DQR Pedigree Matrix: scoring inventory data quality on 5 dimensions
How does EN 15804+A2's mandatory DQR system work? The 1-5 scale on each of five dimensions, the aggregate score, and triggering sensitivity scenarios.
By clca Editorial TeamLast updated
DQR (Data Quality Rating) scores each inventory item's data quality on a standardised 1-5 scale. The Pedigree Matrix method was developed in 1995 and is now mandatory under EN 15804+A2 and PEFCR. Computing DQR correctly is one of the first things a verifier checks.
The five dimensions
- TiR (Temporal Representativeness) — data age. 1: under 3 years, 2: up to 6 years, 3: up to 10 years, 4: up to 15 years, 5: over 15 years
- GeR (Geographic Representativeness) — geographic match. 1: exact country match, 2: sub-region, 3: region, 4: continent, 5: global average
- TeR (Technological Representativeness) — technology fit. 1: same technology + scale, 2: similar technology, 3: different technology same category, 4: category-generic, 5: untyped proxy
- C (Completeness) — missing values. 1: all relevant flows included, 2: 95%+ coverage, 3: 80-95%, 4: 50-80%, 5: below 50%
- P (Precision) — numerical accuracy. 1: measured, 2: calculated, 3: balanced estimate, 4: industry estimate, 5: rough estimate
Total DQR — arithmetic mean
EN 15804+A2 defines total DQR as the arithmetic mean of the five dimensions: DQR = (TiR + GeR + TeR + C + P) / 5. Example: TiR=2, GeR=3, TeR=2, C=2, P=3 → DQR = 12/5 = 2.4. This is "good". Inventory items with DQR > 3.5 trigger an automatic sensitivity scenario — the impact of varying that item by ±10% is reported on the total. This is concrete evidence the verifier wants to see.
Different focus for foreground vs background
The manufacturer's own process data (foreground) must be high quality — target TiR=1-2, GeR=1, TeR=1, C=1-2, P=1-2. Annual production data typically lands at DQR=1.5-2.0. Background data (ecoinvent, GaBi) comes from external sources; its DQR is usually 2.5-3.5. The weighted DQR average of your EPD blends foreground and background. Most PCR Part B documents require foreground share > 50%.
Strategies to improve DQR
- Request supplier-specific data (move from background to foreground)
- Check for local geographic variants (Europe → TR)
- Get a modern technology certificate (BAT — Best Available Technology reference)
- Refresh data collection date (replace 5+ year datasets)
- Measure for missing flows; replace industry estimates with measurements
- Replace P=4-5 estimates with P=2-3 calculated values
Tags
- DQR Pedigree Matrix
- data quality rating LCA
- inventory quality
- representativeness LCA
- sensitivity scenario LCA
Related reading
All articles in Data & Methods- Data & Methods
ecoinvent activity selection: choosing the best fit on 3 criteria
Picking the right activity from ecoinvent's 21,000+ entries is a methodological decision. Geographic, technological and temporal representativeness.
10 min read - Data & Methods
Digital EPDs: machine-readable formats and why PDF is not enough
EPDs are no longer read but processed. The ILCD+EPD data format, the digital product passport and direct import into building LCA tools.
9 min read - Data & Methods
Foreground vs background data: getting the distinction right
The split between foreground (manufacturer-specific) and background (generic) data is the foundation of an LCA's scientific credibility. Practical rules and edge cases.
8 min read - Standards
Filling the modulary table — 8 common mistakes that get verifier rejection
The most common mistakes when filling EN 15804+A2 Tables 8-11: MNR/MND confusion, module-chaining errors, GWP-fossil/biogenic imbalance, transport assumptions.
11 min read