Case studies/Crisp
Online grocery
Netherlands
Cradle-to-shelf CO₂ · LCA
ISO 14067 · verified by SGS
ISO 14067 · verified by SGS
An entire grocery catalogue footprinted. 3,873 SKUs, 100% coverage.
- Crisp, the Dutch online supermarket, wanted a carbon footprint for every product it sells: not a sample, the whole range
- Product metadata shared in early December
- By February, all 3,873 SKUs had a cradle-to-shelf CO₂ analysis and LCA reporting, straight through the retail peak
3,873SKUs analysed: Crisp's entire range, food and non-food
100%coverage: every SKU received a final CO₂ value
36categories · 171 subcategories · 1,153 product types
45countries of origin across 109 sourcing cities
10+LCA databases and studies: Agribalyse 3.1, Poore & Nemecek, Concito and more
~11 wksfrom data handover to full final delivery, over the holidays
The deployment
Delivered through the seasonal peak, not after it
- Timing was the hard part: export shared in the first week of December, peak season
- A traditional programme would have scheduled kickoff for the second quarter
From handover to full coverage
Early DecemberData handover
- Full product set: names, ingredients, weights, origins, packaging
- No reformatting asked for
Four weeks laterFirst insights delivered
- Full data-quality readout of the range
- Sourcing mapped across 45 countries
FebruaryFull range, final numbers
- Final cradle-to-shelf CO₂ per SKU with uncertainty range
- Subcategory benchmark rank and reduction levers
Later that yearRefreshed
- Re-run as the assortment evolved
- A batch job, not a new engagement
Crisp on working with GreenSwapp
- 75+ seasonal SKUs in scope: kruidnoten, chocolate letters, oliebollen, stollen, Easter chocolate, summer wines, barbecue lines
- On a consultancy timeline their footprints would have arrived after delisting
- Here, the Christmas assortment was footprinted while it still mattered
What we did
Grocery-grade LCA, at catalogue scale
- Grocery data is messy: ingredient lists that do not add up, missing barcodes, partial origins
- The pipeline is built for that, and gap-filling was our job, not Crisp's
1Structure and translate
- Raw Dutch export: names, ingredients, nutrition tables
- Parsed, translated, normalized
2Match every product to a footprint path
- Most accurate route per SKU: product-type match to full ingredient breakdown
- Input-output analysis for non-food
- Nothing left unfootprinted
3Compute baseline CO₂
- Baselines from ingredients and origin
- Agribalyse 3.1, Poore & Nemecek, Concito, EU SuEatable and more
- 25th to 75th percentile uncertainty bounds
4Correct for how it is actually made
- Organic certification, greenhouse-grown, transport mode
- Down to shipping routes and fishing zones
5Add packaging and retail
- Packaging per SKU: primary, cover, label
- Store losses included, broken down by lifecycle stage
6Benchmark and recommend
- Ranked against subcategory average
- Reduction levers quantified per lifecycle stage
The results
A footprint map of the whole store
- Every SKU on one methodology, so category-level truths show at a glance
- Actionable per product: each number traces to its own ingredients, packaging and supply chain
Where the carbon sits in Crisp's catalogue
Median cradle-to-shelf footprint, kg CO₂e per kg of product. Ten largest categories by SKU count.
Catalogue-wide median: 4.3 kg CO₂e per kg. The spread across individual products runs across four orders of magnitude, a range no category average could capture.
2,377Organic sourcing, actually credited
- Baselines adjusted for organic supply chains
- Sourcing choices show up in the numbers
3,497Every product knows where it stands
- Benchmarked against subcategory average
- A ranked to-do list for buyers
Like for likeThe swap argument, quantified
- Animal products and plant-based twins on one methodology
- The difference is a defensible number, not a slogan
ISO 14067 methodology, verified by SGSEvery footprint on this page was computed with GreenSwapp's carbon footprint methodology, independently verified by SGS in accordance with ISO 14064-3 as meeting ISO 14067 and aligned to ISO 14040/14044. The numbers are defensible in front of a buyer, a regulator or a customer.
Time and cost
The math that makes full coverage possible
- Bespoke food LCA: €2,500 to €10,000 and 4 to 12 weeks per product1
- 3,873 SKUs the old way: a €10M+, multi-year programme, so nobody does it
- GreenSwapp: the entire catalogue in about eleven weeks, roughly 70 products per working day, at a small fraction of the cost2
- Refreshes are incremental, not a re-engagement
The traditional route
- €2.5k to €10k and 4 to 12 weeks per bespoke food LCA
- Full catalogue = €10M+ and years, so nobody does it
- Fallback: generic category averages, blind to sourcing
- Seasonal SKUs delisted before their footprint arrives
- Every assortment change re-opens the engagement
With GreenSwapp
- All 3,873 SKUs in about 11 weeks, 100% coverage
- SKU-specific: ingredients, origin, packaging, supply-chain tags
- Uncertainty ranges and multi-source corroboration per SKU
- Christmas assortment footprinted in season
- Refreshes are a batch job, not a new project
The hassle Crisp skipped
What Crisp's team never had to do
✕
No data cleanupThe export went in as-is, Dutch names, incomplete ingredient lists, missing fields and all. Gap-filling was our job, not theirs.
✕
No methodology decisionsDatabase selection, matching strategy per SKU, allocation methods, packaging models, loss rates: handled and documented, with the "how did we find this?" recorded on every row.
✕
No sampling compromisesNo "let's start with the top 200 SKUs". The long tail, the non-food aisle and the seasonal range were all in scope from day one.
✕
No holiday-season standstillThe analysis ran while Crisp's team ran their peak. The busiest weeks of their year cost the project nothing.
✕
No dead-end reportResults came with per-stage reduction levers and subcategory rankings: a roadmap, not a PDF that ages in a drawer.
Why this scales
The catalogue is the unit, not the product
- Multiple matching routes, so every product gets the most accurate method its data supports
- Reusable assets: ingredient dictionary, product-type hierarchy and packaging models for 1,153 product types
- The next grocer starts warm, not cold
- Twice the range takes more compute, not twice the engagement
More case studies
Full-catalogue coverage, this quarter.
Methodology. Product carbon footprints per ISO 14067 (methodology verified by SGS in accordance with ISO 14064-3). Cradle-to-shelf carbon footprints per SKU: ingredient- and origin-based baselines corrected with supply-chain attributes, plus per-SKU packaging (primary, cover, label) and retail store losses. Lifecycle coverage spans land use change, feed, farming, processing, packaging, transport and losses. Sources include Agribalyse 3.1, Poore & Nemecek, Clune, Concito, EU SuEatable, CarbonCloud and Cool Food, with 25th to 75th percentile uncertainty bounds per SKU.
1 Typical published ranges for consultant-led single-product food LCAs; varies by product complexity. 2 Exact engagement pricing available on request. Figures shown are aggregates from the Crisp delivery dataset.