How We Calculate Your 0–100 Score
A transparent, data-driven approach to home energy performance. Every formula, weighting, and data source is published. Built on open data, continuously evolving with community input and measured validation.
- 01 6 weighted components (Envelope 30%, Heating 25%, Ventilation 15%, Water 10%, Solar 12%, Storage 8%)
- 02 Continuous scoring curves — no step jumps like EPC A–G bands
- 03 Primary sources: EPC descriptions, PVGIS solar data — heuristic where direct measurement is unavailable
- 04 Goes beyond EPC A — rewards generation, storage, and grid flexibility (prosumer features)
- 05 Climate-relative rank + 90+ readiness context included (v2.3) — additive modeled/heuristic layers beside the core score
How the 0–100 Score Works (methodology v2.3)
Our score is a continuous 0-100 metric that builds on government data like EPC, enriched with real-world sources. It's designed to reward every improvement, updating as your home evolves.
Unlike static ratings, our score uses continuous curves for smooth progression — no discrete jumps. For example, improving insulation from R-20 to R-30 gradually increases your score based on performance curves.
Continuous Scoring Curves
Metrics like U-values (thermal transmittance) map to scores via smooth curves. Lower U-value = higher score, with diminishing returns at extremes.
COP to Score Mapping
Heat pump COP (Coefficient of Performance) is scored on a curve: ASHP from 3-4.3, GSHP 4+, benchmarked against regional averages.
Weighted Aggregation
Component scores are weighted and summed to your final 0-100 score. Region-specific normalization ensures fairness.
6 Component Breakdown
Your score is composed of six weighted components, each fed by multiple data sources for accuracy.
Envelope
Assesses insulation, airtightness, and thermal bridging of walls, roof, floors, and windows.
⚠ U-values are inferred from text descriptions, not measured. This is heuristic, not a full SAP fabric calculation.
Data Sources
- EPC (U-value descriptions → lookup table)
- Airtightness from EPC when present (defaults to ACH50=10)
- SAP 10.2 (benchmark ranges)
- OSM geometry (when available for bridging estimate)
Heating
Evaluates heating system type and efficiency, including boilers, heat pumps, and distribution.
⚠ Rule-based keyword matching from EPC descriptions. Not a full plant simulation.
Data Sources
- EPC (system type, descriptors)
- COP band mapping (ASHP/GSHP)
- Weather compensation / low-flow bonus flags
Ventilation
Measures air quality systems and heat recovery efficiency.
⚠ Rule-based scoring. MVHR efficiency is banded, not measured.
Data Sources
- EPC (vent type field)
- MVHR efficiency band mapping
Water
Analyses hot water heating system type and efficiency.
⚠ Pattern-matched from EPC descriptions.
Data Sources
- EPC (hot water field)
- HPWH / stratified tank / DWHR pattern matching
Solar
Calculates on-site renewable generation based on enriched data and system flags.
⚠ Without a full demand profile, solar utilisation is estimated from defaults. Generation data requires PVGIS enrichment.
Data Sources
- PVGIS (solar irradiance + annual generation)
- EPC (existing PV flag)
- DC coupling / bifacial / tracker flags
Storage / Grid
Evaluates battery storage, grid flexibility, and virtual power plant participation.
⚠ Capability score, not a verified operational optimisation model.
Data Sources
- Battery capacity bands
- VPP participation flag
- TOU optimisation flag
- Thermal storage flag
Our Technology Database
200+ technologies researched across 7 categories. Continuously updated from research pipeline (YouTube, ArXiv, Google Scholar).
Envelope and Insulation
35 technologies
Heating and Cooling
45 technologies
Ventilation and Air Quality
20 technologies
Water Heating and Efficiency
15 technologies
Renewables and Solar
25 technologies
Storage and Grid Integration
35 technologies
Smart Controls and Automation
25 technologies
Scoring Curves Explained
We use mathematical curves to map raw metrics to 0-100 sub-scores, ensuring smooth progression and realistic diminishing returns.
U-Value → Score (Envelope)
Lower U-values (better insulation) yield higher scores via a sigmoid curve. Example: Wall U-0.18 (R-31) scores ~85, improving to U-0.12 (R-47) reaches 95.
COP → Score (Heating)
Heat pump efficiency on a linear-to-exponential curve. ASHP COP 3.5 scores 70, 4.3 scores 90; GSHP 4+ can exceed 95.
Other Examples
- MVHR recovery: 90% → 80 score, 95% → 95 score
- Battery self-consumption: 30% → 40 score, 60% → 85 score
- Solar yield: Normalized to roof potential via PVGIS
Climate-Relative Rank
Your score report includes a climate-relative rank that compares your home against similar properties in the same climate zone, region, and building type. This is an additive context layer — it does not change your score.
⚠ Important: this comparison is modeled, not empirical
The distributions used for ranking are synthetic — derived from building stock data (UK EPC register, NREL ResStock for US, IS 5281 norms for Israel). They are not computed from real user scores in this system. Ranks are never described as percentiles; they use named tiers to avoid false precision.
How Cohorts Work
Your home is placed in a cohort defined by:
- Climate band — simplified Köppen-Geiger from your coordinates (e.g. temperate, cold, mediterranean)
- Region — UK, US, or IL (distributions differ per country)
- Property type — flat, terrace, semi, detached
- Floor area band — small (<70m²), medium, large, very-large
Example cohort key: temperate:UK:semi:medium
Tiers, Not Percentiles
- LeadingTop ~10% of modeled cohort
- Above averageTop 10–35%
- TypicalMiddle 35–65%
- Below averageBottom 35–65% (15–35th)
- LaggingBottom ~15%
Data confidence: UK = high (EPC register), US = medium (ResStock), IL = low (code norms). Confidence is shown alongside your rank.
Unsupported countries currently return a note asking where you're scoring from so we know which markets to expand next.
Potential Score: Likely vs Stretch
Alongside your current score, we show a potential range — what your score could reach with targeted improvements.
Likely
Top 3 high/medium-confidence recommendations applied to your current score. Represents a realistic improvement from 2–3 targeted upgrades. Capped at 95.
Stretch
Top 6 feasible recommendations applied (including caveated ones; excludes infeasible). Represents the outer bound if all practical measures are pursued. Capped at 98.
⚠ Potential is bundle simulation, not a certified forecast
Score impact values in our recommendation catalog are manually assigned estimates — not derived from physics calculations or SAP. Interaction effects between co-installed measures are not modelled. Treat the range as directional, not as a guaranteed savings figure.
Beyond Net Zero: Buildings as Batteries
EPC A is not the finish line — it's just where the government stops counting. Our score goes to 100 because homes can do more than minimise consumption. They can generate, store, and export energy; respond to grid signals; and become active nodes in a distributed virtual power plant.
A score of 100/100 represents an energy-positive prosumer home — one that generates approximately 6,000 kWh/year from solar while consuming only 4,000 kWh, resulting in a net export of roughly −2,000 kWh/year for a typical UK 3-bed semi. The home doesn't just pay its own energy bill — it credits it.
Thermal Mass as Storage
Buildings store energy as heat in walls, floors, and thermal mass — free batteries that need no chemistry. Hedar et al. (2023, Building Simulation) quantified this: homes provide grid flexibility through thermal inertia, pre-heating or pre-cooling during cheap/clean grid windows and coasting through peaks. Our score rewards homes with high thermal mass and smart scheduling capability.
Heat Pumps for Load Shifting
A heat pump paired with a well-insulated building envelope is a controllable thermal load. Power-to-Heat during surplus renewable periods stores energy in the building fabric itself. On Octopus Agile or similar dynamic tariffs, a smart ASHP can run predominantly on cheap overnight electricity, shifting kilowatt-hours from grid-stress periods to off-peak abundance.
Solar + Battery + V2G
Rooftop solar generates. A home battery (10–20 kWh) stores and time-shifts. Vehicle-to-Grid (V2G) adds 40–80 kWh of EV battery into the equation. Together, these create a system that can island during grid stress events and sell power back at peak prices — turning household energy into a revenue stream rather than a cost centre.
Every Home as a VPP Node
A single prosumer home is interesting. One million of them, coordinated via demand response APIs, form a virtual power plant (VPP) that can provide gigawatt-scale grid balancing services. Evolving Home's score is designed to track and reward each home's contribution potential — not just its own consumption efficiency.
What Score 100 Actually Means
(UK 3-bed semi)
vs 4,000 kWh consumed
prosumer asset
Research basis: Hedar et al. (2023) "Buildings as Batteries: Leveraging Thermal Inertia for Grid Flexibility," Building Simulation journal. Heat pump Power-to-Heat storage validated against UK Climate Change Committee demand flexibility estimates.
What is heuristic vs live vs planned
Live (actively used in scoring)
- EPC fabric descriptions → U-value lookup → component score
- EPC system type fields → rule-based component scores
- Airtightness from EPC when present
- PVGIS solar generation when enriched
- Storage/grid capability flags (battery, VPP, TOU)
- Climate-relative rank (v2.3, modeled distributions)
Heuristic (in use, but approximate)
- U-value inference from EPC text (not measured)
- COP scoring from qualitative descriptors
- Thermal bridging default junction count
- Potential score impact values (manually assigned)
- Confidence / uncertainty (completeness-based, not error-calibrated)
- Climate rank distributions (synthetic, not from real user scores)
Planned (not yet in core scoring)
- Smart meter time-series calibration
- Full OSM geometry coupling in every score
- Thermal-camera / retrofit proof ingestion
- Calibrated uncertainty from real measured error
- Public contribution pipeline for model updates
Known limitations (current model)
- Some component mappings are still rule-based and text-driven.
- Confidence and uncertainty are estimated from input completeness, not measured calibration error.
- Validation dataset includes both measured and benchmark archetypes; measured sample is still small.
Our Open Methodology Commitment
Everything here is open. Full formulas, weights, and update logs are in the repo. We believe energy scoring should be auditable — not a black box. Where something is heuristic, we say so.