Methodology
How we compute climate contribution.
FRAME-NL · v1.0 · published 2026-04-26 · CFNL = 0.93
Kavel aligns with FRAME — the Food Recovery Avoided Methane Emissions methodology used by Global FoodBanking Network — and applies a Dutch-specific counterfactual based on RIVM Afvalmonitor 2024 + CBS Waste Statistics. Every number on this site can be traced back to the field it was extracted from, and the source it was extracted from.
Ingestion pipeline
From an annual report PDF to a contribution disclosure.
- 01
Annual report
Foodbank submits its annual report PDF (Dutch).
- 02
AI extraction
Five parallel section-specific prompts extract typed measurements with provenance.
- 03
Provenance ledger
Every field is tagged: extracted, inferred (national avg / category split / calculation), or manual.
- 04
FRAME compute
kg × emission factor × NL counterfactual (0.93). Per-category and total CO₂e baseline persisted.
- 05
Contribution disclosure
On corporate purchase, allocation engine ranks food banks; AI composes ESRS E5 + S3 climate-contribution disclosure, streamed back via SSE.
Core formula
The arithmetic, in one line.
CO₂e = Σ(kgi × EFi) × CFNL
where CFNL = 0.93 (NL incineration with energy recovery)Each food category contributes its mass times its emission factor (kg CO₂e per kg food). The sum is multiplied by the Dutch counterfactual — the emissions that would have occurred if the food were sent to the dominant Dutch waste route instead of being rescued.
Emission factors
Per category, with sources.
| Category | EF (kg CO₂e / kg) | Visual | Source |
|---|---|---|---|
| Produce | 1.0 | FAO FWF (2013) Table 4.2 | |
| Bakery | 1.5 | WRAP Courtauld Commitment 2030 | |
| Dry goods | 2.0 | FAO FWF + Poore & Nemecek (2018) | |
| Prepared | 3.0 | Poore & Nemecek (2018) | |
| Dairy | 3.2 | FAO FWF + RIVM Dutch dairy LCA | |
| Eggs | 4.5 | FAO FWF + Dutch egg sector | |
| Meat | 8.5 | FAO FWF (2013) weighted NL consumption mix |
NL counterfactual
0.93 — not 1.0. Why it matters.
Most international food rescue calculators apply a counterfactual of 1.0, assuming food would otherwise rot in landfill — generating significant methane. The Netherlands incinerates virtually all unrecycled waste with energy recovery, which produces less methane than landfill.
We apply CFNL = 0.93 per RIVM Afvalmonitor 2024 and CBS Waste Statistics. This makes Kavel conservative compared to FRAME defaults — by design.
Why the math compounds
The Netherlands wastes 70× more food than its foodbanks rescue.
Foodbanks recover 1.43% of NL food waste. Kavel funds the network to rescue more — paying foodbanks for verified contribution capacity, instead of asking corporates to pay landfill avoidance fees that never reach the social side.
Source: Stichting Samen tegen Voedselverspilling, cited in VBN F&C 2023
See the full data narrative on /why — poverty gap, decade of pressure, CSRD wave, contribution-capacity gap.
Provenance ledger
Every measurement carries its source.
Read directly from the annual report. Pinpointed to a page or table cell in the source document.
Filled from a Dutch national average when missing — e.g. CBS household size for individuals served.
Distributed across categories using a national typical mix when only a total is reported.
Derived from other extracted fields (e.g. residual = total − sum of categories).
Manually corrected after audit — used sparingly, always with a reviewer note.
Claim type
Climate contribution. Not offsetting.
Kavel packages fund verified climate impact at Dutch foodbanks. Corporates disclose this as a climate contribution under ESRS E5 (resource use & circular economy) and ESRS S3 (affected communities) — not as Scope 1/2/3 reduction or offset under ESRS E1.
Avoided emissions are reported separately per EFRAG E1-4 §AR-58. We align with VCMI & Oxford Net Zero contribution-claim guidance and the EU Green Claims Directive — no offset, neutrality, or compliance-substitute claims are made or supported.
Trust pillars
What this approach buys you.
Deterministic compute
Same input → same output. The FRAME formula is a multiplication, not a model. Auditors can reproduce every number.
Source citations
Each emission factor carries its source. Each measurement carries its provenance. Every figure is one click from its receipt.
Conservative baseline
NL counterfactual of 0.93 (incineration with energy recovery), not the global default of 1.0. Less generous on purpose.