What an EPC rating actually tells you about your bills

Somewhere on the listing there's a small coloured graph with a letter on it. Usually a C or a D, usually ignored, occasionally the thing that makes someone walk away from a flat.

It's worth about thirty seconds of your attention, and those thirty seconds are better spent on the second page than the letter. Because the letter is answering a question you probably didn't ask.

What the rating actually measures

An EPC rates the building, deliberately and by design, as if a standard household lived in it. It is not a prediction of what you will spend.

The methodology says so in as many words. The Standard Assessment Procedure, which is what generates the number, describes itself as "a rating of the asset, so is independent of factors related to the individual characteristics of the household occupying the dwelling". It then lists what it ignores: household size and composition, which appliances you own, and "individual heating patterns and temperatures".

So the model heats the living area to 21°C, for nine hours a day, two in the morning and seven in the evening, whatever you actually do. If you're out all day and wear a jumper, your bills will be lower than the certificate. If you work from home with the heating on constantly, higher. Neither means the EPC is wrong. You're just not the household it modelled.

Two more things people get wrong:

It isn't a survey. Existing homes are assessed using RdSAP, which BRE describes as "a system of data collection together with defaults and inference procedures". Where the assessor can't see something, the software assumes a value from the property's age and construction. That's a screening tool, and a reasonable one. It is not somebody investigating your house. If you want that, you want a proper survey, which is a different exercise with a different price.

It's valid for ten years. The certificate you're looking at may well have been produced for a sale two owners ago, before the boiler was replaced or the loft was done.

The bands, and why a C isn't good news

The letter is a band, and the band is a range of points. Under the current specification:

A property on 69 points and a property on 80 points are both a C, and there's a real gap between them. If the certificate gives you the number, use the number.

Here's the part that reframes the letter. Across the active listings we hold, in London and the surrounding commuter belt, in September 2026:

C and D together are seven listings in ten. A is about one in 250. So a C isn't a good rating, it's the modal rating, and a D is entirely ordinary. Reacting to a D as though it were a defect will rule out a third of the market for no reason.

The band that should actually make you slow down is F or G, which is roughly one listing in eighty. Not because the home is unliveable, but because it usually means something structural and expensive: solid walls with no insulation, electric storage heating, single glazing, or all three.

The pounds on the certificate aren't your bills

The certificate estimates annual costs for heating, hot water and lighting. Those figures are not priced at what you'll pay, and they're not priced at today's rates either.

The specification is blunt about it: "The fuel prices given are averaged over the previous three years and across regions. Other prices must not be used for calculation of SAP ratings."

Read that again. The assessor is forbidden from using real prices. Every pound figure on an EPC is a three-year regional average frozen as at the day that certificate was lodged, which on a ten-year-old certificate makes it a historical artefact. The median modelled heating cost across the certificates we hold is £500 a year, which should tell you on its own that it isn't anybody's actual bill.

Use the pound figures to compare two properties assessed at roughly the same time. Don't use them to budget.

What "potential" actually promises

The right-hand side of the graph shows the potential rating, the band you'd reach if you did everything on the recommendations page. It reads like a promise. It's better understood as a ceiling.

Across the same listings, if every recommendation on every certificate were carried out:

• Only 1.5% would reach band A. Nine in ten would land at B or C.
• On 45% of certificates, the potential band is identical to the current band. You could do all of it and the letter would not move.
• The modelled heating cost falls on 66% of them, doesn't move on 27%, and goes up on 7%.

That last group sounds like an error and isn't. Some recommendations improve the rating or the carbon figure without cutting the modelled heating cost, and some swap a cheap fuel for a cleaner one.

The recommendations page is still the most useful part of the document, but read it as a to-do list with prices attached rather than a forecast. It comes with indicative costs, and on a solid-walled Victorian terrace those run to five figures.

What changed in June 2025

RdSAP 10 replaced a methodology that had been in place since 2012. The current specification is dated 9 June 2025 and the new version went live that month.

It collects considerably more data, handles ventilation properly, and treats heat pumps and electric heating far more sensibly than the old version did, which had a habit of punishing them. It doesn't fix the underlying data collection problem: the rating is still only as good as what the assessor recorded on the day.

The practical consequence for you: certificates last ten years, so the market is full of both regimes at once and will be for years. Two properties with the same letter may have been assessed under different rules. If you're comparing closely, compare the assessment dates too.

One more thing worth knowing if you might ever let the place out. Since 1 April 2020 a property below band E can't be let, exemptions aside. The government has said it wants privately rented homes at band C or equivalent by 2030, but that is an aim rather than a requirement, and the policy is still being designed. Don't buy or avoid anything on the strength of it.

What to do with the certificate

In the order worth doing them:

1. Find it yourself rather than trusting the listing graphic, which is one of the few numbers on a listing that actually has to mean something. The register at GOV.UK is free and searchable by postcode.
2. Check the assessment date. Older than about five years, and it predates any work the current owner has done, and possibly the previous methodology.
3. Read the points, not just the letter. The bottom of a band and the top of it are different homes.
4. Go straight to the recommendations page and read it as a cost list. The expensive items are wall insulation and anything involving the heating system.
5. Compare the current and potential bands. If they're the same, the certificate is telling you there's no cheap headroom.
6. Look at what's driving a low rating. Solid walls, electric storage heaters and single glazing are expensive to fix. Loft insulation and lighting are not.
7. Ask the seller for actual bills. A year of real usage from the current occupants beats the model, with the obvious caveat that they may live very differently from you.

Nookaly surfaces the EPC band alongside council tax, tenure and area data on the listings it scores, which saves the lookup. The certificate itself is still worth opening.

The honest summary

An EPC rates the building under standardised assumptions, so it's a fair way to compare two homes and a poor way to predict your spending. A C is average, a D is ordinary, and F or G is the one that signals real money. The pound figures are averaged, regional and frozen at the certificate's date, so treat them as relative, not absolute. And check the potential band before you get excited about it, because on nearly half of certificates the potential is exactly what you've already got.

---

Figures for the band distribution and the current-versus-potential comparisons are from active listings in and around London and the commuter belt, sampled September 2026. These are listings rather than unique properties. EPC methodology and band boundaries are from the SAP 10.2 and RdSAP 10 specifications published by BRE, checked 3 September 2026. This is general information, not energy, financial or legal advice.