UpGreen matches each address against the MHCLG Open EPC Register, which holds records for every property in England and Wales with an EPC certificate issued since 2008.

For each address, UpGreen returns:

• Current EPC band (A–G)

• Inspection date and expiry date

• Floor area

• Property type (domestic / non-domestic)

• Building archetype

• Original assessor recommendations

On matching accuracy. Address matching is the single hardest problem in portfolio EPC analysis. Commercial addresses, multi-let buildings, suite numbers, and inconsistent naming conventions across systems mean that no matcher gets every record on the first pass.

UpGreen runs deterministic, statistical matching rather than language-model matching. This means the same address list run twice returns the same result every time — no variance between runs. Where addresses don't resolve cleanly, the platform surfaces candidate certificates for human review rather than guessing.

Every match comes with a downloadable match report: the original address you supplied, the matched EPC address, the reason for the match, and the band and inspection date. You can review every decision and archive any false positives — the system learns from the corrections.

For each property, UpGreen models the relevant retrofit interventions in a defined sequence:

Demand-side first.

• LED lighting and controls

• Building management systems

• Low-cost behavioural and operational measures

On-site generation.

• Solar PV (where the building archetype supports it — UpGreen excludes flats and unsuitable roof types automatically)

• Solar thermal and hot water systems

Fabric and heating.

• Wall, roof and floor insulation

• Glazing upgrades

• Mechanical ventilation with heat recovery (MVHR)

• Air-source and ground-source heat pumps

The sequencing reflects how retrofits actually get done — quick wins first, capital-intensive interventions last. Each intervention is modelled against the building's archetype, floor area, current EPC band and regional climate. The model predicts EPC band uplift, energy savings, carbon savings and capex for each option.

The underlying physics — U-value calculations, heat loss modelling, solar gain — uses standard methodologies. The assumptions about building fabric (wall thickness, glazing type, roof area) are based on archetype-level defaults, refined with thermal imagery and Google Maps API roof area data where available.

These assumptions are accurate enough for portfolio-level prioritisation. They are not a substitute for an on-site survey. UpGreen tells you which buildings warrant the survey.

Capex estimates are built from cost curves trained on real quantity surveyor and contractor pricing from completed UK retrofit projects. The curves are continuously updated as new project data is added.

Validated accuracy: within 5–10% of implemented project quotes when the same interventions are used.

The cost curves are not generic benchmarks. They reflect actual UK market pricing for the specific intervention, building archetype and region. Where partners upload their own QS quotes against assets, the platform recalibrates against those figures for that partner's book.

Each intervention's effect on asset value is modelled per asset using assumptions developed with lender panels. UpGreen produces:

• Valuation uplift from improved EPC band

• Brown discount exposure for sub-C assets

• Lender covenant headroom impact

• MEES enforcement risk position

• Rental impact — built from public rental data and capitalised against current and post-intervention EPC bands

For owners who hold private rental data, UpGreen can take rent rolls as a direct input and produce actual ROI rather than modelled ROI. Outputs are designed for investment committees and lender conversations, not just technical reports.

For each asset, UpGreen runs the MEES exemption tests:

• Seven-year payback test — the most commercially important. On portfolios analysed via UpGreen, this test alone has cut active retrofit capex liability by more than 70%.

• Consent-related exemptions — tenant or third-party consent denied

• Devaluation exemptions — where works would reduce property value by more than 5%

• Wall insulation exemptions — where wall insulation would damage the property

Assets likely to qualify are flagged with the supporting data needed to file an exemption application.

EPCs issued before 2014 used different methodologies. Many properties currently rated D or below would already meet C under reassessment.

UpGreen flags every certificate where reassessment is likely to be cheaper than retrofit, surfacing it as the lowest-cost compliance path where it applies. On a recent 600-property analysis, 241 assets fell into this category — a fraction of the original retrofit Capex bill.

The Phase 0 analysis is the entry point. From there, UpGreen keeps the portfolio picture current.

Public data refresh. New EPCs lodged on the MHCLG register against assets in your portfolio are picked up automatically. Stale certificate flags update as reassessments are recorded.

Document upload. Survey reports, retrofit completion certificates, MEES exemption filings and asset management plans can be uploaded against individual assets. The platform integrates them and refines the assumptions for that asset and across the portfolio. Over time, the picture moves from predicted to actual.

Rent and tenancy data upload. Owners and managers can upload rent rolls, lease data and tenancy schedules. Outputs recalibrate against actuals rather than public assumptions.

Portfolio change. Acquisitions and disposals are added or removed in real time. The virtual estate reflects the current portfolio, not the one that existed when the first analysis was run.

BD and renewal triggers. Expiries, lapsed EPCs and clustered renewal windows are surfaced across the portfolio — useful for owners planning Capex and partners managing client books.

Audit trail. Every change is logged with date and source. You can see what the portfolio looked like at any past point — useful for lender reporting, board updates and regulatory submissions.

A portfolio dashboard, exportable as PDF or structured data. Filterable by EPC band, archetype, region, exemption status, capex band, expiry date. Every figure links back to its source.

Outputs include:

• Portfolio overview — every asset, every band, every expiry, every MEES position

• Compliance roadmap — costed, sequenced interventions per asset

• Exemptions report — assets likely to qualify, with supporting data

• Reassessment candidates — stale certificates likely already at C

• Postcode and cluster views — for routing surveyor visits or scoping geographically clustered work

• Summary statistics — portfolio-wide cost to C, cost to B, full retrofit, valuation uplift, opex reduction, carbon reduction

• Optional AI qualitative summary — for board and committee narratives

For partners, all outputs are white-labellable.

The first delivery arrives within 48 hours of the address list being submitted. Once populated, the virtual estate is available on demand.

UpGreen does not deliver EPC certificates. It does not produce SAP or SBEM calculations. It does not deliver DECs, TM44 inspections, or PAS 2035 Level 5 retrofit assessments. It is not a digital twin and does not model individual-building physics at high fidelity.

UpGreen is the upstream triage layer that tells you where to deploy those tools and services. It is built to be fast and broad rather than deep and individualised.

For deeper work on individual buildings, UpGreen integrates with EPC assessors, surveyors and PAS 2035 retrofit coordinators — pointing them at the right assets, then holding their output afterwards.

Data sources

UpGreen pulls from a defined set of public registers, project datasets and user uploads. Every figure on the platform traces back to one of these sources.

MHCLG Open EPC Register. The public register of every EPC issued in England and Wales since 2008. UpGreen uses it for current EPC band, expiry date, floor area, building archetype and the original assessor's recommendations on each property.

QS and contractor project data. A continuously updated dataset of real quantity surveyor and contractor pricing from completed UK retrofit projects. This feeds the Capex cost curves and intervention pricing — the figures behind every cost estimate the platform produces.

Lender panel valuation models. Valuation assumptions developed with UK lender panels, applied per archetype and region. These drive the asset value impact, brown discount exposure, and lender covenant headroom outputs.

Google Maps API. Used for roof area calculations and solar potential modelling. Where the API returns usable roof geometry, the platform refines its solar PV recommendations against the actual building footprint.

Companies House. Used to derive an address list from a registered company or registered society number, for portfolio analysis where ownership is the starting point.

Ordnance Survey and OS Open Data. Used for address matching and regional climate data. Address matching is deterministic against the OS dataset rather than language-model-based, which is why the same address list returns the same matches every time.

Public rental data. Used to model rental impact and capitalisation across EPC bands. Drives the rental ROI and valuation uplift figures for assets where the user hasn't uploaded actuals.

User uploads. Survey reports, retrofit completion certificates, MEES exemption filings, rent rolls and asset management documents uploaded by owners and partners. These recalibrate the platform's assumptions for the specific assets and portfolios they relate to — moving the picture from predicted to actual.

The first six sources are public or licensed; the last two — QS project data and lender panel models — are proprietary to UpGreen and not available elsewhere. The user upload layer is what closes the loop between portfolio-level modelling and on-the-ground reality.

• Capex: within 5–10% of QS quotes on the same interventions

• EPC matching: deterministic and reproducible; every match auditable

• Valuation impact: modelled per asset against lender panel assumptions

• Limitations: without an uploaded survey, all building-fabric assumptions are archetype-level defaults. UpGreen flags where survey data would materially improve the picture.

The platform does the best it can with the data available. Where partners and owners feed real data back in — surveys, completed retrofits, rent rolls — the picture sharpens. The system is designed to improve continuously as the data flowing into it improves.