# Data quality in emissions accounting: primary data and estimates

> What data quality means in an emissions inventory: primary data, estimates and uncertainty. How to back data with documents and make your calculations more reliable.

Source: https://7g.carbonplatform.asia/en/kb/data-quality

Data quality in emissions accounting is the extent to which the source figures are complete, accurate, backed by documents and true to the actual activity. Primary data (invoices, certificates of completion, meter readings) is more reliable than estimates, and uncertainty shows the range within which the true value may lie. Data quality determines whether a bank, a customer or a verifier will accept your report.

## Primary and secondary data

**Primary data** consists of actual measurements and documents for your facilities: meter readings, invoices from utilities, delivery and acceptance certificates, vehicle trip logs.

**Secondary data and estimates** are values obtained indirectly: calculations based on floor area or equipment capacity, industry averages, extrapolation for missing months.

Estimates are acceptable when primary data is not available, but their share should be visible and should shrink over time. The report states which sources were calculated from estimates and by what method.

## Principles of a high-quality inventory

The GHG Protocol and ISO 14064-1 rest on shared principles:

- **completeness**: all significant sources within the boundaries are included;
- **consistency**: the same methods are used year after year;
- **accuracy**: there is no systematic overstatement or understatement;
- **transparency**: data sources and assumptions are disclosed;
- **relevance**: the data meets the needs of the report's users.

## Uncertainty

Uncertainty has two components: the uncertainty of activity data (meter error, estimates) and the uncertainty of emission factors. When quantities are multiplied, their relative uncertainties combine approximately as the square root of the sum of their squares.

Example: gas metering error is ±2% and emission factor uncertainty is ±5%. The combined relative uncertainty is ≈ √(2² + 5²) = √29 ≈ ±5.4%. If consumption is 100,000 m³ and the emission factor in this worked example is 2.04 kg CO₂e/m³, emissions of 204 t CO₂e lie roughly in the range of 193–215 t.

In this example, refining the emission factor clearly has a bigger effect than replacing the meter. For more, see [Emission factors](https://7g.carbonplatform.asia/en/kb/emission-factors).

## How to improve data quality

1. Link every value to a document: an invoice, a certificate of completion or a photo of the meter readings.
2. Collect data monthly, as a single missing month is easier to spot and fill in.
3. Reconcile meter readings with the supplier's invoices.
4. Flag estimated values and the estimation method.
5. Check for sharp deviations from the previous month and from the same month last year.
6. Assign a person responsible for reviewing and approving each period.

The process for collecting documents is described in [Collecting energy and fuel data](https://7g.carbonplatform.asia/en/kb/energy-data-collection).

## Data quality for financial institutions

Banks and investors assess the quality of their financed emissions data on the PCAF scale from 1 to 5, where 1 is the best (verified company data) and 5 is an estimate based on economic activity. For more, see [PCAF data quality](https://7g.carbonplatform.asia/en/kb/pcaf-data-quality).

## How it works in the platform

- On the [Data entry](https://7g.carbonplatform.asia/data-entry) screen, data is entered month by month for each facility, and supporting documents are attached to records: invoices, certificates of completion, photos of meter readings, contracts.
- A month with no consumption is marked “No consumption” so that it does not look like a gap.
- The completeness check for a facility's year shows what data is missing. The source list is confirmed manually or automatically, once the year has ended and all Scope 1 and 2 sources are complete.
- Each facility can have a responsible employee; the year is closed by an administrator or by the responsible employee with edit rights, and a change history is kept for each facility's year.

## Frequently asked questions

### Can estimated data be included in the report?

Yes, if primary data is not available. You need to describe the estimation method and, over time, replace significant estimates with actual data.

### What share of estimates is acceptable?

There is no single threshold: it depends on the reporting program and the verifier's requirements. As a rule of thumb, significant sources should be based on primary data.

### Do we need to calculate uncertainty?

ISO 14064-1 provides for an assessment of inventory uncertainty. Even a simplified assessment helps you see which data to refine first.