process end to end, and the filed statement is the last step, not the only one.

01

We build your due-diligence file. Art 9

The products and quantities, the country and plots of production, the production period, your suppliers, your import and EORI details. You bring the source information (the geolocation your producers hold and the documents only your suppliers can provide), and we assemble it into the record the regulation expects: structured, complete, and ready to defend.

02

We assess every plot for deforestation. the hard part

Each plot is cross-read against three authoritative datasets to help distinguish genuine post-2020 forest conversion from the ordinary management of shade-grown coffee and cocoa: the distinction a single map cannot make on its own, and the part we built glebora around. Our assessment is type-aware, and anything ambiguous goes to a person, not an automatic verdict. Every result records the exact datasets and versions it relied on.

03

We structure and assess your legal-production evidence. Art 9 · legal production

The regulation also asks whether your coffee or cocoa was produced legally: land-use rights, environmental rules, third-party and labour rights in the country of origin. You supply the documents your suppliers hold; we organise them into your file, assess what they show, and flag what’s missing or weak. You’re never left to work out what “evidence of legal production” means on your own.

04

We weigh the risk, and tell you what to close. Art 10-11

The whole picture becomes a documented risk assessment. Where the risk isn’t negligible, we tell you precisely what needs closing and how, because a statement filed over an open risk is a statement that won’t hold. We file when it’s ready to stand up.

05

We file as your Authorised Representative, and hand you the record.

We submit the DDS in the EUDR Information System (the EU’s EUDR filing system within TRACES) under your operator identity, retrieve the reference number your buyers and the authorities can ask for, and hand you the complete, cross-referenced dossier behind it: the answer, already written, for the day a Competent Authority asks how you concluded no or only negligible risk was found. We file with your operator identity in the payload, never our own.

You bring your supply chain. We carry the due diligence, not a dashboard that leaves the hard parts on your desk.

The assessment

Why one map isn’t enough for coffee and cocoa.

Coffee and cocoa are often grown under shade trees. To a single satellite map, the pruning and replanting of those shade trees can look like forest loss, and a naïve one-map overlay will flag a perfectly legal agroforestry farm as a problem.

We cross-read several authoritative sources instead of trusting one. A baseline forest map sets the 2020 line; a change layer shows disturbance since; and a third classifies whether that disturbance was actual deforestation or the ordinary management of a tree-crop system. Confirmed post-2020 conversion is flagged with the evidence behind it; anything ambiguous is surfaced for a human to look at, never silently passed or silently failed.

Copernicus GFC2020 Hansen Global Forest Change JRC Tropical Moist Forest

How we read the evidence

01 · baseline2020 forest baselineCopernicus GFC2020
02 · changePost-2020 changeHansen GFC
03 · classifyDisturbance classifiedJRC TMF
Type-aware cross-read: no single map trusted No one map is treated as binding. The three layers are read together to tell genuine post-2020 conversion apart from the ordinary management of a shade-grown, tree-crop system.
No conversion indicatedNothing in the sources points to post-2020 forest conversion on the plot.
Needs review · humanAn ambiguous signal: canopy change that could be shade management. Surfaced to a person, never auto-failed.
FlagConfirmed post-2020 conversion of forest to agriculture, raised clearly with the evidence behind it.

Real forest data

We read the real forest behind your beans.

Two-panel Hansen Global Forest Change map of the Ahafo cocoa frontier in Ghana: forest canopy in green and tree-cover loss since 2000 in red, with a regional view beside a close-up of the assessed cocoa plot outlined in white. Data: Hansen/UMD/Google/USGS/NASA, rendered on Google Earth Engine.
Hansen Global Forest Change v1.13 across the Ahafo cocoa frontier in Ghana, rendered on Google Earth Engine: forest canopy in green, tree-cover loss since 2000 in red. The plot outlined at right is the CERSGIS reference farm assessed in the record below: ~0.72 ha of shade-grown cocoa with genuine post-2020 loss on its edge. We isolate the post-2020 portion of that loss and cross-check it against JRC TMF, which tells real forest conversion apart from ordinary agroforestry management: the difference between a flag and a plot held for review.

What you receive

A record you can hand to an auditor, not a report to decode.

Every plot comes back as a plain record: what was checked, what the sources said, and a clear verdict, with anything ambiguous held for a person rather than force-passed or force-failed. Below is real, unedited output from our engine: two public reference plots, run and rendered field for field.

glebora Plot deforestation assessment
Baseline
31 Dec 2020
Analysis
plot boundary, post-2020
Datasets
GFC2020 · Hansen v1.13 · TMF 2025

Plots assessed: 2

Ghana · Ahafo, shade-grown cocoa Needs review

Location 6.738522, −2.561955| Plot size 0.72 ha| Forest present in 2020 none

FindingMethod / datasetResultReading
Tree-cover loss, post-2020 Hansen GFC 2025 v1.13 0.35 ha Disturbance signal, not itself deforestation
Confirmed deforestation JRC TMF vs 2020 baseline 0.00 ha No TMF-confirmed forest conversion

Held for human review. Satellite shows 0.35 ha of tree-cover loss since the cut-off, but cross-checking with JRC TMF confirms none of it is forest conversion, consistent with the ordinary management of a shade-grown farm. Surfaced to a person, not auto-failed.

Brazil · Sul de Minas, coffee Pass

Location −21.715207, −46.438349| Plot size 4.59 ha| Forest present in 2020 none

FindingMethod / datasetResultReading
Tree-cover loss, post-2020 Hansen GFC 2025 v1.13 0.00 ha None detected
Confirmed deforestation JRC TMF vs 2020 baseline 0.00 ha No forest conversion

Pass. No post-2020 tree-cover loss on any source. Clears without needing human review.

Datasets & method

Baseline map
JRC Global Forest Cover 2020, Copernicus JRC/GFC2020/V3
Loss detection
Hansen Global Forest Change 2025, CC-BY 4.0 v1.13
Conversion class.
JRC Tropical Moist Forest, 2025 release JRC/TMF/v1_2025
NRT alerts
RADD / WUR Sentinel-1 forest disturbance, CC-BY 4.0
Criteria
Reg. (EU) 2023/1115 Art. 2: forest ≥0.5 ha / ≥5 m / ≥10% canopy; cut-off 31 Dec 2020
Method
Type-aware cross-validation: TMF-confirmed conversion flags; unclassified loss held for review

Filing

Once a plot clears, we file the Due Diligence Statement in the EUDR system as your Authorised Representative, under your operator identity and EORI. TRACES issues the DDS reference and verification number on submission; you keep the whole dossier (plot geometry, assessment, evidence and reference) for the five years the regulation requires.

Filed: Due Diligence Statement Acceptance · test submission
DDS reference
26DK0KD5X095AP
Verification no.
22NXW0L5
Status
AVAILABLE
Filed as
Authorised Representative: operator identity in the payload, never ours
Operator
Provenleaf Coffee ApS · EORI DK39184756
Submitted
31 Jul 2026

The coffee plot above cleared, so we ran it end to end through the European Commission’s EUDR Acceptance environment: the official replication system that mirrors production TRACES but issues test statements, not legally valid ones. This is a genuine sandbox submission that proves the filing path works, not a client filing. Live filings are made only on production TRACES, under a signed mandate.

glebora · deforestation due-diligence assessment Engine: GFC2020 / Hansen v1.13 / TMF 2025 / RADD

Real, unedited output from our deforestation engine, run on two public reference plots. The Ghana plot is a public reference farm, not a client plot: CERSGIS, Reference Dataset for Land Use Change Mapping in Ghana’s Cocoa Landscape (2024-2025), Zenodo doi:10.5281/zenodo.15778396 (CC-BY 4.0). The DDS reference shown was issued by the EU’s EUDR Acceptance (replication) environment: a real test submission, not a live client filing. The due-diligence declaration, and legal liability for it, remain with the operator.

Keep this as a standalone due-diligence file you can print or share →

Get in touch

Ready when you are.

Tell us what you import and where from, and we’ll show you exactly what your filing would involve.

contact@glebora.com