What your certificate states — method version cert-3.1.0

This page explains certificates of method version cert-3.1.0. The version is stated on every certificate. It is tied to the signed wording: a certificate issued today is to be read against this explanation in five years' time, not against a later edition. The calculation procedure is set out in our methodology, the revision history in the change log.

New compared with cert-3.0.0: in the edge zones of the state of charge (the very bottom and the very top), the curve follows the discharge side of the measurement runs. The certificate states extractable energy; charging and discharging read measurably differently in these zones, and what counts is the energy the battery actually delivers. Where both directions cover the same segment, their measured difference additionally enters the uncertainty band. How the headline figure is derived in principle — from the measured progression of stored energy over state of charge, with a stated measured range and the previous calculation as a comparison line — is unchanged from cert-3.0.0.

Each figure is explained in three sections: what is stated, how the value is derived, and which claim it does not make. The third section is the material one.

Energy retention (SoCE-like)

What is stated. The estimated usable energy of the battery at the reference date, relative to a documented as-new reference energy for the same model variant. Alongside it stands an uncertainty band ("± 3.1%"), which forms an equally important part of the figure.

How the value is derived. From the vehicle's charging and driving runs, the progression of stored energy over state of charge is determined as a curve: each run establishes how much energy lies between two states of charge, and assembled together these stretches form the curve from 0 to 100%. The headline figure is the total area under that curve — the energy that actually lies between a full charge and the discharge limit. Regions the vehicle's everyday use has not reached are filled in by a versioned template per cell chemistry (see measured range). In the edge zones the discharge side of the runs determines the curve, because the stated quantity is extractable energy and charging and discharging read measurably differently there; without a discharge measurement, the template applies in an edge segment, and charging runs alone do not establish it. No blanket correction factor is applied at any point: every departure from the straight line is either measured on the vehicle or stated as a per-segment template that measurement can replace.

For comparison, the document carries the previous calculation as its own line: the average slope of the curve, projected across the full charging range — the procedure used by certificates up to cert-2.3.0. Because the curve runs flatter at the edges than in the middle, that figure comes out systematically somewhat higher. Both quantities remain visible side by side, so that earlier certificates stay comparable and the difference between the procedures is in plain view.

Which claim it does not make. The figure is an estimate with a stated uncertainty, not a laboratory measurement. The term "SoCE-like" indicates that the quantity is modelled on the State of Certified Energy under UN GTR 22 but is not determined by that procedure: no test cycle was driven and no type approval granted. Certificates from different providers are comparable only insofar as they use the same reference quantity; the one used here is stated immediately alongside.

Relation to the term "SoH". In everyday use, a figure like this is called State of Health. This document does not call it that, because SoH presupposes the as-new state of this particular vehicle, and that vehicle's initial capacity was never measured. The figure answers the same question (how much energy the battery still takes in today compared with new), but it discloses which reference quantity it is measured against.

Measured range

What is stated. The range of state of charge over which the curve is backed by real measurement runs — for instance "measured from 10 to 95%". A bar next to the headline figure shows the measured zone and the model zones segment by segment. Where the full range is covered, the document says so as explicitly as it flags a measured zone with gaps.

How the value is derived. The curve is divided into fixed segments of state of charge. A segment counts as measured once the runs cover it sufficiently; in the edge zones, the discharging runs are what counts. For the remaining segments (in everyday use, mostly the very top and the very bottom) a versioned template per cell chemistry supplies the shape of the curve. Without a documented template, a straight line is the starting assumption; documented templates come from vehicles whose everyday use has actually reached the edge zones. The template's uncertainty enters the uncertainty band in full.

Which claim it does not make. A limited measured range is not a fault in the vehicle and not a deduction from the headline figure — it quantifies which part of the figure rests on measurement and which on the stated model. Making fuller use of the charging range (a drive from above 90% down to below 15% without charging in between) replaces model with measurement and narrows the band. The document claims no measurement for any individual segment where none took place: gaps can exist even within the measured range; they are visible in the bar and accounted for in the band.

The uncertainty band

What is stated. The span within which the true value is expected given the evidence — as the "±" figure next to the headline value, with coverage factor k = 1.

How the value is derived. The band of the curve-based figure combines the measurement scatter of the runs with the template terms of the unmeasured segments; the blanket surcharge for the unused charging range that the previous calculation carried is replaced by these per-segment terms. Where charging and discharging runs cover the same segment, their measured difference is added as a term of its own; it replaces the blanket direction surcharge of the previous calculation. The absolute figure additionally includes the uncertainty of the as-new reference energy; the narrower measurement band is stated alongside as the measurement precision of the vehicle — the figure to use for trend and comparison.

Which claim it does not make. A wide band quantifies the remaining measurement uncertainty, not a poor battery condition. It is not a 95% confidence interval; the coverage factor is stated.

Retrospectively calculated curve figures

What is stated. Where a certificate carries the note "calculated retrospectively from the stored raw data", the curve-based figure for an already issued document was determined after the fact from the same stored measurement data.

Which claim it does not make. The note changes nothing about the measurement data. It records that the curve figure was not fixed at the time of issue — for the ongoing validation it therefore does not count as a figure specified in advance.

Usable energy

What is stated. The numerator of the percentage, in kWh: the amount of energy available between a full charge and the discharge limit indicated by the vehicle. A note beneath it states the value under the previous calculation.

How the value is derived. As the total area under the curve described above — measured segments from the runs, unmeasured segments from the stated template.

Which claim it does not make. The figure does not denote the gross cell capacity, nor the manufacturer's data sheet value. Between gross and usable capacity lie reserves that the vehicle keeps in hand by design.

As-new reference energy

What is stated. The reference quantity of the quotient: the usable energy of this model variant when new. A quality note may be added:

• Variant resolved — the specific variant is identified; the reference quantity is a documented manufacturer figure.
• Variant band · midpoint of two documented packs — two different battery packs are documented under the same model designation and cannot be distinguished from the available characteristics. The reference quantity is their midpoint; the spread is accounted for in the uncertainty band. No claim is made as to which variant is fitted.

The basis of the figure. The reference quantity and the measured usable energy stand on the same basis: the energy between an indicated 100% and an indicated 0%. The reserve the vehicle keeps below the indicated 0% — usually two to four kilowatt hours — is included in neither figure, so the quotient compares like with like. Figures on other bases come out higher: the battery management system's full-pack energy counts this reserve, and the gross capacity additionally counts the parts of the cells locked away by design. A comparison with such figures only holds after converting to the same basis.

How the value is derived. From a versioned table of reference quantities, not from the vehicle's model designation. The same designation covers technically different vehicles across model years.

Which versions apply. For this method version the table of reference quantities is fixed at veh-energy-1.1.0, SHA-256 checksum 0dbebc608277f354e5487de800203a6def0a99fe7e0ebf44f4e4729390bcc4b4. For vehicles outside the ongoing validation study whose battery pack has been identified by a documented pack code, the pack table veh-packs-0.5.0, SHA-256 checksum 8477e511f01888a68f4552821bf6e1a61e936ff7accb04da8c207386ac2e72ba, applies in addition since 31 August 2026; it lists one reference quantity per battery pack rather than per model variant. If either table changes, its version and this page change with it — a document already issued keeps the details of the version it was calculated against.

Which claim it does not make. The reference quantity is not a measurement of the particular vehicle when new. Manufacturing spread between identical packs is accounted for in the uncertainty band rather than in the figure. Altering the reference quantity changes the percentage without any change in the condition of the vehicle; the table is therefore versioned and held fixed for the duration of the ongoing validation study.

Charging sessions or individual data points

What is stated. The number of observations entering the estimate. The label identifies the unit counted:

• Charging sessions — each entry corresponds to one complete charging session admissible for the analysis, each underpinned by a large number of individual readings.
• Individual data points — each entry corresponds to a single reading within a measurement run.

How the value is derived. Only observations that passed the admissibility check are counted; runs that are too short, too widely scattered, or mixed across data sources are excluded beforehand.

Which claim it does not make. The figure does not state the total number of charging sessions of the vehicle, nor does it indicate a thin evidence base. Whether the evidence supports the estimate is read from the uncertainty band and the measured range, not from this count.

Trend per 10,000 km and measured decline

What is stated. The change in usable energy across the distance and time observed to date, given in kWh per 10,000 km and in kWh per year.

How the values are derived. From a regression across the measurement runs, kept separate from measurement scatter. An axis is stated only where it could actually be estimated: below a span of 5,000 km or 90 days, no slope is determined. Where the certificate instead notes an insufficient span, that is the accurate statement — and is not to be equated with a trend of zero.

Which claims they do not make. The values are not a forecast. Battery ageing is not linear; the first percentage points decline more rapidly, after which the curve flattens. Extrapolating the observed decline linearly systematically understates remaining service life.

Data provenance and quality tier (Q1–Q4)

What is stated. The source of the measured energy and the robustness of that channel. Without this information a percentage cannot be interpreted.

How the value is derived. From the channel that supplied the stated figure. Where the underlying history contains several sources, all are named.

Which claim it does not make. The tier ranks data quality, not the independence of the measurement; a higher tier therefore does not denote the more independent figure. Q1 offers the highest data quality — high sampling rate, energy directly in kWh, no conversion constant — yet remains a quantity the vehicle estimates itself. Q2 is the only tier that captures a quantity the vehicle does not estimate: the energy actually delivered. Both characteristics are therefore stated separately, and a Q2 certificate carries the note "measured independently of the vehicle's own estimate".

Q4 is the weakest tier: deriving the figure from the range display via a constant can yield values above 100%. It serves solely as a plausibility check and not as the primary estimator.

Battery status date, issue date and odometer

What is stated. The time of the last measurement together with the corresponding odometer reading — not the time at which the page is opened. Where that date lies several months back, the certificate flags this actively.

Which claims they do not make. A certificate remains bound to its reference date. The stated odometer reading should be compared with the vehicle's own: if several thousand kilometres lie between them, the document describes an earlier condition.

VIN checksum and signature

What is stated. The binding of the document to one particular vehicle and to an unaltered wording.

How the entries are derived. The full vehicle identification number is not contained in the certificate; only its SHA-256 digest is. Anyone who knows the full number can check on the certificate page whether it matches that digest. The wording itself is signed with Ed25519; verification is performed automatically in the browser when the page opens, using an embedded public key and transmitting no data.

Which claims they do not make. A valid signature establishes that the document originates from us unaltered — not that the battery is in good condition. The percentage remains an estimate even where the signature is valid.