Your clients now have to report the CO2 of every device they replace. You hold the numbers.
Circulife does two things. Case by case it says whether repairing or replacing is the better call on climate impact. At the end of the quarter it adds them up: what was bought to settle the claims, in kg CO2, and how much repairing instead of replacing saved. Written so your client’s sustainability team can file it, and so it holds up when somebody checks.
Built on manufacturers’ own footprint data. Independent review in scoping; first pilot running.
Two numbers, kept apart on purpose
What you bought on their behalf
Every spare part and every replacement device, added up in kg CO2, for every case. This is the number their sustainability team is missing. In their vocabulary it belongs to Scope 3, the part of a company’s footprint that comes from what its suppliers buy and make. You are that supplier.
What repairing saved
How much less CO2 a repair caused than the replacement the device would otherwise have got under your client’s own policy. With the margin of error stated, and only counted for cases where the answer is clear. Reported on its own line, never mixed into number one.
The same way every time, and you can see it
Your client’s own policy
A new device, or a refurbished one, whichever their policy actually pays for. We do not pick the flattering option.
Worked out, not guessed
From the type of repair, the device’s age and its condition. Nobody types this number in, because it is the number the whole result hangs on.
Counted honestly
The parts at their real footprint, new or reclaimed, plus the transport and the work. Subtracted, not ignored.
A report worth filing says so
A flat figure per repair cannot tell a battery from a mainboard. Ours can, and the spread is wide: on the same device, the best repair is worth ten times the worst. Mid-range models in good condition, against a new replacement. In brackets, the same repair against a refurbished one, which is the much harder test.
| Repair | Advice | Saved, 2 years old | Saved, 4 years old |
|---|
Every row is measured against a policy that pays for a new device, and against that test a repair usually wins. The bracket is the hard test: a negative number means a refurbished unit would have caused less CO2 than this repair, so a client whose policy pays for refurbished gets a “renew” on that same case. Batteries are the only repairs that win both outright.
A tool that only ever says “repair” is not worth reporting from
Real results from the method. It opens on a case where it advises against repairing.
Who sees a “renew” first? You do. The platform is yours and nothing leaves it until you issue a report. Your client’s report shows the totals, how many cases replacing was the better call, and the repairs that count. It does not list your partners’ misses one by one. What you cannot do is drop cases from the total, because a total that only counts the wins is the one that gets rejected.
What the law does, and what it does not
EU reporting rules put everything your client buys to settle a claim inside their own mandatory CO2 inventory. They can estimate it from invoices, which treats a repaired phone and a new one as the same euro. Which device, which part, repaired or replaced: that sits in your claims file and nowhere else.
Your client has to report what it buys
Companies above 1,000 staff and 450 million euro turnover, which is most principals worth having. For an insurer, the claims supply chain can weigh more than ten times its own offices and operations.
Nothing in law forces you to hand it over
In law your client asks. From your biggest client, you already know, asking is ordering. Either way the administrator who can answer, with a method behind the number, is the one who keeps the account.
We will not tell you that right-to-repair rules or product passports require this, because they do not. The full version, with every directive, article and date, is on the Why it is needed page. It is written for your client’s sustainability team. Forward it to them.
Every number names the document it was read from
Not “based on industry data”. Each figure is a row in a register that names the manufacturer’s report or the published study, its edition date and a link. The list is public in the state it is actually in, holes included, so anyone can check the sources before they trust the arithmetic.
Each figure carries the uncertainty its own document publishes, so a device whose maker is vague reports a wide band and the report says which is which. The advice is steadier than the kilograms: moving every footprint from half to double changed a handful of recommendations out of , because the device’s total sits on both sides of the comparison and cancels out.
Who is behind it
Bring us a hundred old cases
We run them through the method and send back the repair or renew call on each, both numbers per client, the margin of error, and a frank note on where your data was thin. About two weeks, and nothing for you to build.
What European law actually asks for, and where the claims file is the only source of primary data
This page is written for the sustainability or reporting team of an insurer, manufacturer or retailer, and for their auditor. Claims administrators: forward it. We would rather be accurate than urgent. No rule obliges a claims operation to report avoided emissions. The rules do something more useful to you: they put the goods you procure inside your principal’s mandatory inventory, and only you can say what those goods were.
Your principal must report gross Scope 3, including purchased goods and services
Under CSRD as amended by Omnibus I, undertakings above 1,000 employees and 450 million euro turnover report against ESRS. ESRS E1 requires gross Scope 1, 2 and 3 for the significant categories, in disclosure E1-6 of the 2023 standard and E1-8 of the revised standard that applies from financial year 2027. At that size there is no Scope 3 phase-in left to lean on.
Claims fulfilment sits in Scope 3 category 1, and for an insurer it is large
Replacement devices and repair parts bought to settle claims are purchased goods and services. For a typical property and casualty insurer, claims supply-chain emissions can exceed its operational emissions by more than an order of magnitude, and repair over replace is a named lever.
They can estimate that line on spend. Only the claims file can supply primary data.
ESRS lets them estimate Scope 3 from spend or other proxies, but they must also disclose how far the figure rests on primary supplier data rather than proxies, and what they intend to do about it. A spend estimate treats a repaired phone and a replaced one as the same euro. What was actually bought to settle each claim is the primary data that moves that disclosure, and it exists only in the claims file. Our per-case statement records it and the period report rolls it up with full coverage.
The value-chain cap will mean they may ask, not compel. So answering is an advantage.
Once transposed, by 19 March 2027, Omnibus I limits what an in-scope company may require from a partner of up to 1,000 employees to the voluntary SME standard, whose basic module has no Scope 3 at all. It carries exceptions, and a TPA above 1,000 staff is not covered. In practice: your principal’s sustainability team will still ask, and the TPA that can answer with a method behind it is the one that keeps the account.
Whatever is quoted to a consumer has to survive the green-claims rules. That includes what you say to a claimant.
From 27 September 2026 the Empowering Consumers Directive bans generic environmental claims, offset-based neutrality claims and unmonitored sustainability labels in anything said to a consumer. It binds every trader who speaks to one, and a TPA handling a claim for an insurer is a trader. Every figure we issue carries its baseline, its range and its limitations on the face of the document, is never a label and is never netted, so it can be quoted without becoming a liability. Whether to quote it to a policyholder is still your call, with your own advice.
What we will not claim
The Right to Repair Directive is consumer law. It obliges a seller to tell a consumer about the choice between repair and replacement, and reaches manufacturers of phones and tablets only where ecodesign repairability rules already apply. None of it touches insurance-funded replacement and none of it is about CO2e. Digital product passports for electronics are not adopted: the Commission’s working plan foresees a repairability measure in 2027 and product rules for phones and tablets in 2030, and neither places a duty on repairers. So neither is a reason to buy this, and we do not use them as one.
Dates, verified 2 September 2026
| Instrument | Applies | Binds | What it means for this report |
|---|---|---|---|
| CSRD as amended by Omnibus I; ESRS E1 | In force; scope above 1,000 employees and EUR 450m turnover; revised ESRS apply from FY2027 | Your principal | Gross Scope 3 for significant categories, purchased goods and services included (E1-6, becoming E1-8). Figure one is primary data for it. |
| Value-chain cap; voluntary SME standard | Directive (EU) 2026/470; transposition by 19 March 2027 | Your principal’s requests to partners of up to 1,000 employees | They may then only require VSME-level data, with exceptions. VSME basic has no Scope 3. You are not compelled; you are advantaged. |
| Empowering Consumers Directive 2024/825 | 27 September 2026 | Every trader communicating with consumers, a TPA acting for an insurer included | No generic claims, no offset neutrality, no unverified labels. Our documents are none of those; quoting them to a consumer is your call, with advice. |
| Right to Repair Directive 2024/1799 | 31 July 2026 for sale-of-goods contracts; manufacturer repair duty for phones and tablets placed on the market from 20 June 2025 | Sellers and manufacturers, to consumers | Consumers are informed of the choice; not insurance-funded replacement; not about CO2e. We do not cite it as a reason to buy. |
| ESPR digital product passport, electronics | Horizontal repairability measure foreseen 2027, scope open; phones and tablets 2030; computers under a separate review | Manufacturers | No repairer obligation. We watch it; we do not claim it. |
What your client actually receives
One document per client per period, written so their sustainability team can file it without translating it first, and so an auditor can recompute any line in it.
What is on it
- Number one, what was bought. The CO2 of everything bought to settle the period’s claims, every case. This is what their sustainability team needs for the supplier part of their footprint (Scope 3, category 1).
- Number two, what repairing saved. Compared with their own replacement policy, with the margin of error, the cheapest-alternative comparison, and a clear rule for which cases count that adds up on the page.
- Coverage. The period as dates, the data cut-off, and the walk from every case in the period to the cases that count, negatives included.
- Per repair partner, reconciling to the totals, and an annex listing every included case.
- How it was prepared. Method version, dataset, which replacement policy was used and why, service lives, decision rule, margins, review status.
- How to use each figure, and the limitations verbatim from the method, including the inconvenient ones.
Results the method produced, including the ones against us
Each is a real result under the default settings, shown as the statement a claims handler would attach to the file. The interactive tool, with your own device list and your clients’ own replacement policies, is in the platform.
Every statement carries a code. Check one here.
The code on a statement is a fingerprint of its inputs, its settings and the method version. Enter what the statement says and the same code should appear. If it does, the document came from these inputs under this method and nobody changed a figure afterwards. If it does not, something was edited. A matching code does not prove the inputs were true; that is what the source register, the stated policy and, in time, the independent review are for. Your client’s auditor is the person most likely to use this page.
Settings assumed at their defaults unless the statement says otherwise: circular hierarchy 15%, overhead ×1, reclaimed factor 15%, bands ±30% and ±40%, ceiling factor 2.0, no extension override. A statement with an override cannot be reproduced here; it names the reason and the value, and the platform reproduces it.
How the figures are made, and where the numbers come from
The structure is public because it is what makes the number defensible. The parameter tables that feed it are disclosed to clients and to the reviewer, and printed on every document a client receives, so anyone holding a statement can check its arithmetic.
Per case: repair or renew
From the device, its age, its condition, what failed and what would be fitted, against your client’s own replacement policy. No personal data goes into it.
Every case on its own, never an average
Not a case count multiplied by a typical saving. Each case carries its own counterfactual, and the ones the method cannot settle are marked as needing an input rather than guessed.
Per period: what the quarter actually saved
The two totals per client, with the range, the inclusion basis, the limitations, and a code your client’s auditor can re-enter to confirm nothing changed after issue.
How that reaches you is deliberately still open: a report we produce, a tool your own team runs, or a calculation layer inside a system you already use. We would rather fit the way you work than sell you ours, and we would rather agree it with the first few clients than guess. The rest of this page is the arithmetic behind it, written to be checked.
The calculation
Annualised embodied impact in kg CO2e per year of service, following the ADEME approach to refurbished goods and Boavizta’s linear amortisation of the manufacturing footprint. Use-phase is equal across scenarios by assumption, excluded from the decision and the reduction, and reported separately.
| Quantity | How it is established |
|---|---|
| Procured | Embodied footprint of what was actually bought: parts and the repair act, or the replacement device. An inventory quantity, all cases, no baseline. |
| Repair, per year | Parts at their origin-adjusted footprint plus overhead, spread over the years the repair buys. |
| Refurbished replacement | Residual embodied share plus reconditioning, spread over its service life. |
| New replacement | Manufacturing footprint spread over the client’s stated service life. |
| Years the repair buys | Derived from the repair type, the device’s condition and its age, capped by a technical ceiling set as a multiple of the stated service life. Never typed in. An override is bounded and must carry a reason, which is printed. |
| Advice | Repair, if it is no more than 15% worse on embodied impact than the replacement the policy would otherwise provide. Or strictly the lowest, if your reviewer prefers. The lowest-impact alternative on the market is computed alongside on every case as a conservatism check. |
| Avoided | Baseline rate times years bought, less the footprint of the repair. Against the principal’s evidenced counterfactual; the best-available figure printed beside it as a sensitivity. |
| Range | The device footprint shocked as one quantity, because every share-based part moves with it; years bought, refurbishment residual and overhead shocked independently; combined by root sum of squares. |
| Reduction | How much repair saves against new, as a percentage. Deliberately not a letter and not a class. |
Limitations, printed on every statement
What we will not claim
Does your tool just always say repair?
No, and the home page shows where it does not. Cheap parts on young devices win comfortably; screens and mainboards frequently lose to a refurbished replacement. About a third of demo cases sit close enough to the line that the advice moves inside the uncertainty band, and each of those says which single input would settle it.
Where is the certification?
There is not one. What there is: a versioned method, published assumptions, a range on every figure, and an independent ISO 14040/14044 critical review being scoped. A panel review, because the tool makes comparative assertions disclosed to third parties.
Can my principal book the avoided figure as a Scope 3 reduction?
No. Avoided emissions are a comparison against a baseline, not a change in an inventory. ESRS keeps them out of gross figures and targets, and following the GHG Protocol and the WBCSD guidance we report them separately, with the baseline stated. What your principal can use in their inventory is figure one, the procurement figure, which is a purchased-goods quantity and needs no baseline.
Why are the parameter values not on this site?
Because the tables are the work. The structure of the method is public so it can be judged; the calibrated values are disclosed to clients on every document and to the reviewer in full. That is the same line a testing laboratory draws between its method and its instrument.
Why is there no A to E label?
We withdrew it. A letter scale in coloured chevrons is the visual grammar of the EU energy label and the repairability class carried on smartphones and tablets since June 2025. A percentage against a stated baseline says the same thing and cannot be mistaken for a regulated label or lifted into consumer material.
Does it cover cost?
Not as an input, deliberately. The Right to Repair Directive already makes cost the legal test. One climate metric alongside it stays explainable.
Every number the method uses, and where it comes from
The source type says how good each one is: a manufacturer’s own product carbon footprint, a published study, or an estimate for a class of device. The values are shown to clients on every document and to the independent reviewer; what is public is the inventory of what we rely on and how much of it is measured rather than estimated.
How good each number is, and whether it matters
Manufacturers do not publish on one method, and the difference is not small: the same laptop can carry a footprint that differs by a factor of two between a streamlined datasheet and a full life-cycle assessment. Averaging that away would be dishonest, so we do not. Each row carries the spread its own document publishes, never narrower than our own floor, and the family is printed on every statement.
The advice
The kilograms
A portfolio drawn from a manufacturer that publishes conservatively reports a wider range than one drawn from a manufacturer that publishes precisely. That belongs in the range, not hidden in the total. Where a client’s fleet spans more than one family, the report says so on its face and states that no manufacturer-against-manufacturer conclusion can be drawn from it.
The years a repair buys is the assumption we cannot source, and it is most of the margin
Three quarters of the width on the reported figure comes from one input: how long a repaired device actually stays in service. No manufacturer report and no public study contains it, because it is a property of a repair population rather than of a device. Everyone in this market is estimating it, including us, and we say so on every document. It is also the one input a claims administrator can settle from their own history, which is why the first thing we ask a client for is which repairs came back and after how long. That is what turns a defensible estimate into a measured one.
Assumptions, as distinct from measurements
Not taken from anyone’s document. Choices we made, printed on every statement, and exposed as settings so a client can substitute their own.
Version log
Circulife
Circulife is a Belgian company, based in Bonheiden, that helps organisations get more life out of their smartphones, laptops and tablets through local repair and reuse. This tool grew out of that work: it makes what a claims operation procures reportable and what it avoids defensible, without claiming more than the method can deliver.
What this website collects
Nothing. That is not a slogan, it is a description of the code, and you can check it by reading the page source.
This site
The website is a single static file. It sets no cookies, loads no analytics or tracking script, stores nothing in your browser, and sends nothing anywhere. There is no account, no login and no visitor profile. The demo request form does not submit: it composes a message in your own mail client, which you then choose to send or discard. Until you press send in your own mail programme, we have not received anything.
One exception, stated because it is one: the page loads its typefaces from Google Fonts, so Google sees the IP address of anyone who opens the site. We are looking at serving the fonts ourselves to remove that. Our hosting provider, Vercel, processes request logs to serve and protect the site.
The method, and client data
The method itself takes no personal data. A case is assessed from the device, its age, its condition, the fault and the origin of the part. None of that identifies a person, and the platform is built so that it cannot: where a client's own reference travels with a case, it is their reference, and we ask them not to put a person's name in it.
Where we process a client's case history under a contract, we do so as a processor on their instructions, for the period the agreement runs, and we return or delete the file at the end of it. The terms are in the agreement, not on this page.
Writing to us
If you mail hello@circulife.eu we keep your message and your reply address for as long as it takes to answer you and to keep a record of the conversation, and no longer than we have a reason to. We do not add you to a mailing list, because we do not have one. You can ask us what we hold, ask for it to be corrected, or ask us to delete it, by writing to the same address.
The controller is Circulife BV, Peerlaarstraat 80, 2820 Bonheiden, Belgium, enterprise number 0805.197.493. If you think we have handled your data badly you can complain to the Belgian Data Protection Authority, Drukpersstraat 35, 1000 Brussels.
Last changed 8 September 2026. If this page ever stops matching the code, the code is the thing to believe, and we would like to know.
Twenty minutes, then a hundred cases
Tell us roughly how many devices you handle and who you report to. We will show you the method on one of your own cases and tell you plainly whether we can help.
What happens next
- We reply personally within two working days.
- A twenty minute call where you talk and we mostly listen.
- You send an anonymised extract of a hundred closed cases. Within two weeks you get both figures per principal, the statements and the range.
- We will ask whether you can see which repairs came back. That is the one dataset that lets us stop estimating the most important input, and it is worth asking for early.
- If the number holds up, we work out together how it should fit the way you work: something we run, something your team runs, or something that plugs into a system you already have. That is genuinely open, and we would rather settle it with the first few clients than guess. If the number does not hold up, you have lost two weeks and learned something true.