Buying a used Zero: what the logs prove

Buying a used electric motorcycle splits cleanly into two questions, and confusing them is what makes the whole exercise feel unknowable. The first is: how has the expensive part been treated? That question has an answer in the bike’s own memory, and it is the one this site can help with. The second is: what state is the rest of the motorcycle in? That question has never had an answer in a log file, on any electric vehicle, and pretending otherwise is where a buying guide starts doing harm.

The main board and the battery management system write a running record: charge sessions with their power and their duration, pack temperatures, cell voltages, faults with a timestamp, the odometer. Pull that file and you are reading the same source a dealer reads. Zero says as much itself — the warranty clause below names a "battery management system log data extraction" as the way a pack’s capacity is established.

Everything mechanical is silent. There is no sensor on a brake pad, a tyre, a wheel bearing, a fork seal or a drive belt, so none of them appears in a log at any mileage. Neither does a drop, a bent frame, a cracked fairing or the corrosion under a bracket. A motorcycle that has been dropped writes exactly what one that has not writes.

The index is a single number out of 100, and it is the mean of the checks that could be answered — published with that count attached, always, in the same sentence: "82 / 100 over 3 of the 4 checks this motorcycle allows". A reader who quotes the score has quoted the coverage with it.

One rule holds the whole thing up: every check measures a distance from a limit Zero itself publishes, and every check declares the value that scores 100 and the value that scores 0. Nothing is a percentile against our own fleet. Nothing is a weighted blend of opinions. There is one curve — a straight line between the two declared ends, clamped at both — because a reader with a calculator has to be able to reproduce the number.

That rule is not aesthetic. This project has already shipped a health score that graded how a pack behaves — balance, temperatures, deep discharges — and never asked how much of it was left. It printed "97 Excellent" three centimetres above the same page’s own 70% state of health. Every step of its arithmetic was correct. What it lacked was a denominator anybody could check.

The first build of this index averaged whatever could be scored. That is the obvious design, and it is wrong in a way that only shows up when you point it at a real motorcycle owned by somebody with an incentive.

On one bike in this fleet the three checks that answered scored 30, 88, 88, and the index read 69 over 3 of the 4 checks. The pack size on file for that motorcycle is a field its owner can type. There is a rule that exists to protect the buyer from exactly that: when the size on file is not the one the VIN decodes to, state of health is withheld, because state of health is a measured energy divided by that size and a value computed from a declared input stays declared however much arithmetic sits on top.

Withhold it, and a mean over the checks that remain reads 88. The safeguard handed the seller 19 points, because dropping the lowest number raises an average. The protection arrived through the door marked "safety" and walked out with a better score.

It cannot be fixed by weighting. Any average over a varying subset does this: whatever is missing is implicitly assigned the mean of what remains, which on a document a seller publishes is the most flattering assumption available. So the battery check is decisive. Without it there is no index — and the sentence that replaces the number says which check was missing and why.

The other checks are unaffected and still publish. Only the single headline is withheld. And there is a floor underneath all of it: below 2 answered checks there is no index either, because a mean of one value is that value wearing a different name — and two motorcycles both reading 82 might have scored on different single checks, which is the opposite of what an index is for.

State of health is the energy the pack measures now over the energy it held when new. It survives scrutiny for one reason: both of those are real quantities somebody published, so the ratio can be checked rather than believed.

The limit is Zero’s. The warranty section of the owner manual says the company will repair or replace a pack "that exhibits a nominal storage capacity reduction of greater than 20% of the published nominal capacity, as measured by an authorized Zero dealer", and one sentence later names the measurement as "a battery management system log data extraction". The capacity clause is word for word in all 38 owner manuals in our corpus and in none of the 8 service manuals, which carry no warranty section at all. The sentence naming the log extraction is in 35 of the 38: it is absent from LS1, XB and XE, Zero’s newest EU manuals, which end the paragraph one sentence earlier. Those three also carry the shorter warranty of the light-EV line, so the whole passage is worth re-reading rather than assumed if the motorcycle in front of you is one of them.

The 80exactly 80% still delivers four fifths of its original range, which on a seven-year-old bike is an ordinary motorcycle rather than a broken one. Scoring 0 there would flatten precisely the older, cheaper bikes a buyer is most often looking at, and stop the check discriminating between a tired pack and a dead one.

The zero end follows from the two anchors rather than from an opinion: the straight line through 100% and through 8060%. It is derived in code so that moving either anchor moves it too, and it is never presented as a Zero figure, because it is not one.

The two denominators had to be checked before any of this was legitimate, because Zero’s terminology is the reverse of this application’s. The manual calls 6.3 kWh "Nominal Capacity" and 7.2 kWh "Maximum Capacity"; this site calls those usable and nominal. The usable-when-new figure here is the badge times 0.88, and against the manuals’ own Nominal rows that lands within 0.94% on every pack checked. The state of health published here sits on the warranty’s own denominator.

This check publishes the share of charges we could place that ran at or below the onboard charger rating Zero publishes for that model year. A charge at or below that rating needed no charging hardware Zero did not fit to the motorcycle.

It says nothing about harm, deliberately. High current at low temperature can plate lithium metal on the anode instead of storing it, and that is permanent — but we have not measured that correlation on THESE motorcycles, and a threshold tuned without a positive example is a guess wearing a measurement’s clothes. So the check is a record of how the bike was charged, with the mechanism named beside it, and no weight is shipped ahead of the study that would justify one.

The scored quantity is a share rather than a charge rate, and the reason is a limit on what Zero publishes rather than a preference. A score needs two ends. Zero publishes the onboard rating and no maximum charge rate at all: the rapid-charge field is empty on all 90 owner-manual specification sheets. A proportion needs no invented end — its ends are 0% and 100% — so the one published number does all the work and nothing is made up to anchor the other side.

The charge rate as a fraction of pack size is the better VARIABLE, and it is published beside the score as context that moves nothing. One kilowatt into a small pack is about 0.16C and irrelevant; four kilowatts into the same pack is 0.63C. The same kilowatts are a gentle charge for a large pack and a hard one for a small one. That figure is withheld when the pack size was typed by hand, because the pack is its divisor — and the share is unaffected, since no pack size appears anywhere in a count of sessions over a count of sessions.

The scale runs from 50 °C, where Zero’s manual says "the BMS will no longer allow charging", to 60 °C, the top of the range the pack "should not be used outside of", where the BMS turns the power controller off and the motorcycle stops being a motorcycle. Both figures are stated at the PACK, and both come out of the manual through a generator that confirms the page in the PDF before publishing it.

If your motorcycle is a 2014 or 2015 model, your own manual says 55 °C rather than 50, and you are not misreading it — Zero changed the figure from 2016. Every motorcycle here is scored against the STRICTEST figure any manual states, which is the lower one, for a reason worth knowing: the model on a report can be corrected by hand by whoever owns the account, and 55 is the flattering end — a 52 °C peak scores 100 against 55 and 80 against 50. Letting the anchor follow a field a seller can type would let them choose their own scale. The cost falls on genuinely old bikes, graded slightly harder than their own manual asks, and it is written here rather than absorbed quietly.

The anchor used to be 43 °C and that was a defect, fixed on 2026-09-06. Zero prints the 43 °C figure in the same paragraph, but attributes it to AMBIENT temperature and to the charger rather than to the pack — while this check reads a pack sensor. It compared a pack reading against an air temperature, and it moved the answer: a 51 °C peak scored 53 on the old anchors and scores 90 on these. The battery check had verified the direction of its own terminology; this one had not.

The published peak is the hottest temperature reached in 2 SEPARATE sessions, which is the second-highest session reading, and the single hottest reading is shown beside it where it moves nothing. A bare maximum over raw telemetry is a defect this codebase has already shipped and paid for: the battery page once headlined 16,546 mV for one cell of a 28-cell pack, from 582 corrupt rows out of 726,902 — measured and fixed on 2026-08-12. The tail here has the same shape: re-measured 2026-09-07 over the 58,090 pack-temperature samples the local log corpus decodes through the binary route, nine sit above 45 °C and six of those are single readings — 46 °C four times, then 47, 48, 51, 53 and 58 once each. Requiring two sessions costs a genuinely hot motorcycle a degree and costs a corrupt frame everything.

Most motorcycles score high here, and the check says so itself rather than letting a high number lift the index quietly. Zero’s thermal envelope is wide and these packs live well inside it: across 59,970 pack readings decoded from real logs, not one reached the 60 °C shutdown, and the maximum was 58 °C. The check earns its place by noticing the one motorcycle that did not.

Those are two different populations two paragraphs apart, and the page owes you the difference rather than a guess. The 58,090 above are what ONE decode route reads out of the binary logs — 41 content-distinct files from about a dozen motorcycles. The 59,970 here is that set plus 1,880 readings from a SECOND and independent route, the PackTemp: h …C, l …C line a serial console prints. Both were measured on 2026-09-07 and both come from the local log corpus rather than from the database, which is why neither carries a fleet denominator. The two routes agree on the thing that matters here: neither reaches 60 °C. An earlier draft published 53,219 and 55,099 from a run whose file selection was never written down; those two totals did not reproduce, and the distribution behind them did, reading for reading.

Chassis isolation is whether the high-voltage system is leaking to the frame — the fault a wet week produces and the one nobody can see. It is the fourth check in this index, and it never scores, on any motorcycle. This is the clearest worked example of the rule the whole page rests on: a component can be absent, and its absence is information rather than a gap.

The reason is about Zero, not about your bike. Every other check measures a distance from a number Zero publishes. For isolation there is no such number, and that is a measurement rather than an impression: run pdftotext over all 46 manuals we hold — 38 owner and 8 service — and the word "ohm" appears zero times, in any spelling. So do "megohm", "insulation resistance", "isolation resistance", "insulation test" and "continuity". Zero names the CONDITION, in the troubleshooting table of 22 of those 46 manuals, as dash codes 20 "BMS Low Isolation", 28 "BMS Isolation Fault" and 29 "BMS Isolation Danger". It publishes no resistance at which any of them fires.

There is a widely used figure for a high-voltage traction system — 100 ohms for every volt of pack, which would put the line near 10 kilohms on these motorcycles. It is a general requirement, not this manufacturer’s, and the index does not score against it. The moment one of four spokes is graded against somebody else’s limit, "check each number against what Zero publishes" stops being true of the number a reader actually quotes. The per-bike report publishes the readings against that band with the disclosure attached, which is the right home for a judged-but-not-by-Zero figure.

Coverage is not a caption printed above the verdict. It is what decides which questions can be asked. A single file tells you the state on the days it saw; two years of history tell you how the motorcycle was USED — and it is the use, rather than the state, that explains why two packs of the same age measure differently.

It is counted in months that actually contain data: distinct calendar months carrying at least one log row. Never the span between the first and last date, and never the union of the report windows. Zero downloads are cumulative, so each later file begins at or before where the previous one did, the windows merge into a single island, and their union collapses to "last date minus first date" — which measures how long the motorcycle has existed.

The difference is not academic. Measured across production on 2026-09-05, per bike, the union of report windows reached six months on 140 of 174 bikes, while the median number of days actually carrying data was 43. The ratio between the two has a median of 10.2 and a maximum of 259.6. One motorcycle showed 886 days of span over 13 days of data. A criterion that believed the span would answer a question it has no evidence for.

This section is mandatory on the per-bike report and it is mandatory here, for the same reason: a clean report is the one a reader is most likely to mistake for a clean motorcycle.

An unanswered check is not a pass. Where a report says nothing about a fault, it means we saw none in the log lines we were given — not that the motorcycle does not have one. Absence of evidence is not the same claim as evidence of absence, and on a cumulative, time-bounded ring buffer the difference is large.

Nothing here is an appraisal, an insurance value or a book value. There is no published book value for a used Zero at all: J.D. Power declines to issue one, stating "Due to limited resale market, we are unable to determine used sales." That absence is a large part of why this exists — and it is also why this page carries no prices. The condition judgement and the money are kept apart on purpose, so that a verdict about a pack cannot quietly become an estimate about a price.

Most people reading this site own the motorcycle rather than shop for one, and the same file works in the other direction. Zero’s warranty clause is adjudicated by "a battery management system log data extraction" — which is precisely the artefact this application produces — so a documented state of health is not a nice extra. It is the thing that moves a price, because it is the only claim in the advertisement that a buyer can check.

The same holds for a defect. A fault documented with its date, declared by the seller, is worth more than the same fault discovered by the buyer three weeks later — and the logs date these things precisely. One owner’s onboard charger stopped answering on a specific day: the bike still charged from an external charger and nothing on the dash said anything, and the log shows 69 wake-ups on the onboard charger up to that date and none in the 68 covered days afterwards, with key-switch power-ons continuing throughout.

What a report like that cannot do is name the broken part, and it should not pretend to. A dead charger, an interrupted signal wire, mains that never reached the charger and a failed input on the main board are observationally identical: when the board is never woken, it writes nothing. Two of those four cost far less than a charger. The honest claim is a fact about the vehicle — "this motorcycle has not used its onboard charger since this date, across this many covered days" — and the cheap checks come first.

The file answers the pack. The visit answers everything else, and it is where most of the money actually is on a used motorcycle of any kind. Do them in this order, because the file can be read before you travel.