Version 2.1 — final method for the 2026 edition
First published 30 August 2026 · This revision 31 August 2026 · Antwerp, Belgium
This method is fixed. It was published before any product was assessed and will not change for this edition.
We rate what a buyer can verify before purchase. We do not rate product performance.
A battery may be excellent and still score poorly here. A low score means one thing: the manufacturer does not publish enough, in a form a buyer can rely on, for that buyer to check the claims before signing. It does not mean the product is bad.
This distinction is not a disclaimer. It is the whole design, and every product page repeats it.
Why this, and not efficiency or capacity? Those are measured well already — laboratory programmes test them under controlled conditions. What no one measures is the thing a European household actually faces: a €8,000–15,000 purchase, a ten-year commitment, and a set of documents that may or may not exist, may or may not apply in their country, and may or may not still say the same thing next year.
Fixed-installed residential battery storage systems, usable capacity ≥5 kWh, requiring professional installation. This covers AC-coupled integrated systems, DC-coupled battery modules, and battery systems sold for use with a specific hybrid inverter.
Commercial and industrial storage. Assessed in a pilot and not published as a rating. C&I buyers obtain documents through tender processes under confidentiality, so public disclosure is a much weaker proxy for what a buyer actually receives. Rating them on the same basis would misrepresent both.
Portable power stations. Not a fixed installation; different category, different use.
Balcony and plug-in storage. Excluded from this rating, and the reason matters: it is not that our indicators do not apply — most of them apply, and some apply more sharply, since these buyers have no installer standing between them and the manufacturer. It is that a €500–1,500 plug-in unit and a €10,000 installed system are different purchase decisions, and ranking them in one table would compare products that do not compete. This category will be assessed separately.
Selection is by published rule, not editorial preference, and applied Europe-wide without regard to the manufacturer’s country of origin.
A brand is included if it meets all three:
1. It sells a product within the category defined in 2.1
2. It has a verifiable sales or service channel in at least three European countries
3. It operates a European website or European-language product pages
“European country” means the EU, the EEA, the United Kingdom and Switzerland.
A sales or service channel means a local commercial or service presence: a distributor, an authorised service partner, or a local legal entity. Shipping to a country is not a channel — the point of the criterion is whether a buyer has someone to reach when something fails.
Brands are then ranked for inclusion by countable, checkable proxies, in this order: number of European countries with a verifiable channel; appearances in national grid-connection, subsidy or certification registers; existence of a European legal entity confirmed in an official company register. Where a manufacturer claims a distributor count without a country-level breakdown, only countries with evidence are counted.
The top 30 are rated, and ranks 31–40 are published with their evidence. If a manufacturer asks why it is not in the rating, the answer is already public.
We do not claim to know European market share. No reliable public source exists for residential storage market share by brand, and we will not rank by a figure we cannot show you.
Every indicator is answered from documents, not from claims. A document qualifies as evidence — grade E2 — only if it meets all three conditions:
| Condition | |
|---|---|
| Official | Published by the manufacturer, a certification body, or an official register, on its own domain |
| Complete | Contains the actual terms or figures for this product, not a framework, a summary, or a marketing statement |
| Fixed | Carries a version number or effective date, so it can be cited |
Anything else is graded E3 and earns no points, with the reason recorded and published:
| Reason | Meaning |
|---|---|
official_undated |
Official and complete, but carries no version or date |
incomplete_official |
Official, but does not contain the specific terms or figures for this product |
mirror_only |
Found only on a third-party site — a distributor, a download portal, a forum |
ND means no document answering the indicator could be found. ND is a finding, not an unfinished cell.
Product pages distinguish three kinds of ND:
| No document found | Nothing answering the indicator was located; the search paths are recorded |
| Document exists, no clause addresses this | The right document was found, but it is silent on the question |
| Verifiable, but no compliant document could be archived | The fact can be established, but no official document meeting our evidence standard was obtainable — typically where a national register charges for extracts |
The third kind reflects a limit of our method, not of the manufacturer’s disclosure, and is labelled as such. We would rather show you where our own reach ends than let it look like a company’s opacity.
Because a buyer cannot prove what it said on the day they bought. A warranty page that can be edited silently is not a commitment a household can rely on in a dispute three years later. This is a real loss to the buyer, not a formatting preference, and it is scored on that basis.
The distinction between “official but undated” and “marketing claim only” is real, and we report it — in the Disclosure Completeness Index (section 7), not in the score.
Many manufacturers sell both inverters and batteries and publish separate warranties for each. An inverter warranty is not a battery warranty, and the reverse is equally true.
This matters more than it sounds. In our pilot, one manufacturer’s inverter warranty runs two years from dispatch, while its battery warranty runs ten years with an 80% capacity guarantee and a published throughput table. Both documents are official, current, versioned and downloadable from the same site. Reading the wrong one would have understated that product by eight years.
A document is evidence for an indicator only if it names this product or its product family. Where a manufacturer publishes a combined warranty with separate sections, only the section covering the rated product is used.
Where a manufacturer publishes warranties for its inverters but none for its battery, that is recorded as an absence — a finding, not a gap in our search.
Adjacent clauses in the right document do not substitute for each other. The extract we publish must be the passage that answers the indicator.
Two examples from our pilot, both of which we caught and corrected in our own work: a clause continuing the warranty on additional modules bought later is not a clause on transferability to a new owner; a requirement that claims be filed through an authorised partner is not a condition making the warranty valid only if installed by one. The legal consequences differ.
Where a document exists but no clause addresses the indicator, the result is ND, and the product page says which.
An entry in an official company, producer or grid register is evidence when we record the register’s own entry identifier and archive a capture of the official page with its URL and retrieval date.
Where a register publishes only paid extracts and offers no free official view, we record the third kind of ND above rather than cite an aggregator. National registers differ in what they publish free of charge, and we will not let a registry’s pricing be read as a company’s opacity.
Every document we cite is retrieved, archived and machine-checked before it is read. A file qualifies only if it returned HTTP 200, has a content type matching its claimed format, passes a format signature check, and opens as a readable document. Every scored indicator is traceable to an archived file.
This is not housekeeping. A server that returns an error page under a filename ending in .pdf produces a file that looks, in a directory listing, exactly like a warranty. Read carelessly, it becomes “this manufacturer publishes no warranty” — a false finding about a real company. Files that fail the check are quarantined, never cited, and the indicator is recorded as not retrieved.
If two sources give different figures or terms for the same indicator, both are recorded and the indicator is graded ND. We do not pick one.
Manufacturers frequently differentiate warranty terms within Europe. One brand guarantees 80% capacity retention in nine countries and 70% in five others, and excludes the Nordics, Poland and Czechia from coverage entirely. Another differentiates by country group. A third gives 80% in Germany and 60% elsewhere in the EU for the same product.
A single European figure therefore does not exist for many products. We resolve this by rule:
We state the reason for choosing Belgium, and its consequence. BessCare’s editorial team and readership are based in the Benelux. Belgium also falls in the more favourable tier for most manufacturers that differentiate. We do not pretend that no choice was made.
Forty indicators across six dimensions. Each is scored 0–5, and only E2 evidence can earn points.
| Indicator | Scoring | |
|---|---|---|
| 1.1 | Warranty document downloadable without registration | 0 / 5 |
| 1.2 | Warranty term in years | ≤5: 0 · 6–9: 2 · 10: 3 · 11–14: 4 · ≥15: 5 |
| 1.3 | Guaranteed capacity retention at end of term | <60%: 1 · 60–69%: 2 · 70–74%: 3 · 75–79%: 4 · ≥80%: 5 |
| 1.4 | Guaranteed throughput (MWh) or cycle count | none: 0 · one: 3 · both, with basis of measurement stated: 5 |
| 1.5 | Relationship between term, throughput and cycles stated explicitly | 0 / 5 |
| 1.6 | Transferable to the next owner of the property | no clause: 0 · conditional: 3 · automatic: 5 |
| 1.7 | Named legal entity performing the warranty in Europe | none: 0 · named, non-EU: 2 · named EU entity: 5 |
| 1.8 | Exclusions listed in the document itself | referenced elsewhere: 0 · partial: 3 · complete: 5 |
| 1.9 | Warranty conditional on an authorised installer | yes, network not searchable: 1 · yes, searchable: 3 · no: 5 |
On 1.9: requiring professional installation is sound safety practice and is not penalised. What is penalised is conditioning the warranty on a closed network the buyer cannot inspect — because when an installer leaves the market, the buyer is left holding the consequence.
| Indicator | Scoring | |
|---|---|---|
| 2.1 | European legal entity and registered address | none: 0 · name only: 2 · name, address, verifiable in an official register: 5 |
| 2.2 | EU Authorised Representative published | none: 0 · name only: 3 · name, address, contact: 5 · N/A for EU-established manufacturers |
| 2.3 | Location of European spare-parts stock published | none: 0 · “Europe”: 2 · specific country or city: 5 |
| 2.4 | Service response commitment expressed in time | none: 0 · qualitative only: 1 · stated in hours or working days: 5 |
| 2.5 | Authorised service network published and searchable | none: 0 · static list: 3 · searchable: 5 |
| 2.6 | Producer / EPR registration verifiable in a national register | no: 0 · one country: 3 · two or more: 5 |
| 2.7 | Support in a language other than English | none: 0 · one: 3 · two: 4 · three or more: 5 |
| 2.8 | Claims and RMA process published before purchase | none: 0 · outline: 3 · full process with timescales: 5 |
| Indicator | Scoring | |
|---|---|---|
| 3.1 | IEC 62619 | no certificate document: 0 · covers family, model not named: 3 · model named: 5 |
| 3.2 | IEC 63056 | as above |
| 3.3 | VDE-AR-E 2510-50 or TÜV 2PfG 2698 | as above |
| 3.4 | UN 38.3 | as above |
| 3.5 | EU Declaration of Conformity | as above |
| 3.6 | Grid-connection compliance in European countries | none: 0 · one: 3 · two or more: 5 · N/A where the product does not itself connect to the grid |
3.6 is N/A for DC battery modules, where grid compliance sits with the inverter. A product cannot be credited or penalised for a certification it is incapable of holding.
| Indicator | Scoring | |
|---|---|---|
| 4.1 | Full datasheet available without registration | 0 / 5 |
| 4.2 | Installation manual available without registration | 0 / 5 |
| 4.3 | Modbus register map or API documentation published | none: 0 · on request or under NDA: 2 · openly downloadable: 5 |
| 4.4 | Owner can export their own data in an open format | none: 0 · vendor app only: 1 · CSV export or open API: 5 |
| 4.5 | Firmware changelog published | none: 0 · undated: 2 · version, date and changes: 5 |
| 4.6a | Cloud dependency disclosed, including consequences of losing connectivity | not disclosed: 0 · mentioned: 3 · consequences stated: 5 |
| 4.6b | System operates without the manufacturer’s cloud | no: 0 · partial function: 3 · full local operation: 5 |
This dimension scores whether a figure is accompanied by the basis on which it was determined. It never scores the figure itself. A round-trip efficiency of 89% and one of 96% score identically if both state the standard and test conditions.
| Situation | Grade | Score |
|---|---|---|
| Figure published with the applicable standard or full test conditions | E2 | 5 |
| Figure published, partial conditions stated | E2 | 3 |
| Figure published, no basis stated | E2 | 1 |
| Only a qualitative phrase, no figure | ND | 0 |
| Nothing published | ND | 0 |
Applied to: 5.1 usable capacity · 5.2 round-trip efficiency · 5.3 expansion limit · 5.4 backup and off-grid capability · 5.5 operating temperature range.
A ten-year warranty is only as good as the entity behind it. In April 2026 a major European home-battery manufacturer entered administration and its warranties stopped being honoured; the hardware kept working, the commitment did not.
| Indicator | Scoring | |
|---|---|---|
| 6.1 | Years trading | <3: 1 · 3–9: 3 · ≥10: 5 |
| 6.2 | Ownership and listing status transparent | opaque: 0 · partial: 3 · listed or fully disclosed: 5 |
| 6.3 | Financial information published | none: 0 · summary: 3 · full annual report: 5 |
| 6.4 | Evidence of European installed base | none: 0 · company-wide figure only: 3 · European figure with basis stated: 5 |
| 6.5 | Years since the European entity was established | <2: 1 · 2–5: 3 · >5: 5 |
On 6.4: contracted is not installed. Where a source says “contracted”, “ordered” or “under agreement” rather than “installed”, it does not answer this indicator.
Dimension score = points earned ÷ (5 × number of scored indicators in that dimension) × 100.
Indicators marked N/A are removed from both numerator and denominator, so an N/A neither helps nor harms. There are exactly two grounds for N/A — 2.2 for EU-established manufacturers, 3.6 for products that do not connect to the grid — both determined by objective attributes and both disclosed.
Overall score = 0.30 × Warranty + 0.25 × European service + 0.15 × Safety + 0.10 × Documentation + 0.10 × Declared basis + 0.10 × Corporate standing.
The weighting is a judgement and we say so. Warranty and European service together carry 55% because they are what a household is exposed to for a decade after the installer’s van drives away. Readers may weigh these differently; the dimension scores are published separately so they can.
Three findings cap the overall score at 44 regardless of the calculated total. The trigger is stated at the top of the product page.
| Condition | |
|---|---|
| K1 | No EU Declaration of Conformity obtainable, and neither IEC 62619 nor IEC 63056 obtainable |
| K2 | No European legal entity identified, and no named European warranty entity |
| K3 | No public warranty document exists for the rated product |
Band names describe the state of documentation. They are not verdicts on product quality.
| Band | Score | Name |
|---|---|---|
| A | ≥75 | Fully documented |
| B | 60–74 | Well documented |
| C | 45–59 | Partly documented |
| D | 30–44 | Poorly documented |
| E | <30 | Largely undocumented |
Band boundaries were set before any product was assessed and will not be moved.
Published beside the score, never folded into it.
DCI = (E2 count × 1.0 + E3 count × 0.5) ÷ scored indicators × 100
E3 earns half weight here because “complete terms, not versioned” is genuinely better than “marketing claim only” — a difference the score does not capture and this index does. The breakdown by E3 reason is published, so a reader can see whether a manufacturer discloses imperfectly or barely discloses at all.
We expect manufacturers to disagree with their result. There is one way to change it, and we state it plainly because the alternative — a private channel — would destroy the rating’s only asset.
A document that is published becomes evidence. A document sent to us privately does not.
This is not bureaucracy. This rating measures what a buyer can verify before purchase. A warranty PDF emailed to our editors does nothing for a household comparing systems in Rotterdam. If it does nothing for them, it cannot change the score.
So: publish it on your own domain, dated or versioned, covering the rated product. Then tell us where. We re-assess that indicator and record the change with its date.
The only route to a better score is better public disclosure. That is the entire point.
Factual corrections and notifications of newly published documents go to hello@besscare.eu and are handled by editorial staff.
Our commercial staff do not receive, forward or evaluate evidence, and cannot initiate a re-assessment. If a manufacturer sends documents to a salesperson, that salesperson will redirect them to the editorial address and will take no part in what follows.
Licensing (section 11) is the right to reproduce a result that has already been published. It is not a service, a review, or an appeal. A licence purchase does not trigger, accelerate or influence any re-assessment, and a manufacturer that never licenses is published identically to one that does.
We will not preview scores, share drafts before publication day, offer paid pre-assessment, or advise a manufacturer on how to score higher beyond what this document already states publicly. The method is the advice, and it is free.
BessCare is published by Vigla Media OÜ, with its editorial team in Antwerp. An affiliated company, BessRe, provides after-sales and field service for battery energy storage systems in Europe and has, or is negotiating, commercial relationships with some manufacturers in this rating.
Our response is not to exclude those brands — excluding them would itself distort the rating. It is to disclose:
Our licensing revenue is independent of scores. Licensing is open to every rated product at any score (section 11), so we gain nothing from scoring a product higher and lose nothing from scoring it lower.
Scores and rankings are not for sale at any price. Advertising and sponsored content on BessCare are always labelled, and neither placement nor spend affects a score, a rank, a publication date or a position on the page.
Documents are assessed as published on the date we retrieved them, and every indicator shows its own retrieval date. We do not claim a single assessment date for the whole rating. Where a manufacturer published or revised a document after our retrieval date, section 8.1 applies.
Any manufacturer whose product is rated may license the right to reproduce its published result in its own materials. The tariff is published, uniform, and not negotiable.
| Basic licence — score and band | €2,500 per product per year |
| Full reproduction — including dimension scores | €6,000 per product per year |
| Eligibility | Any rated product, at any score |
Conditions: the overall score and the assessment date must always be shown; where dimension scores are shown, all six must be shown; the licence lapses if the product changes.
An earlier draft of this method restricted licensing to products scoring 60 or above. We removed that restriction before publication, and the reason is one we would rather state than be asked about.
A minimum score would have given us a financial interest in higher scores. The more products cleared the bar, the more we could earn. That is the wrong incentive to build into a rating whose entire value is that it scores strictly.
With no threshold, our licensing revenue does not depend on how any product scores. We have no financial reason to prefer a high score to a low one, which is the only position from which we can credibly claim to score without fear or favour.
The consequence is that a licence is not a badge of quality. It is the right to reproduce a published result accurately — including a low one, if a manufacturer chooses to show it. Some will. Publishing your own weak result alongside a plan to improve it is a more credible thing to do than saying nothing.
Licensing confers no influence whatsoever over the assessment. See section 8.3.
On publication day, every rated manufacturer is notified and receives its full assessment, including every source cited.
Manufacturers have ten working days to identify factual errors. Where a document exists that we did not find, or where we cited a superseded version, we correct the assessment and republish with the change and its date recorded.
We correct facts. We do not adjust scores in response to disagreement with the method, and we do not remove published assessments.
Anyone — reader, installer, manufacturer — may report an error at any time at hello@besscare.eu.
This method is final for the 2026 edition. It was published on 30 August 2026, before any product was assessed, and it does not change for this edition.
Where an assessment shows that an indicator is poorly specified, the flaw is recorded, published in the edition’s notes, and corrected in the 2027 edition — never applied retrospectively, and never after scores are known.
The 2027 edition’s method will be published before its assessments begin, as this one was.
We would rather state these than have them discovered.
We measure public pre-purchase disclosure, not everything a buyer eventually receives. A manufacturer may supply excellent documentation on request, or through its installer network, and still score poorly here. We think what is public before purchase matters — it is what a household compares on — but it is not the whole picture.
We do not test products. No capacity, efficiency or safety measurement is performed. Where others measure those well, we say so rather than duplicate it.
Certificates are assessed as documents, not audited. We record whether a certificate exists, is obtainable, and names the model. We do not verify the testing behind it.
National registers are not equally open. A manufacturer registered where company extracts are paid may show ND on entity indicators that a manufacturer elsewhere would pass. We label these separately, but the asymmetry is real and we cannot remove it.
The reference geography is a choice. Using Belgium produces different figures than Italy would for manufacturers that differentiate. Every tier is published so a reader elsewhere can read their own.
Scores in this first edition are expected to be low across the field. Since only verifiable documents earn points, a product’s ceiling is set by how much of its documentation is publicly obtainable in citable form. In pilot assessments this ranged from roughly 20% to 70%. We have not adjusted the scale to produce a more comfortable distribution. If most of the European home-storage market lands in the lower bands, that is the finding, and it is the reason this rating exists.
| Version | Date | Status |
|---|---|---|
| 1.0 – 1.2 | Aug 2026 | Internal drafts, not published |
| 2.0 | 30 Aug 2026 | Final method for the 2026 edition, published before any assessment |
| 2.1 | 31 Aug 2026 | Licensing minimum score removed (11.1); third kind of ND introduced (3.1); official-register protocol added (3.6); register-openness limitation added (14). No indicator, weight or band changed. No product had been scored at the time of this revision |
BessCare · Noorderlaan 147, 2030 Antwerp, Belgium · besscare.eu
Editorial and corrections: hello@besscare.eu
Published by Vigla Media OÜ (Estonia). Editorial team in Antwerp, Belgium.