How We Calculate Your Energy Predictions

If you're a council, housing association, or fuel poverty charity, you're not just looking at a dashboard. You're deciding whether to hand a resident an improvement plan, or which homes in your stock get investment first. Before any of that happens, you need to trust the number in front of you. Here's a plain-language look at where our figures come from, without giving away the exact mechanics a competitor could copy.

Why this matters

An energy prediction is only useful if it's defensible. If a resident asks "how do you know my bills will drop by that much?", or a board asks "why should we prioritise this street over that one?", the answer can't be "the software said so." It needs to trace back to a recognised methodology, real property data, and a process that catches numbers that don't make physical sense.

That's the bar we hold ourselves to on every property in Energicore.

The foundation: RdSAP

Our estimates are built on RdSAP, the Reduced Data Standard Assessment Procedure. This isn't a name we invented. RdSAP is the UK government's own methodology for producing official Energy Performance Certificates, used by every accredited domestic energy assessor in the country.

Building on RdSAP means Energicore isn't running a proprietary black box that happens to output an EPC-shaped number. It's the same underlying framework assessors use for lodged certificates, which is why our indicative estimates often sit close to a formal assessment, but what we show is still a modelled estimate, not an official or certified Energy Performance Certificate.

What goes into a prediction

We don't ask for (or need) an engineer's survey to produce a useful estimate. Instead, we work from a handful of categories of information, most of which we can already infer from official property and EPC data, topped up by whatever details you or the resident can add:

• Property archetype: the age, type, and construction of the building. A 1930s semi behaves very differently to a 2015 new-build flat, and archetype is the single biggest driver of a home's starting point.
• Fabric: the walls, roof, and glazing, and how well they retain heat. This is where most of the variation between two similar-looking homes actually comes from.
• Heating system and controls: what's generating the heat and how it's controlled, which shapes both running cost and carbon output.
• How the household uses energy: occupancy patterns and usage habits, which affect real-world bills even when two homes are otherwise identical on paper.

We don't publish the exact weighting we place on each of these, or the internal calculations that combine them. That's the part of our process we keep to ourselves. But the categories above are the honest, complete list of what shapes every number you see.

What comes out

From those inputs, Energicore produces:

• Energy use: estimated consumption for the property as it stands
• Running costs: what that energy use translates to on a typical tariff
• CO2 emissions: the carbon footprint of the home's current energy use
• EPC band and score impact: how a given improvement measure would move the property up the ratings scale

These figures don't sit in isolation. They're what your improvement plans, grant eligibility checks, and portfolio analytics are built on elsewhere in Energicore, so getting this layer right is what makes everything downstream trustworthy.

How we keep it honest

A methodology is only as good as the checks around it. We treat every prediction as something that has to earn its place, not something we publish and forget:

• Figures are sanity-checked against patterns we see across real, verified EPC assessments and known UK housing stock data. If a result looks physically implausible for a given archetype, it gets flagged rather than shipped.
• As more verified assessments and outcomes come through the platform, we use them to refine how our estimates behave over time, rather than treating the model as fixed.

We're deliberately not detailing the exact checking mechanism here, for the same reason we don't publish our weightings. But the short version is that no number reaches your dashboard without having been checked against reality first.