How the sheets are made

The data comes from the source, not from us

The specifications come from the manufacturer's official site, manuals and datasheets: next to every value are the page it was read from and the date we read it. We do not write from memory and we do not copy from shops. If a figure cannot be verified it stays empty: a missing field beats an invented number.

What counts is the document, not the server hosting it. An original datasheet or manual holds good even when downloaded from a retailer's site — in this market the official PDFs often sit there — provided it is the manufacturer's document, with its masthead and its version, and provided the sheet records where it came from. A shop's own page, by contrast, is never a technical source.

Three numbers nobody states: we calculate them ourselves

A calculated value has no source to cite: it has a formula, and the formula is printed on the page along with the assumptions it had to make. There are three of them.

The cost per kWh — the list price divided by capacity. We always say which of the two capacities we divided by: the nominal one, which is nearly always given, or the usable one, which is the real figure and which many do not state. Writing «usable» and dividing by nominal is the comparison sites' sleight of hand, and it is not done here.

The battery life cycles — not the brochure's. A warranty ends at the first limit to run out: the years, the total energy throughput or the guaranteed remaining capacity, whichever comes first. Six thousand stated cycles with a two-year warranty, at one cycle a day, are seven hundred and thirty covered cycles. They are two promises that never meet, and the sheet says so.

The cost of stored energy — what it costs to pass one kWh through that battery across its whole working life: the price divided by the energy that will actually pass through it. This is the number that changes the conversation, because it can be set against the grid electricity price in your country. When it is higher, storing costs more than buying, and the page says so.

What a manufacturer leaves unstated is data too

A dash in a table looks like an omission on our side. Here absence has three states, and they are not interchangeable: «the manufacturer does not state it» — we looked in a specific document and it is not there, and the sheet keeps the URL of where we looked; «not applicable» to that product; «not verified by us yet», which is no fault of the manufacturer's. Saying the first when the third is true would be inventing a figure in reverse.

This is also where the transparency index comes from: how many of its category's decisive attributes a product states. It is always computed within the same category, because comparing balcony storage with a home battery tower would measure two documentation cultures, not two manufacturers. And it measures disclosure, not merit: a mediocre product that publishes everything scores highly, and rightly so.

The class says how big it is, not who it is for

From F1 to F5 the class states the size: outlet-sized (up to 1 kWh), room-sized (1–3), flat-sized (3–8), house-sized (8–20), large-house-sized (over 20). It exists so that comparable things are compared — a 2 kWh power station and 2 kWh of balcony storage answer the same question, a 15 kWh tower does not — and it is not derived from price.

On prices we use the list, not the deal of the day

Comparisons and the cost per kWh are computed on the manufacturer's official list price, which is stable and has a source: a deal does not change a ranking. The best available price appears only in the buying block, with its label and its date. Where there is no official list price, the sheet shows no price and stays out of the cost calculations: we prefer a gap to an estimate dressed up as data.

The country changes the maths

Between Europe's dearest and cheapest kWh there is a factor of four, and export rules and incentives change every year. That is why the country is not a language setting but a calculation parameter: energy price, export regime, incentives and grid code are recorded per country and per date, each with its source, and the tools ask where you are. Product sheets carry no payback figure, because it would depend on your country and your installation: they carry the cost of stored energy, which does not depend on the country.

When it does not pay off, we say so

Storage does not always pay for itself, and quote portals never say so. Our calculations may conclude that it does not — and in that case the page says so, with the numbers that conclusion comes from. It is the thing that sets us apart, and the only one that earns a reader's trust.

Do we use AI? Yes, to write better

It helps us draft and translate the sheets' texts and catch mistakes. The numbers, no: they always come from the cited source, and in review we remove any sentence the data does not support. Comparisons between two products are composed automatically from the sheets, and say so at the top of the page.

Every sheet says who made it

Spec sheets and guides carry the JouleSpecs byline and the date of the last check. The bio next to the byline says how we work, not how good we are: we are a group, and anyone who joins with a role of their own will appear under their own name. Comparisons composed from the data carry no byline, and say so at the top of the page.

Data and opinions stay separate

A technical figure always has a source. An editorial judgement — strengths, caveats, who a product suits — is declared as such. We avoid manufacturers' vocabulary ("revolutionary", "up to 6,000 cycles"): if something is good the numbers show it, and a number without its conditions is not a number.

How we sustain ourselves

Affiliate links keep JouleSpecs running: if you buy through us, a small share comes back to the site and covers its costs, at no extra charge to you. No commission buys a judgement, and the technical data are the same whether or not an affiliate programme exists for that product.