Roughly one in seven public charging attempts does not result in a charge. That is not an impression from a forum thread. It is the finding of a 2025 study of 7,428 electric and plug-in hybrid owners, and it puts a number on the thing every EV driver already suspects. More usefully, it identifies the dominant cause, and the cause is not what most people assume.
The number, and what it actually measures
J.D. Power's 2025 U.S. Electric Vehicle Experience Public Charging Study, fielded January through June 2025 across 7,428 BEV and PHEV owners, found that 86% of charging attempts succeed and 14% fail.
Two pieces of context matter before anyone quotes that figure. First, it is improving: failed visits fell by five percentage points from 2024, which is a substantial year-on-year move and worth remembering when the topic comes up as though nothing ever gets better. Second, it is United States data. European networks differ in density, operator mix, payment norms and roaming arrangements, so treat 14% as a well-measured indication of the scale of the problem rather than a figure to apply to a specific country. We are not going to pretend a US sample describes Poland.
The dominant cause: the charger is simply broken
Here is the finding that should reshape how drivers plan. The most common reason for a visit that produces no charge is the charger being out of service or not working properly: 60% of failed charging visits.
Not queueing. Not the wrong connector. Not payment refusal. Not a car-side fault. Six out of ten failures are a unit that was, in some sense, already dead when the driver arrived.
That single statistic explains why range anxiety persists even as networks expand. Adding chargers to a map does not help if a meaningful share of them do not work, and it explains why experienced EV drivers behave in ways that look paranoid to newcomers, arriving with more reserve than the arithmetic requires, and treating any single charger as a plan that might not survive contact with reality.
The uncomfortable implication: if most failures are broken hardware, then the useful question about a charger is not "is it there?" or even "is it free?": it is "did it recently work for somebody?" Those are different questions, and most tools answer the first two.
Why an app can say “available” about a dead charger
This is worth understanding, because it is a structural problem rather than sloppiness by any one company.
Charger status usually travels a long chain: the hardware reports to the operator's back end, the operator exposes that to roaming platforms and aggregators, and an app you are using reads it from there. Every link introduces both delay and interpretation. Worse, the statuses in that chain describe session state (is a session open, is the connector engaged) rather than health. A unit that is powered on, reporting cheerfully, and physically incapable of delivering a charge because of a payment terminal fault, a locked connector, an authentication failure or a firmware hang can appear as a perfectly ordinary available.
So "available" honestly means nobody is currently plugged in. It does not mean this will work for you. The only source of information about whether a charger actually delivered electricity is somebody who tried.
A second dataset, and a much harsher number
The 14% figure comes from a driver-experience survey. A separate study measured the same problem from the hardware side, and it does not agree, which is itself the most informative thing about it.
ChargerHelp! published its EV Charging Reliability Report on 24 September 2025, built from more than 100,000 charging sessions across 2,400 chargers. Its headline finding: only 71% of charging sessions succeed, close to one attempt in three failing. Its measure of first-time charge success rate averaged 85% at newly installed stations and fell to roughly 70% by the third year.
Set that against what the same chargers were reporting about themselves: 98.7% to 99.9% uptime. Some report 100% uptime and still fail without retries, resets or errors. The report's own framing is that more than a third of failures happen on chargers that appear entirely operational.
Why the two numbers differ so much: uptime asks whether the machine is powered and reachable. Success rate asks whether a driver got electricity into a car on the first try. A charger can be flawless by the first measure and useless by the second, and the metric operators publish, and that apps repeat to you, is almost always the first one.
There is a second lesson buried in the age curve. A charger's reliability is not fixed at installation; it decays. An 85% first-time success rate at a new site becoming 70% by year three means the map entry that served you well two years ago is a weaker prediction today than it was then. Static databases have no way to represent that. A report from this week does.
What actually reduces your risk
None of this is an argument against electric driving; it is an argument for a specific kind of planning. What experienced drivers do, in rough order of usefulness:
- Never plan on a single charger. Pick a primary and a genuine alternative within your remaining range. The 14% figure applies to attempts, so a second independent option collapses your practical failure rate dramatically.
- Prefer sites with multiple units over single-unit sites. A four-bay location where one bay is dead is an inconvenience. A single-unit location where the unit is dead is a towing conversation.
- Weight recent human confirmation above reported status. If you can see that a driver charged there today, that is stronger evidence than any status flag, for the structural reasons above.
- Arrive with reserve, not with precision. The arithmetic that says you can make it with 4% left assumes the charger works. Build in the failure case rather than the average case.
- Carry the payment fallbacks. A contactless card as well as the network app, since authentication is one of the ways a working charger becomes unusable.
- Report what happened. Whether it worked or did not, the next driver is making the same decision you just made, with the same information you just wished you had.
Why this is exactly what driver reports are for
The 60% figure is the entire argument for crowdsourced confirmation, stated by a third party with a large sample.
If failures were mostly queues, live occupancy data would solve them. If they were mostly connector mismatches, better filtering would solve them. But when six in ten failures are hardware that does not work, the only reliable signal is a report from somebody who plugged in, with the time attached, because a charger that worked three weeks ago tells you nothing about tonight.
That is why Refuelia treats a driver's "this charged at 148 kW, card worked first try" as more valuable than any status field, and why a charger with no recent confirmation is shown as exactly that, no recent report, rather than dressed up in a colour that implies a promise nobody made. The alternative is the failure mode the study measured: a driver arriving somewhere on the strength of a green dot that never meant what it appeared to mean.
Fourteen percent is a solvable number. It gets solved by better hardware and better maintenance from operators, and in the meantime, by drivers telling each other the truth about what happened when they plugged in.