Our sources and methodology

VersusIndex compares products on their specifications. We don’t test anything ourselves: we collect data from official registers and from manufacturers, check it, and show where every value comes from. This page explains which sources we use, what they do and don’t guarantee, and how we check manufacturer data.

The registers we use

There is no UK or European database of technical data for EV chargers, home batteries and inverters comparable to EPREL. For those, manufacturer datasheets are the main source, supplemented by the OZEV, ENA and Australian lists. Why use German, Dutch or Australian registers on a UK site? Because the same models are sold across these markets and those registers publish detailed, structured data that UK lists often don’t. We use the Australian data under CC BY 4.0: Based on Clean Energy Regulator (Commonwealth of Australia) material licensed under CC BY 4.0 ¹⁰. We never copy addresses or contact details from registers.

What a register does and doesn’t guarantee

Registers are not all equal. It helps to tell three kinds apart:

• Third-party certification – Heat Pump KEYMARK. The values are measured to European standards and certified by a certification body ³. This is the strongest assurance of measured performance.
• Legal registration – EPREL. The supplier registers the data behind the energy label itself; market surveillance authorities check samples ⁴.
• Grant and approval lists – OZEV, ENA, BAFA, RVO, CEC and CER. A model is listed because it meets the requirements of a grant scheme or grid connection standard. The technical values usually come from the manufacturer and have been checked against those requirements, not re-measured.

No register tells you the price, UK availability, installation quality or how much a product will actually use or generate in your home. Standard ratings are measured under fixed conditions; real-world results vary with your home, climate and use. An OZEV listing, for example, does not mean a charger qualifies for every grant, and an MCS certification for your installation is a separate matter.

Which source wins when they differ

Many products appear in several sources. For each specification we use the source with the highest priority: EPREL, then BAFA, Heat Pump KEYMARK, RVO, OZEV and ENA, then the Australian and Californian lists, and finally the manufacturer. Each value on the site shows which source was used, with a link. An official value is never overwritten by manufacturer data; if they differ, the discrepancy goes on a review list we check by hand.

How we check manufacturer data

Manufacturer data is essential but the least controlled, so stricter rules apply:

• Datasheets. For every value taken from a PDF datasheet we record a verbatim quote. A script downloads the PDF again and accepts the value only if the quote really appears in it, the value appears in that quote and it falls within plausible limits.
• Nightly check. Every night an AI model (Google Gemini) looks up specifications for a batch of products on the manufacturer’s official website. The model decides nothing: we fetch that page ourselves and accept a value only if the domain belongs to the manufacturer, the model name and the quote with the value appear verbatim on the page, and the value is plausible. Retailers, marketplaces, forums and other comparison sites never count as a source.
• New models. Each night we look at a few brands’ websites for models we don’t know yet. A new model only gets a page once its key figure – for example a battery’s usable capacity – has been confirmed.

New and incomplete products are checked first; manufacturer data we have already confirmed is re-checked after six months.

Values we calculate

Some figures we calculate ourselves, always with a fixed formula:

• SCOP from ηs. If a source only gives the seasonal efficiency ηs, we use the conversion from the energy labelling regulation: ηs = SCOP ÷ 2.5 × 100 − 3 ¹¹. For ground and water source units this slightly understates SCOP.
• Energy class from ηs, using the class limits in the same regulation ¹¹.
• GWP from the refrigerant name.
• Solar panel efficiency from peak power and panel area, when the source doesn’t state it.

We deliberately do not use part-load COP values from Keymark certificates as full-load COP: they are different test points.

How often we update

Every product page shows when its data was last updated.

When a product gets a page

A product only gets its own page once we have at least 8 comparable specifications. Models with fewer go on a waiting list that the nightly check fills in. Company names from registers (often a holding or national subsidiary) are converted to the everyday brand name. If a model disappears from the data, its old page redirects to the category page, or to the product the variant has been merged into, so links keep working.

What is in the database today

At the end of September 2026 there are nearly 1,800 products on the site. For heat pumps, values come from the German BAFA list (just over half) and from Heat Pump KEYMARK certificates, supplemented by the Dutch RVO list; we only show models available in the Netherlands, Germany or the UK (the model or brand is listed by BAFA, ISDE or MCS). EV chargers, home batteries, inverters and solar panels are mostly based on manufacturer datasheets, supplemented by the OZEV and ENA lists. We will fill the air conditioner category once the EPREL data is in.

Spotted a mistake? Let us know via our contact page. More about who we are is on our about page.