Your A-rated buildings are the ones to check first.

If you manage a portfolio, you probably worry about the bottom of it. You know, the old buildings. The Es and Fs. The ones you already know need work.

‍

However, research suggests that you should look up. Not literally. Just further up your portfolio. 

‍

Buildings with the best EPCs (energy performance certificates) are often the ones where the label is least likely to match reality – and the gap runs in the direction you might not expect. They tend to use more energy than their certificate predicts, not less.

‍

Surprised? Here's the evidence (and why it stops being such an academic oddity from next year).

‍

The label is a calculation, not a measurement

An EPC doesn't tell you what a building uses; it tells you what a building would use – if it were occupied by a standard family behaving in a standard way in standard weather. Standardly.

‍

Keep in mind that this isn't our standpoint, mind you. It's what the people who run the system say.

‍

Denmark's own building regulation puts it plainly: the standard assumptions fit an average family, and since no such family exists, there will always be deviations. It even quantifies how easily the calculation shifts – turn the thermostat from 20°C up to 22°C and calculated energy use rises about 16%. Nothing about the building changed.

‍

And when Nykredit commissioned MOE to build a knowledge base on the energy efficiency of Danish property, the report said it in a single sentence: the label indicates the energy standard of the building, but not its actual consumption.

‍

So far, so … reasonable. A model is a model.

‍

But the error isn't random

If certificates were simply imprecise, the errors would be scattered. Some buildings above, some below – no discernible pattern. You could treat the label as noisy but even-handed, and a ranking built on it would still come out roughly right.

‍

But this isn't what the research finds.

‍

In 2012, Sunikka-Blank and Galvin examined 3,400 German dwellings and found measured consumption running about 30% below the certificate's prediction on average. But the size of the gap tracked the rating. 

‍

Poorly rated homes, around 500 kWh/m² a year, used roughly 60% less than predicted. Homes with better ratings, around 150 kWh/m², were out by about 17%. And for the most efficient dwellings, below 100 kWh/m², the pattern flipped: they used more than the calculation said they would. The authors called the first effect the prebound effect; the reversal at the top is the rebound effect. And it wasn’t just Germany; the findings held in the Netherlands, Belgium, France and the UK too.

‍

Nine years later, Coyne and Denny tested the same question in Ireland, with 9,923 households and 149,518 meter readings, and the results were the same. A and B-rated dwellings consumed 39.6% more than their certificate predicted. F and G-rated dwellings consumed 56% less.

‍

Interestingly, across the whole rating scale, best to worst, actual consumption varied by just 457 kWh a year. The labels spread buildings out. The meters didn't.

So the ranking is wrong, not just the numbers

The two studies together give us more than the basic claim of "certificates are inaccurate".

‍

The error is systematic, and it runs in opposite directions at the two ends of the scale. Worst-rated buildings are overestimated. Best-rated buildings are underestimated. 

‍

So when you sort a portfolio by label, you don't get a slightly messy version of the true order. You get an order that's been pulled the wrong way at both ends.

‍

We see the same thing

While we don’t have as much data as these studies, the data we do have points the same way. 

‍

Across eight buildings in Danish portfolios, every A-rated building used more energy than its certificate estimated. One used more than three times as much. Three of the four C-rated buildings used less than predicted.

‍

And while it’s not 1000s of properties, what this result does is confirm that a pattern documented across six countries and more than thirteen thousand dwellings turns up here too – in Danish stock, in buildings we work with every day.

‍

Why this stops being academic once Denmark publishes its thresholds

Until recently, a mis-ranked portfolio was just an annoyance. Now it's becoming a compliance issue. Aka a real drag. 

‍

The recast Energy Performance of Buildings Directive (EPBD) works through the bottom of the stock. For non-residential buildings, member states set minimum standards pinned to the worst-performing slice – 16% by 2030, 26% by 2033 – with the threshold derived from energy use. For homes, each country runs a national trajectory instead, and at least 55% of the required reduction has to come from the worst-performing 43% of the stock.

‍

Both regimes do the same thing. They identify the bottom of a ranking and attach obligations to it.

‍

Which puts the problem in one sentence. 

‍

The instrument that decides your obligations is a calculation with a known, directional error – and there's no measured backstop.

‍

In Denmark, the number a building gets judged on will come from its certificate rather than its meters. The certificate is the official figure on file. Actual consumption does appear on it – there's a section for reported consumption from invoices – but it isn't what sets the letter. The calculation is.

‍

So the mis-ranking doesn't just mislead you privately. It carries through into who picks up an obligation and who doesn't. Two buildings can swap places: a genuinely heavy consumer with a good label escapes, and a modest consumer with a poor label gets caught. Neither outcome has much to do with how much energy either building actually uses.

‍

What to do about it

But none of this makes the certificate worthless. It describes the fabric of a building under fixed conditions, and that's a genuinely useful thing to know. It just isn't a measurement, and it shouldn't be used as one.

‍

So what can you do that will help?

‍

Start by putting actual consumption next to the label for every building you own. But read that comparison carefully, because most of what you'll see is the pattern described above rather than anything specific to your portfolio – poor labels overestimating, good labels underestimating. 

‍

The buildings worth your attention are the ones that diverge more than their label band would predict. That's where something building-specific is going on: insulation that isn't what the paperwork claims, a system running differently from the record, occupancy nothing like the standard case.

‍

Even then, a divergence tells you where to look rather than what to do. Turning it into action needs the things the model doesn't have – what you know about the building, how it's actually run, and who's in it.

‍

And there's one more consequence most owners haven't worked through: improving actual consumption doesn't necessarily improve your rating.

‍

Better operation, better tenant behaviour, a mild winter – none of that moves a number built from fabric and systems under standard assumptions. 

‍

So you have two targets, and they aren't the same one: what lowers your energy bill, and what moves your label. Knowing the gap between calculation and reality is how you tell them apart.

‍

That comparison is what comundo keeps current. We connect directly to Danish utility companies for electricity, district heating, water and gas, so the real number stays live instead of ageing for a decade. No hardware. No meters to install. No invoices to chase.

‍

The certificate stays what it is, a snapshot; the live picture runs alongside it, and that comparison serves as a very useful indicator of where to look for optimisation. 

‍

That comparison is what comundo builds. We connect directly to Danish utility companies – electricity, district heating, water, gas – so real consumption data flows automatically. No hardware. No meters to install. No invoices to chase. The certificate stays what it is, a snapshot; the live picture runs alongside it.

‍

Start at the top of your portfolio. That's where the surprises are.