Data source
All domestic Energy Performance Certificates for England and Wales, downloaded from the Open Data Communities EPC register. The archive contains one folder per local authority (347 total), each with certificates and recommendations CSV files linked by a unique key.
Date of download: February 2026. Data includes lodgements from 2008 through early 2026.
Scale & deduplication
The raw dataset contains 29,069,352 EPC lodgements. Since a property assessed multiple times appears multiple times, we deduplicated by keeping only the most recent EPC per unique building reference number. This reduced the dataset to 20,589,231 unique properties — close to the ~24M dwellings in England and Wales.
For payback analysis, certificates missing cost data or with zero/negative savings are excluded.
Important caveats
Costs are modelled, not real. Heating, hot water, and lighting costs are SAP (Standard Assessment Procedure) estimates based on standardised occupancy, heating schedules, and fuel prices at the time of assessment. They do not reflect actual energy bills or current prices.
Improvement costs are fixed. The "indicative cost" for each recommendation is selected from a predetermined list — the same range regardless of property size, location, or complexity. In many cases they significantly understate actual installation costs. Payback calculations are therefore optimistic — real-world paybacks will generally be longer.
Tenure classification
Tenure is recorded at the time of EPC lodgement. The dataset uses two naming conventions across its history (e.g., "rental (private)" and "Rented (private)") — both variants are included. Approximately 1.1M certificates with unknown or blank tenure are excluded from tenure-specific views.
Payback calculation
For each certificate with valid cost data in all six fields (heating, hot water, and lighting — both current and potential):
- Annual saving = total current costs minus total potential costs
- Total improvement cost = sum of all recommendation indicative costs (ranges converted to midpoint)
- Payback = total improvement cost / annual saving
Certificates with zero or negative savings, or with no parseable recommendation costs, are excluded from payback analysis.
Age bands
Construction age is taken from the CONSTRUCTION_AGE_BAND field, which uses predefined ranges for most records (e.g., "England and Wales: 1900-1929") with specific years for newer builds. These are consolidated into 12 age groups. Combinations with fewer than 50 certificates are excluded.
What this tool shows
All views in this dashboard are pre-computed cross-tabulations of the full dataset. Selecting filters retrieves the pre-aggregated statistics for that combination of tenure, property type, and construction period. No individual property data is stored or transmitted.