We Hadn't Even Touched Anything Yet: What 44,000 Data Points Revealed on One German Asset
Zurück zu EinblickeCase Study

We Hadn't Even Touched Anything Yet: What 44,000 Data Points Revealed on One German Asset

neos Team·September 11, 2026·4 min read

We recently analysed 44,000 data points from one of our newly onboarded multifamily assets in Germany: 18 months of heat meter, gas meter, and outdoor temperature sensor data. The data alone already pointed to an annual savings opportunity of roughly €3,500–4,000 on a single asset, about 20% of its gas bill.

We're sharing the outcomes of what was merely a preliminary analysis, because this is the kind of thing asset owners should want to know before spending money on the wrong intervention.

The 3 key findings

1. The building fabric was not the villain. The measured heat-loss coefficient — i.e., the rate at which a building loses heat — was strong. The problem was the way the plant was being run.

2. The boiler was running at a fixed 83°C all year. Deep winter? 83°C. Mild spring day? 83°C. In July, with zero heating demand, the plant was still burning through ~17 m³ of gas per day.

3. Boiler efficiency sat around 87%, where ~95% should be achievable — likely due to inefficient cycling, i.e., the boiler switching on and off too often.

The key takeaway

Retrofit investments are great and undeniably the long-term path to net zero, but throwing more money at a problem doesn't necessarily lead to better outcomes. Measured data helps separate expensive problems from cheap ones.

A fabric-led intervention would probably not have been the most impactful place to look, and even new systems can fail to meet efficiency standards without active management.

Active management doesn't have to be resource intensive either. At least not with neos. We built an AI-autopilot that identifies cost and carbon-saving opportunities in real time and implements the optimisation autonomously, 24/7, without requiring manual intervention.

Don't worry, neos does not require 18 months of historic data to work. It learns the building during operation. Historic data just lets us spot the obvious waste before we even start.

If you're curious, and hold 12–18 months of meter data for a building or portfolio, we're happy to take a look for you. Get in touch with the neos team.

Jetzt starten

Bereit, Ihre Energie zu optimieren?

Schließen Sie sich Branchenführern an, die ihre Kosten mit KI-gestütztem Energiemanagement um über 30 % senken.